mechanical properties of stitched multiaxial fabric reinforced composites from mannual layup process

Applied Composite Materials 9: 81–97, 2002.© 2002 Kluwer Academic Publishers. Printed in the Netherlands.

81

Mechanical Properties of Stitched MultiaxialFabric Reinforced Composites From MannualLayup Process

YOUJIANG WANGSchool of Textile & Fiber Engineering, Georgia Institute of Technology, Atlanta,GA 30332-0295, U.S.A.

(Received 20 July 2000; accepted 19 December 2000)

Abstract. This paper summarizes the consolidation behavior of E-glass stitched multiaxial non-crimp fabrics (NCFs) and the mechanical properties of the resulting composites from the manuallayup process. NCFs offer many advantages over conventional woven and nonwoven fabrics interms of processibility and properties. The reinforcement structures studied include six NCFs withdifferent fiber orientation combination, number of sub-ply layers, and unit weight. The compressiveand recovery behavior of the dry preforms is evaluated and related to their processibility in the handlay-up process. Mechanical properties of specimens from the hand layup process are evaluated intensile, compressive, and flexural tests. The test results and failure mechanisms are reported. Theeffect of consolidation quality on the mechanical properties is discussed.

Key words: failure, manual layup, mechanical properties, multiaxial warp knitted fabric, NCF, non-crimp fabric, stitched multiaxial fabric, testing, textile preform.

1. Introduction

Manufacturing cost represents a significant portion of the overall costs of com-posites, and textile technology has been turned to for manufacturing cost reduction[1, 2]. The multiaxial warp knit (MWK) process is a new development which offersanother means of controlling the fiber architecture. The fabric structures consisttypically of two to four layers of straight fiber strands held together by a chain ortricot stitch through the thickness [3–5]. The process involves arrangement of fiberlayers followed by stitching [5]. The fibers in each layer can be oriented in thewarp (0◦), filling (90◦), or a bias direction (typically between 30◦ to 60◦). Unlikea woven fabric in which yarns are crimped due to interlacing, the multiaxial warpknitted fabrics preserve the unidirectional characteristics of each fiber layer. Suchfabrics are also known as non-crimp fabrics (NCF). Figure 1 shows a typical NCFfabric. A nonwoven layer can also be incorporated into the fabric. Such fabrics areavailable with different fibers such as glass, carbon, aramid, or custom blends.

NCF fabrics have good dimensional stability that allows them to be handledeasily in the composites manufacturing processes. The stitches allow relative fiber

82 YOUJIANG WANG

Figure 1. Schematic of a triaxial NCF.

movement in the fabric while at the same time maintaining uniform fiber spacing.The fabrics’ excellent conformability makes them suitable for making compositeparts of complicated shapes (e.g. parts with double curvatures) without excessivecutting, joining, and post-consolidation machining. Since multiple fiber layers arehandled in a single step, the composites manufacturing process is significantlysimplified. The mechanical properties of the NCF composites, especially in com-pression, may be superior to those of conventional woven fabric composites due tothe elimination of fiber crimp. A thin, textured polyester yarn is often used as thestitching yarn. A fabric’s dimensional stability and conformability can be altered bycontrolling the stitching yarn density or the stitching pattern. A high performancearamid or glass yarn may be used as the stitching yarn to improve the interlaminarproperties and damage tolerance of the composites.

Composites with NCF can be made by processes such as traditional lamination,resin transfer molding (RTM), pultrusion, vacuum bagging, centrifugal casting,and filament tape winding. In addition to traditional composites applications, NCFcomposites are experiencing growth in such emerging markets as marine, sports,transportation, and infrastructure applications [6].

The literature on the mechanical properties of NCF composites is very limited.In an earlier study, the mechanical properties of NCF composites from the resintransfer molding process were evaluated [7]. Due to their close fiber packing anddense structures, reasonably high fiber volume fraction (about 50%) in the finalcomposite part can be obtained even from the wet manual layup process. This

STITCHED MULTIAXIAL FABRIC COMPOSITES 83

Table I. Reinforcement fabrics.

