bending, compression, and shear behavior of woven glass...

Bending, compression, and shear behavior of woven glass ®ber±epoxycomposites

B. Yanga, V. Kozeya, S. Adanurb, S. Kumara,*

aSchool of Textile and Fiber Engineering, Georgia Institute of Technology, Atlanta, GA 30332, USAbDepartment of Textile Engineering, Auburn University, Auburn, AL 36849, USA

Received 20 May 1999; accepted 25 August 1999

Abstract

The mechanical properties and failure mechanisms of through-the-thickness stitched plain weave glass fabric±epoxy composites were

studied. Unstitched plain weave and biaxial non-crimp fabrics were used for comparison. Composite panels were fabricated using Resin

Transfer Molding. Z-directional stitching increased the delamination resistance and lowered the bending strength of the composites.

Composites made from through-the-thickness stitched fabrics demonstrated improved compression after impact behavior as compared to

the unstitched fabrics. The results presented in this investigation should be useful in tailoring textile composites to achieve speci®c property

goals. q 2000 Elsevier Science Ltd. All rights reserved.

Keywords: Bending

1. Introduction

Textile technologies such as weaving, stitching, braiding,

knitting are being employed to fabricate advanced compo-

sites with conformability, quality and integrated mechanical

properties [1±10]. One of the objectives of using textile

reinforcement is to take advantage of through-the-thickness

arrangement of ®bers to enhance interlaminar strength and

toughness, compressive strength, as well as compression-

after-impact (CAI) strength [11±13]. Reinforcing ®bers in

the thickness direction also contribute to stiffness and

strength in that direction.

Research efforts to improve interlaminar fracture

behavior have generally focused in the following areas: (i)

the reinforcement material, for example, the development of

three-dimensional (3D) preforms; (ii) the use of tough

matrix material such as PEEK [14]; (iii) the addition of

micro®ber [15]; and (iv) developing good ®ber±matrix

interface with the use of improved coupling agents for

controlled interfacial properties. The improved interlaminar

shear property in the thickness direction gives much

improved fracture and impact resistance over the conven-

tional laminates.

The compressive response of ®ber-reinforced composites

has been the subject of continued investigation [9,16±18].

Many factors in¯uence the compressive response of compo-

site materials. These include the compressive properties of

the ®ber [19] and the matrix [20] as well as the ®ber±matrix

interface. The presence of local inhomogeneities and

defects, which are often dif®cult to characterize and

model, also in¯uence the failure in compression. On a

macrostructure level, laminate orientation, specimen

geometry, method of loading and stress concentrations are

some of the factors that play a role in determining the

compression failure mode.

Conventional laminate composites are sensitive to out-of-

plane loading, as they are weaker in the through-the-thick-

ness direction than in the plane of lamination. Therefore, to

improve interlaminar properties, multi-layer continuous

glass ®ber textile preforms have been stitched using Kevlar

yarn. Epoxy has been used as the matrix material. The

objective of this research is to study the relationship

among textile preform architecture, mechanical proper-

ties, and the resulting failure modes. Signi®cant attention

was focused on the compressive response of the textile

composites.

2. Materials and sample preparation

The following E-glass reinforcements were used in this

study: (1) non-crimp biaxial laminate (LM) with glass ®ber

Composites: Part B 31 (2000) 715±721

1359-8368/00/$ - see front matter q 2000 Elsevier Science Ltd. All rights reserved.

PII: S1359-8368(99)00052-9

www.elsevier.com/locate/compositesb

* Corresponding author. Tel.: 1 1-404-894-7550; fax: 1 1-404-894-

8780.

E-mail address: [email protected] (S. Kumar).

in the 0 and 908 direction; (2) plain weave fabric (PW); and

(3) through-the-thickness biaxial (BS) or uniaxial stitched

(US) plain weave fabrics. The non-crimp biaxial laminate

fabric, containing 1% polyester yarn and 99% E-glass, was

obtained from Tech Textiles USA. The polyester yarn

(usually 70 or 150 denier) was used to stitch bond the

glass ®ber layers. The plain weave fabric of E-glass was

obtained from PPG. The ratio of warp to ®lling yarn was

approximately 1.3:1. Woven fabric structures result in

inherent ®ber waviness. A 156 denier Kelvare yarn was

used for stitching the plain weave fabric layers in the thick-

ness direction (denier is the weight of 9000 m length of yarn

in grams). Three and six layers of plain weave fabrics were

stitched. The stitching lines were 2, 5, 10, 15, and 20 mm

apart. In all cases, both uniaxial and biaxial stitching modes

were used. Kelvare stitching yarn accounted for less than

3% of the total woven fabric weight. Various samples tested

are listed in Table 1.

The glass preform containing three, six, or 12 layers was

placed in the steel mold. The mold dimension was 25:4 £30:5 £ z cm3

: The sample thickness varied between 0.18 and

0.63 mm. Two different molds and an aluminum plate were

used for varying sample thickness. There were two nozzles

in the mold, one for resin injection and the other one for

vacuum line to remove air bubbles. Some of the excess resin

was also removed through the vacuum nozzle. The mold

was impregnated with the epoxy resin (EPON 862 and the

curing agent W in the weight ratio 100:26). Both compo-

nents of the epoxy were obtained from Shell Chemical Co.

Before injecting in the mold, the epoxy resin was heated to

approximately 458C. The equipment used was a VRM 2.5

Resin Transfer Molding machine from Liquid Control

Corporation. The system was totally enclosed and the

mixing of resin and curing agent only took place at the

®nal stage with the mixing chamber at the point of dispen-

sing in the mold. After resin transfer, the specimen was

cured at 1258C for 6 h.

Density was measured by weighing the composite

samples in air and in water. Fiber volume fraction was

determined using the following equation:

rc � rfVf 1 rmVm

where r is the density and V the volume fraction. The

presence of voids was not accounted for. Subscripts c, f,

and m refer to composite, ®ber and matrix, respectively.

The density of the ®ber and matrix were taken to be 2.58

and 1.2 g cm3, respectively. For the stitched and unstitched

woven samples, three, six, and 12 fabric layers were used

for the bending, compression, and grooved shear tests,

respectively. To achieve comparable thickness, for the

non-crimped laminate samples, six and 12 plies were used

for the bending and compression tests, respectively.

3. Mechanical testing

The following test methods were used to determine the

interlaminar shear strength: (i) short beam shear test (ASTM

D2355); and (ii) tensile testing of grooved coupons (ASTM

D2677). The length-to-depth �l=d� ratio in the short beam

shear test was approximately 5. The thickness of the

grooved coupon was 7.2 mm, groove depth was 3.5 mm

and the distance between the grooves was 20 mm.

Low-velocity impact was applied to the specimen using a

small pendulum apparatus. The diameter of the hemi-

spherical impactor was 10 mm. The impact energy ranged

from 0 to 8.7 J, and this corresponds to 0±2.5 J mm21 of the

sample thickness. Impact loading produced near-circular

damage areas. These samples were subsequently used for

compression after impact and bending after impact testing.

Bending strength was measured according to the ASTM

790-91 test method. The three-point loading scheme was

selected, and the test was performed using Instron 5500

universal testing machine. The length to depth ratio was

40:1. Compressive strength was determined using the

IITRI method (ASTM D3410). The sample dimensions

were 120 £ 40 £ z mm3; and the gauge length was 25 mm.

4. Results and discussion

4.1. Interlaminar shear strength

Initially attempt was made to determine the interlaminar

shear strength using the short beam shear test. However, it

was observed that the short-beam shear specimens of the

unstitched fabric composites did not develop shear cracking

in the mid-plane of the specimens as required in the ASTM

test. Apparent shear strength of 47 MPa was calculated for

this composite, however, both the upper and lower surfaces

underwent extensive damage. Damage in the upper surface

of the beam is from compressive failure; damage in the

lower surface is from tensile failure. There was no indi-

cation of shear failure or crack initiation in the mid-plane.

Furthermore, ®ber unevenness deterred fabric plane

slippage, and additional through-the-thickness stitching

practically eliminated the possibility of inter-plane slippage.