Fabric Style Type Ply Orientation Unit Mass g/m2

EBX 2400 Biaxial NCF +45/−45 820

EBX 4800 Biaxial NCF +45/−45 1600

ELT 1200 Biaxial NCF 0/90 445

ETTX 2400 Triaxial NCF −45/90/+45 805

ETLX 7000 Triaxial NCF 0/−45/+45 2342

EQX 2600 Quadriaxial NCF 0/+45/90/−45 865

can result in good mechanical properties as well as significant savings in resinconsumption. This paper summarizes the consolidation behavior and mechanicalproperties of NCF composites from the manual layup, vacuum bagging process.

2. Reinforcing Fabrics

The NCFs used in this study were manufactured by the Johnson Industries Com-posite Reinforcements (formerly Tech Textiles USA), Phenix City, Alabama. Theyincluding biaxial (EBX 2400, EBX 4800, ELT 1200), triaxial (ETTX 2400, ETLX7000), and quadriaxial (EQX 2600) fabrics. (The first two digits in the manufac-turer’s style represent the nominal unit weight of the multilayer fabric in ouncesper square yard. For example, EBX 2400 has a nominal unit weight of 24 oz/yd2:multiply by 33.94 to convert to g/m2.) All the fabrics are made of E-glass fiberswith a multi-resin compatible surface finish. Table I summarizes the characteris-tics of these fabrics. For triaxial and quadriaxial NCFs, there are three and fourunidirectional fiber plies in the fabrics. Each ply in a fabric has an approximatelyequal mass fraction. The knitting yarns in NCFs are thin polyester yarns and thetotal mass of the knitting yarns is about 1% in the fabric. Actual fabric unit mass ismeasured for each fabric and reported in Table I.

3. Fabrication Process

The wet lay-up method is often used to make composite parts such as marine boatsfrom NCF fabrics. In this manual lay-up evaluation, an epoxy system is used as thematrix resin, which consists of EPON 815 resin and curing agent U (weight ratio =4 : 1). A layer of resin is applied to the flat mold or the previous laid fabric layers,and then a piece of dry fabric is added and roller pressed. After the correct numberof fabric layers is reached, the assembly is consolidated in a vacuum-bag. Thepanels are allowed to cure at room temperature for 48 hours before cutting. The sizeof the panel is approximately 330 × 380 mm. The panel thickness is determined

84 YOUJIANG WANG

Table II. Specimen layups, thickness, and fiber volume fraction.

Fabric Nf Laminate Layup t (mm) Vf (%)

EBX 2400 6 (45/−45//−45/45)3 4.03 47.8

ELT 1200 12 (0/90//90/0)6 5.45 38.3

ETTX 2400 6 (45/90/−45//45/0/−45)3 3.81 49.7

EQX 2600 6 (90/−45/0/45//45/90/−45/0)3 4.91 41.5

EBX 4800 3 (45/−45//45/−45//−45/45) 3.72 51.4

ETLX 7000 2 (0/−45/+45//−45/+45/0) 3.57 52.1

Nf = Number of fabric layers, t = panel thickness.

from the average thickness measured at about 80 points on the panel. The fibervolume fraction is calculated from

Vf = Fabric weight (g/cm2) × Number of fabric layers

Fiber density (g/cm3) × Panel thickness (cm).

Table II summarizes the layups for the test specimens which also includes thenumber of fabric layers in the panel (Nf), the panel thickness (t), and the fibervolume fraction (Vf) of the panel.

Like prepreg composites, it is desirable that the layup for NCF laminatesbe balanced and symmetric [7]. But because a group of layers are preassembled inan NCF, sometimes a symmetric layup is difficult or impossible to obtainfrom a given NCF. For example, a symmetric layup is not possible with a fourlayer [0/45/90/−45] quadriaxial NCF. For this reason the mirror image([−45/90/45/0] for the above example) is generally also made available. It can beseen from Table II that all the layups in this study are balanced. The layups for thetriaxial and quadriaxial fabrics are not symmetric due to the absence of their mirrorimaged fabrics in this study. Some of the fabric layers in a laminate are rotated by90◦ or turned over to arrive at the alternative ply orientations in the fabric.

The panel specimens were made in two batches. The first batch includes the firstfour panels listed in Table II, and the second batch includes the last two listed inTable II. The panels from the first batch consolidated with very good quality; theyshow a very consistent, uniform translucent appearance, indicating the absence ofa significant amount of voids. The second batch of panels were compacted by thevacuum bags to a slightly higher fiber volume fraction than the first batch, possiblydue to the difference in fabric construction or unintended variations in processingconditions. The panel quality is lower than the first batch, as white fiber strands notfully wetted with the resin can be seen on both surfaces. This improper wetting andexcessive void content has been found to have a pronounced effect on some of themechanical properties, which is discussed later in this paper.