Thus, interlaminar failure was not observed. Based on these

observations, it was concluded that the short-beam shear test

B. Yang et al. / Composites: Part B 31 (2000) 715±721716

Table 1

Description of the sample codes

Sample code Sample description

LM Non-crimp laminate

PW Plain Weave

US2 Uniaxial stitching 2 mm apart




BS2 Biaxial stitching 2 mm apart




is not appropriate for the shear testing of these woven fabric

composites.

Shear testing was therefore done using the grooved

coupon test. In comparison to the short beam shear test,

the grooved coupon test is purely in a state of Mode II

loading. In the grooved coupon specimens, shear cracking

initiates around the notch and propagated between the fabric

layers. The failed samples were characterized by the full-

length delamination along the plane between the grooves.

Table 2 lists the data on interlaminar shear strength of 2D

woven and through-the-thickness stitched woven compo-

sites measured using the grooved test. Two factors affect

grooved coupon test results: (i) shear stress concentration

predominantly develops around the grooves; and (ii) shear

strength measured by the grooved coupon test also varies

depending on the precision of the cut of the grooves. Some

tearing in the composite takes place during the testing of

undercut specimens. Bending and peeling also occurs

especially for the overcut specimen. Z-directional kevlar

yarn used in stitching ®ber appears to resist shear crack

development. Table 2 shows that increasing density of Z-

directional stitching yarn moderately increases the delami-

nation resistance of the composite. The crack surface was

clear and smooth for the plane weave fabric composite as

seen in Fig. 1, which represents poor adhesion between ®ber

and the matrix.

4.2. Bending strength

Bending and bending-after-impact strength data is

presented in Table 3. In general, stitching adversely affect

the bending behavior. Higher stitching density lowers the

bending strength. Three-point bending load versus displace-

ment curves for the unstitched samples were characterized

by ªkneeº effect and non-linearity. Similar load±displace-

ment behavior is also reported for this sample in tensile test

[21]. The non-linearity is caused primarily by a structural

change in the fabric during deformation. ªKneeº effect

occurs, when layers, whose ®lament axes are perpendicular

to the loading direction, crack or break. Bending curve for

uniaxial and biaxial stitched composites in general are

different from those of the unstitched plain weave compo-

sites. Densely stitched samples did not exhibit ªkneeº effect.

Bending test showed that the outer layer fracture along the

beam axis led to the ®nal failure. A closer observation of the

specimen showed that there was some visible damage on the

compressive side. As loading progressed, cracking ®rst

developed in outer ply on the tensile side. Microcracking

initiated when the stress exceeded the local matrix pocket

strength.

4.3. Impact damage analysis

The impact damage zone is generally complex in nature

and consequently not easy to characterize [22]. Due to the

lack of existing standards or the established test techniques

for impact damage of composite materials, direct compar-

ison of data for various material systems reported in the

literature are often misleading. We have determined impact

damage using a pendulum tester, which generated approxi-

mately round-shape damaged areas. Three energy levels, i.e.

0, 1.65, 2.5 J mm21, were used in the present study. Damage

area increased with the increasing impact energy. At the

impact point on the surface of the specimen, the damage

consisted primarily of crushed material with some delami-

nation between plies at the interface. A cone of damage was

formed beneath the point of impact. The amount of crushed

material decreased with increasing depth whereas the inter-

laminar delamination increased. In specimen without

through-the-thickness reinforcement, the damage cone

angle was much smaller and damage area did not increase

with the depth. An analysis suggests a number of damage

B. Yang et al. / Composites: Part B 31 (2000) 715±721 717

Table 2

Interlaminar shear strength (from grooved coupon test) of woven glass±epoxy composites

Sample code Number of plies Density (g cm23) Fiber volume (%) Composite thickness (mm) Shear strength (MPa)

PW 12 1.95 56 7.2 16.0

BS 10 12 1.96 56 7.2 17.0

BS 5 12 1.95 57 7.2 18.2

Fig. 1. Scanning electron micrograph of shear crack (grooved coupon test)

surface of unstitched woven glass±epoxy composite.

mechanisms. These include delamination, ®ber±matrix

debonding, and interlaminar matrix cracking.