4. Effect of Fabric Bulk on Consolidation Behavior

The compactness or bulk of a preform fabric influences the consolidation character-istics. In RTM and press molding, a bulky preform may have to be pre-consolidated(debulked) to allow its placement in the mold and to avoid excessive distortion ofthe fiber architecture in the consolidation process. In the manual layup process,a bulky fabric must absorb more resin to become saturated, resulting in low fibervolume fractions and inferior mechanical properties.

Fabric bulk can be characterized by measuring the compressive behavior.Wang [8] has proposed a fabric bulk factor and discussed its influence on man-ual layup for woven and nonwoven fabric reinforced composites. Using a Kawa-bata Evaluation System for textile fabrics testing, the fabric thickness (t) at 5 kPa(50 g/cm2) can be obtained. The bulk factor, k, is defined in the follow fashion:

k = t

t50%

t50% = 2 × 10−3 Fabric weigth (g/m2)

Fiber density (g/cm3),

where t50% is the thickness after the fabric is consolidated into a composite panelwith a 50% fiber volume fraction.

The bulk factor values for commonly used nonwoven fabrics range from 2.35to 4.6 with corresponding fiber volume fractions from 30% down to 14% whenconsolidated by manual layup [8]. The k values for NCFs studied are consistentlyclose to 1.25, except a 1.37 value for EQX 2600. As reported in Table II, the fibervolume fractions of panels from manual layup with various NCFs range from 38.3to 52.1%. The lowest Vf (38.3%) is for panels with 12 layers of the ELT 1200 fabric.The low Vf appears to be attributed to the large number of fabric layers involvedand the extra amount of resin trapped between them. EQX 2600 fabric, which hasa relatively high bulk factor (k = 1.37), also results in a low Vf at 41.5%. Manuallayup panels with all the other NCFs have a volume fraction close to 50%.

5. Mechanical Testing

Tensile, flexural, and compressive test specimens are cut from the cured panelsusing a diamond abrasive saw mounted on a horizontal milling machine. Speci-mens are cut along three directions: 0, 90, and 45 degrees. Actual specimen thick-ness is measured using a digital caliper for stress and stiffness calculations. Allthe tests are performed on an MTS 810 Tester at a displacement rate of 2 mmper minute. Figure 2 illustrates these tests. The tensile specimen is loaded us-ing Surfalloy-faced hydraulic grips without tabs. Unlike serrated grip faces, theSurfalloy faces are sandpaper like and cause little damage to the specimen sur-faces. The tensile specimen dimensions are 25.4 mm by 254 mm. The specimenlength between grips is 152 mm. A biaxial extensometer with a 25.4 mm gagelength is used to monitor the strains in the longitudinal and transverse directions

86 YOUJIANG WANG

Figure 2. Test configurations (Unit: mm): (a) Uniaxial tensile test, (b) Compression test,(c) Three-point flexural test.

in the tensile specimen. The flexural test is performed in general accordance withASTM D790-91 in three point bending. The flexural specimen width is 25.4 mm,and the specimen span length to thickness ratio is approximately 16. Compres-sion test is carried out to determine the compressive strength and modulus usingSurfalloy-faced hydraulic grips without tabs. The sample dimensions are 100 mmby 25.4 mm and the specimen gage length (between grips) is about 25.4 mm. Thedisplacement during the test was monitored using a clip gage transducer. All the testresults are recorded by the digital computer of the testing machine at a frequencyof 1 Hz. Elastic modulus and Poisson’s ratio are determined by data regressionanalysis.

6. Results of Mechanical Tests

Experimental results from tensile, flexural, and compressive tests are summarizedin Table III. Each test data point is the average of about four specimens.


Tabl

eII

I.E

xper

imen

talr

esul

ts.

Fabr

icV

fD

irTe

nsile

Com

pres

sive

Fle

xual

%E

(CV

)σ

(CV

)ν

(CV

)E

(CV

)σ

(CV

)E

(CV

)σ

(CV

)

GPa

%M

Pa%

%G

Pa%

MPa

%G

Pa%

MPa

%

EB

X24

0047

.80◦

10.3

(2.9

)10

4(3

.4)

0.57

(1.2

)7.