The energy absorbing capability of composites depends

on the properties of the constituents, and is re¯ected in the

damage area. Table 4 lists relative damage areas for various

stitching density samples. It was clear that stitching reduces

damage area and dense stitching is more effective in the

damage area reduction. Signi®cantly higher damage area

was observed in the laminate composites than the compo-

sites made from the plain weave fabric.

An interesting phenomenon is that increasing the thick-

ness of the specimen resulted in an increase in the delami-

nation area as can be seen from the results of three and six

layer samples reported in Table 4. This increase in damage

area may result from the reduction in energy absorbing

capability as proposed by Dorey [23]. It was suggested

that the number of likely delamination sites increase with

fabric thickness. More manufacturing defects and voids may

be present in the thick samples.

4.4. Compressive strength

Compression test results are given in Table 5 and

compressive stress±strain curves are shown in Fig. 2. The

results indicate that the 0/908 non-crimp laminate has higher

compressive strength as compared to the woven samples, a

result of ®ber waviness in the woven samples. The in¯uence

of waviness parameters, such as wavelength and amplitude,

on the compressive behavior has not been studied exten-

sively. This study only describes some experimental obser-

vations and shows qualitatively that the compressive

strength decreases with the increasing ®ber misalignment

or waviness.

The different compressive response can also be re¯ected

in the failure mode. Fig. 3a shows the IITRI compression

failure of the 0/908 non-crimp laminate. Compressive load-

ing resulted in the formation of a shear band. There was no

relative displacement of the ®bers across the failure zone,

i.e. transverse movement across the failure band. Post-fail-

ure examination showed that shear band was at 458 angle to


Table 3

Bending test results

Sample

code

Density

(g cm23)

Fiber

volume

(%)

Composite

thickness

(mm)

Bending

strength

(MPa)

Bending

strength after

impacta (MPa)

PW 1.76 43 2.35 370 ^ 13 267 ^ 9

BS 5 1.71 40 2.50 220 ^ 10 211 ^ 8

BS 2 1.71 40 2.51 233 ^ 9 215 ^ 3

US 20 1.72 41 2.53 326 ^ 9 247 ^ 6

US 5 1.71 40 2.52 198 ^ 3 180 ^ 5

US 2 1.7 39 2.50 177 ^ 6 174 ^ 4

a Bending strength after impact (energy level 2.5 J mm21).

Table 4

Relative impact damage (impact energy 2.5 J mm21) areas in woven lami-

nate composites

Sample code Damage area (%)

Three layer composite Six layer composite

PW 100 (283 mm2) 100 (414 mm2)

BS 20 87 84

US 20 87 86

BS 10 43 50

US 10 ± 54

BS 5 28 39

US 5 33 43

BS 2 22

US 2 26

± ± 140

Fig. 2. Compressive stress±strain curves for glass ®ber±epoxy composites:

(a) non-crimp 0/908 laminate (LM) composite; (bi) plain weave (PW)

unstitched composite; (bii) uniaxially (US5) stitched composite; and

(biii) biaxially stitched (BS5) composite.

the loading direction. Samples failing in this mode yielded

high compressive strength values.

Fig. 3b shows the IITRI compression failure in the plain

weave fabric composite. The damage appears to be initiated

by interlaminar stress, generated by the waviness of ®ber.

These interlaminar stresses appear to cause local delamin-

ation in the sample, thereby reducing the local transverse

support for the ®bers. The ®bers in adjacent region of low

waviness were thus likely to be overloaded. Finally, global

microbuckling can be observed in these composites.

Delamination and global buckling are in fact a bending fail-

ure due to interlaminar stresses. Composites failing in this

mode yielded relatively low compressive strength.

The compressive failure in the composites made from

loosely stitched woven fabrics was comparable to that of

the unstitched fabric composites. However, the dense stitch-

ing appeared to have different failure mode. Fig. 3c shows

typical failure mechanism for the densely stitched biaxial

woven composite. Localized failure in the form of a kink

band can be seen in the crack initiation. Argon [24]

suggested that the regions in a composite in which ®bers

are not aligned with the compression axis would form a

failure nucleus that undergoes kink band formation.

4.5. Compression-after-impact strength

As discussed previously, impact damage area is a func-

tion of impact energy. The CAI strength is very much

dependent on the impact energy. Compression-after-impact

data at two energy levels (1.65 and 2.5 J mm21) are reported

in Table 5. It is seen that compressive strength of un-

damaged laminate is 15% higher than that of the composite

from woven fabric. However, after being subjected to

impact energy of 2.5 J mm21, CAI strength of woven fabric

composite was up to 37% higher than that of the laminate.

CAI of 5 mm spacing biaxial stitched woven specimen was

over 60% higher as compared to the laminate composites.

This dramatic reversal in the structural performance of the

glass±epoxy composite materials is consistent with

published results for graphite±epoxy composites [25]. The

specimens with stitched through-the-thickness reinforce-

ment exhibited higher CAI strength than the unstitched

woven fabrics did. Loose stitching was not much effective

in improving the CAI strength.

The macroscopic failure modes of CAI samples are

different from the compression behavior of undamaged

samples. Fig. 4a shows interlaminar splitting mode in a


Fig. 3. Compressive failure in: (a) non-crimp laminate composite; (b) unstitched woven composite; and (c) biaxial stitched (stitching lines 5 mm apart) woven

composite.

Table 5

Static compressive strength of glass ®ber±epoxy composites using IITRI test

Sample

codes

Density

(g cm23)

Fiber volume

(%)

Composite

thickness (mm)

Compressive

strength (MPa)

CAIa strength

(MPa)

CAIb strength

(MPa)

LM 1.90 53 4.41 438 ^ 10 176 ^ 7 139 ^ 4

PW 1.98 58 4.40 380 ^ 8 195 ^ 2 190 ^ 11

US 20 2.00 59 4.40 380 ^ 6 187 ^ 6 180 ^ 10

BS 20 1.90 53 4.58 386 ^ 16 192 ^ 10 ±

US 10 2.0 59 4.35 385 ^ 5 192 ^ 2 186 ^ 5

BS 10 1.93 55 4.55 398 ^ 8 256 ^ 8 192 ^ 11

US 5 1.95 56 4.40 400 ^ 9 223 ^ 1 207 ^ 10

BS 5 1.95 56 4.60 427 ^ 8 300 ^ 11 223 ^ 8

a Impact 1.65 J mm21.b Impact energy 2.5 J mm21.

laminate composite. Interlaminar crack is caused due to the

delamination in the impact damaged region. For unstitched

woven and loosely stitched sample, the CAI fracture is

mix-mode as shown in Fig. 4b. Crack was caused by the

buckling±bending close to the damaged region, followed by

debonding. Loose stitching has played no obvious role in

moderating failure propagation. Fig. 4c shows kink failure

for densely stitched sample. The kink band in the damage

sample propagated through the sample thickness at an

angle of 378 rather than the 458 angle in the undamaged

composite.

5. Conclusions

There is no clear evidence of cracking in the mid-plane of

woven glass ®ber±epoxy composites during the short beam

shear (SBS test). The grooved coupon test for interlaminar

shear strength shows that increasing density of Z-directional

stitching ®bers will moderately increase the delamination

resistance of the composites. Composites of stitched 2D

woven samples exhibited lower bending strength as

compared to the unstitched samples. Z-directional stitching

is effective in reducing the impact damage.

The compressive strength of the non-crimp laminate

samples is about 15% higher than that for the woven fabric

composite, a result of waviness in the woven fabric compo-

sites. At high stitching density, CAI has been signi®cantly

improved. Compression failure modes for the variety of

®ber arrangements are different. The experimentally

observed shear failure of ®ber across the specimen thickness

was the failure mechanism of laminate composites. Delami-

nation, followed by microbuckling and global buckling is

the failure observed in woven fabric composites. The

densely stitched composites appear to fail via kinking

followed by ®ber buckling.

CAI samples display different fracture mechanisms.

Interlaminar splitting is the dominant mode of fracture in

unstitched woven fabrics and laminate composites. Densely

stitched samples showed shear failure via kinking.

Acknowledgements

This work was supported by the US Department of

Commerce through the National Textile Center (Grant no.

99-27-07400).

References

[1] Miravete A. 3-D Textile reinforcements in composite materials. Boca

Raton, FL: CRC Press, 1999.

[2] Chou TW. Microstructural design of ®ber composites. Cambridge,

UK: Cambridge University Press, 1992.