5(3

.5)

110

(4.6

)10

.9(8

.2)

251

(3.3

)

90◦

11.3

(1.7

)10

8(2

.0)

0.59

(3.0

)7.

7(4

.5)

114

(1.7

)11

.2(4

.9)

257

(1.2

)

45◦

23.3

(1.4

)24

0(4

.5)

0.13

(3.9

)15

.7(4

.6)

377

(3.3

)19

.0(6

.3)

380

(4.7

)

EB

X48

0051

.40◦

11.3

(5.4

)75

(6.5

)0.

56(1

.9)

6.5

(5.3

)84

(11.

4)7.

218

7

90◦

10.2

(8.8

)73

(5.4

)0.

577.

3(4

.2)

90(1

.6)

8.1

(1.3

)20

1(0

.2)

45◦

22.6

(0.9

)27

4(8

.1)

0.15

(4.1

)16

.1(3

.5)

254

(16.

4)17

.0(6

.2)

531

(7.8

)

ELT

1200

38.3

0◦17

.9(2

.4)

235

(8.0

)0.

1513

.1(3

.9)

330

(3.9

)15

.4(5

.7)

359

(8.0

)

90◦

18.6

(2.9

)21

5(8

.6)

0.16

(2.0

)13

.7(5

.6)

336

(3.6

)16

.8(2

.4)

414

(4.8

)

45◦

10.1

(1.4

)12

8(6

.2)

0.54

(0.8

)7.

1(8

.8)

110

(2.0

)9.

1(2

.2)

211

(4.3

)

ET

TX

2400

49.7

0◦17

.3(1

.9)

239

(2.6

)11

.6(4

.9)

270

(1.7

)16

.4(4

.3)

423

(5.1

)

90◦

16.5

(2.0

)23

1(2

.9)

0.38

(2.5

)12

.2(6

.9)

256

(6.1

)14

.9(7

.3)

440

(5.0

)

45◦

21.4

(1.6

)33

7(2

.5)

0.26

(3.8

)14

.3(4

.5)

353

(6.3

)17

.2(0

.9)

503

(3.0

)

ET

LX

7000

52.1

0◦18

.022

1(4

.1)

0.56

6.0

(1.3

)19

8(0

.4)

18.3

(6.1

)57

0(4

.9)

90◦

11.1

(4.9

)88

(6.6

)0.

447.

5(7

.6)

127

(4.0

)9.

1(5

.6)

221

(11.

0)

45◦

21.5

(4.4

)27

3(4

.8)

0.19

(2.2

)12

.2(1

2.8)

272

(5.8

)11

.1(8

.7)

352

(5.9

)

EQ

X26

0041

.50◦

16.9

(1.0

)22

5(5

.9)

0.32

(0.7

)12

.1(2

.9)

313

(2.1

)15

.0(2

.2)

356

(4.8

)

90◦

16.7

(1.5

)21

7(4

.5)

0.31

(3.8

)11

.4(5

.6)

306

(3.8

)14

.8(2

.3)

368

(4.0

)

45◦

16.8

(2.9

)23

0(2

.9)

0.32

(2.7

)12

.0(2

.4)

323

(1.1

)12

.9(5

.2)

384

(6.2

)

E=

mod

ulus

,σ=

stre

ngth

,ν=

Poi

sson

’sra

tio,

CV

=co

effi

cien

tof

vari

atio

n.

88 YOUJIANG WANG

Neglecting the differences in the stacking sequence, a laminate may have thesame fiber orientation pattern when viewed along several discrete axes. For ex-ample, along both 0 and 90 degree directions, the reinforcing fibers are at ±45◦in a biaxial NCF laminate. The test results are very consistent for cases whensuch axes coincide with more than one of the test directions, viz., 0, 90, and 45degrees. Similar property values are observed for all the biaxial laminates (EBX2400, EBX 4800 and ELT 1200) along 0 and 90 degree directions, for the ETTX2400 triaxial laminate along 0 and 90 degree directions, and for the EQX 2600quadriaxial laminate along 0, 90, and 45 degree directions. However, it should benoted that stacking sequence can have a pronounced effect on the strength andfailure mode, and even on the stiffness under flexural loads [9].

6.1. TENSILE TEST RESULTS

Typical tensile stress–strain curves for selected laminates along 0◦, 90◦ and 45◦directions are shown in Figure 3. During all the tensile tests, a quiet poppingsound could be heard even at relatively low loads, which probably corresponds tomicro-failures such as matrix cracking and fiber debonding. The final failure is ac-companied by a single or multiple loud snapping sound. Although NCF compositesare more resistant to delamination than traditional woven fabric composites, delam-ination is still often observed in the failed specimens. Yarns in a woven fabric arecrimped and follow out-of-plane undulations. Delamination occurs when the yarnsare being straightened, forming a checkerboard pattern on the surface [10]. Thecause of delamination in NCF composites, on the other hand, is mainly interlaminarshear, similar to a prepreg laminate.

The stress–strain curves for the EBX 2400 biaxial laminate are given in Fig-ure 3a. As expected from the fiber orientations, distinctive tensile stress–strainresponses are observed along a fiber direction (45◦) and along the bias directions(0◦ and 90◦). In the fiber direction, the response is near linear, though gradual soft-ening is evident as the load increases. Final failure is associated with delamination,especially along the longitudinal fiber bundles. When loaded in the bias directions,significant nonlinearity in the stress–strain response starts at relatively low load,and the load remains near constant as the specimen extends before its final failureat a strain of about 9%. The nonlinear effect is likely the result of progressivefailure due to various interacting micro-failure modes, such as matrix shearing,cracking, fiber debonding, fiber pullout and interply tearing. Delamination over thegage length section can be seen, though the final failure with severe delamination isconcentrated within a short section, as shown in Figure 4a. Other biaxial laminatesshow similar tensile responses and failure mechanisms along the fiber and biasdirections.

Figure 3b shows the tensile test curves for the ETTX 2400 triaxial laminate.Fibers in the laminate are distributed along the loading, perpendicular to the load-ing, and in the bias directions for all the tests along the three directions. Though


(a)

(b)

(c)

Figure 3. Typical tensile test curves for NCF composite laminates: (a) EBX 2400 Biaxial,(b) ETTX 2400 Triaxial, and (c) EQX 2600 Quadriaxial.

similar stress–strain responses are observed along the three directions, the laminateis stiffer in the 45◦ direction since 33% of fibers are in this direction whereas only17% fibers are parallel to the 0◦ or 90◦ direction. The nonlinearity in the stress–strain responses is more pronounced in this triaxial laminate than in the biaxiallaminates along a fiber direction. This is due to the fact that fewer fibers in thetriaxial laminate are along the loading directions, and the cross and bias plies play

90 YOUJIANG WANG

(a)

(b)

(c)

Figure 4. Typical failure pattern of tensile specimens: (a) EBX 2400 Biaxial, along 0◦ or90◦ direction (fibers are ±45◦ to the loading direction), (b) ETTX 2400 Triaxial, along 0◦direction, and (c) EQX 2600 Quadriaxial, along 90◦ direction.

a larger role. The failure is dominated by delamination concentrated within a shortsection (Figure 4b).

As illustrated in Figure 3c, the tensile responses for the EQX 2600 quadriaxiallaminate are near identical along the three test directions. This is expected be-cause the laminate has a quasi-isotropic layup and it should have the same in-planestiffness along any direction. The fiber distribution also remains unchanged with


respect to three loading directions and thus similar strengths along them are alsoexpected. However, the strength along other directions may be significantly differ-ent than along the three test directions. Had the laminate been tested along the 22.5◦direction, a lower strength and more pronounced stress–strain nonlinearity wouldbe expected. In most cases the final failure due to delamination is within a smallregion rather than throughout the entire gage length. In several cases, however, anumber of discrete delamination regions are developed along the edges before thefinal failure, as illustrated in Figure 4c.

6.2. COMPRESSIVE TEST RESULTS

The compressive moduli measured are lower by about 30% than their respectivetensile values, as seen from Table III. This could partially be due to experimen-tal errors, as the small specimen gage length in compression hinders mountingthe transducer directly on the specimen. Tne compressive strengths are generallysimilar to or higher than their tensile values. Typical compressive test curves areshown in Figure 5. Ductile failure is observed for laminates with all the fibersoriented at ±45◦ to the loading direction. Delamination associated with in-planeshear deformation is evident, and the ×-shaped failure bands are along the ±45◦fiber directions as seen in Figure 6a.