[3] Guenon VA, Chou TW, Gillespie JW. Interlaminar fracture toughness

of a three-dimensional fabric composite. Proceedings of the Society

of Manufacturing Engineers, Dearborn MI 1987;EM-87-551:1±17.

[4] Liu CH, Chou TW. Mode II interlaminar fracture toughness of three

dimensional textile structural composites, Proceedings of the fourth

Japan±US conference on composite materials. Washington: Tech-

nomic Publishing Co, 1989. p. 981.

[5] Byun JH, Gillespie JW, Chou TW. Mode I delamination of a three-

dimensional fabric composite. J Comp Mater 1990;24:497.

[6] Guenon VA, Chou TW, Gillespie JW. Toughness properties of a

three-dimensional carbon-epoxy composite. J Mater Sci 1989;24:

4168.

[7] Kregers AF, Teters GA. Structural model of deformation of anisotro-

pic three dimensionally reinforced composites. Mech Comp Mater

1982;1:14.

[8] Ko FK, Pastore CM. Structure and properties of an integrated 3-D

fabric for structural composites, ASTM STP 864. Philadelphia, PA:

American Society of Testing and Materials, 1985. p. 428.

[9] West AC, Adams DO. Effect of yarn crimping on compressive

strength design allowables of braided composites, Proc Am Soc

CompÐXIth Tech Conf. Lancaster, PA: Technomic Publishing Co,

1996. p. 999.

[10] Naik RA, Ifju PG, Masters JE. Effects of ®ber architecture parameters

on deformation ®elds and elastic moduli of 2-D braided composites. J

Comp Mater 1994;28:656.

[11] Kozey VV, Kumar S, Adanur S, Mohamed M. Proc 26th International

SAMPE Tech. Conf. Covina, CA, 1994. p. 315.


Fig. 4. Failure in CAI samples: (a) non-crimp laminate composite; (b) unstitched woven composite; and (c) biaxial stitched (stitching lines 5 mm apart) woven

composite. Impact energy in each sample, 2.5 J mm21.

[12] Arendts FJ, Drechsler K, Brandt J. Proc 34th International SAMPE

Sym. Covina, CA, 1989, vol. 129. p. 2118

[13] Morales A. Proc. 22nd International SAMPE Tech. Conf. Covina,

CA, 1990.

[14] Yau S, Chou TW. Strength of woven-fabric composites with

drilled and molded, Composites materials: testing and design

(eighth conference), vol. 972. Philadelphia, PA: ASTM STP,

1988. p. 423.

[15] Wang Y, Zhao D. Mechanical and interlaminar fracture properties of

woven fabric reinforced composites. Proceeding of the 39th Interna-

tional SAMPE Symposium and Exhibition. Anaheim, CA, 1994.

[16] Adams DO, Bell SJ. Compression strength reduction in composite

laminates due to multiple layer waviness. Comp Sci Technol

1995;53:207.

[17] Kumar S, Adams WW, Helminiak TE. J Reinforced Plastics Comp

1988;7:109.

[18] Hahn HT, Sohi M, Moon S. NASA Contractor Report no. 3988, June

1986.

[19] Kozey VV, Jiang H, Mehta VR, Kumar S. J Mater Res 1995;10:1044.

[20] Kozey VV, Kumar S. J Mater Res 1994;9:2717.

[21] Do PB. Effect of textile preform structure on the mechanical proper-

ties and failure modes of composites from resin molding process. MS

thesis. Georgia Institute of Technology, 1994.

[22] Avery JG. Design manual for impact damage tolerant aircraft struc-

tures. AGAR DOGRAPH, no. 238, NATO, 1981.

[23] Dorey G. Fracture of composites and damage tolerance in practical

consideration of design, fabrication, and test for composites materials,

AGARD Lecture series 124, 1982.

[24] Argon JP, Guceri SI. Comp Sci Tech 1989;35:317±30.

[25] Madan RC. In¯uence of low-velocity impact on composite structures.

In: O'Brien TK, editor. Composite materials: fatigue and fracture,

ASTM STP 1110, 1991. p. 457.


bending, compression, and shear behavior of woven glass...

Documents