A higher compressive than tensile strength is generally observed when testedalong a fiber direction. In such cases the laminates showed a brittle failure witha sudden load drop after reaching the maximum load. Examination of the failedspecimens reveal that the dominant mode is delamination, and the single failureband is perpendicular to the loading direction. Figure 6b shows the compressivefailure pattern typical of all the laminates tested along a fiber direction. In con-trast, for a woven composite with yarn undulations, it is easier for a ply to buckleunder compressive load. It has been observed that compressive failure in wovencomposites is initiated in the crimped yarn along the loading direction [11].

6.3. FLEXURAL TEST RESULTS

The flexural modulus is similar to their respective tensile modulus, as reported inTable III. The flexural strengths of all the laminates tested are significantly higherthan their tensile strengths, a trend similar to that reported for prepreg laminates[12, 13]. The flexural strengths are also higher than or similar to their compressivestrengths. The flexural load-deflection responses, shown in Figure 7, exhibit lessnonlinearity than the tensile and compressive responses. All the laminates testedalong a fiber direction show a brittle failure, generally by outer ply delamination ona tensile surface. The delamination zone starts at the middle of the specimen wherethe bending moment is maximum, then propagates outwards until significant fiberrupture occurs at the middle section on the tensile surface. This typical failure mode

92 YOUJIANG WANG

(a)

(b)

(c)

Figure 5. Typical compressive test curves for NCF composite laminates: (a) EBX 2400Biaxial, (b) ETTX 2400 Triaxial, and (c) EQX 2600 Quadriaxial.

can be observed in Figure 8a for a biaxial specimen tested along a fiber direction. Incontrast, laminates with all fibers oriented at ±45◦ to the specimen length directionshow a ductile failure, as in tension and compression. Delamination along the fiberstrands can be observed on both the tensile surface (Figure 8b) and the compressivesurface (Figure 8c). Large deflection is reached before the final failure with fiberrupture occurs.


(a)

(b)

Figure 6. Typical failure pattern of compressive specimens: (a) EBX 2400 Biaxial, along 0◦direction (fibers are ±45◦ to the loading direction), and (b) EQX 2600 Quadriaxial, along 0◦direction.

6.4. EFFECT OF CONSOLIDATION QUALITY ON MECHANICAL PROPERTIES

As mentioned earlier, the two panels with EBX 4800 and ETLX 7000 fabrics haveinferior quality, manifested by incomplete fiber wetting. Excessive void content canalso result if the liquid resin is not properly degassed in a wet layup process. In-consistency in consolidation quality is not uncommon in the manual layup process.For glass fiber reinforced neat resin composites, fortunately, it is fairly easy todetermine the consolidation quality, as a well-consolidated part has a uniform,translucent appearance, and the transparency would be reduced if air voids arepresent. It is more difficult and it may require other techniques to assess the con-solidation quality of composites with opaque fibers such as carbon or aramid, orcomposites with resin containing fillers.

It is interesting to note that both the panel with EBX 2400 and that with EBX4800 have the same fiber orientation pattern. When tested along the 45◦ directionwhere the fibers are 0/90 degrees to the loading direction, both panels show similar

94 YOUJIANG WANG

(a)

(b)

(c)

Figure 7. Typical flexural test curves for NCF composite laminates: (a) EBX 2400 Biaxial,(b) ETTX 2400 Triaxial, and (c) EQX 2600 Quadriaxial.


(a)

(b)

(c)

Figure 8. Typical failure pattern of flexural specimens with EBX 2400 Biaxial NCF: (a) testedalong 45◦ direction, tensile surface (fibers are 0◦ and 90◦ to the loading direction), (b) testedalong 0◦, tensile surface (fibers are ±45◦ to the loading direction), and (c) tested along 0◦,compressive surface.

96 YOUJIANG WANG

tensile behavior. However, when tested along 0 and 90 degree directions wherefibers are ±45◦ to the loading direction, the well consolidated EBX 2400 specimensshow a much higher strength than the EBX 4800 specimens which contain not fullywetted fiber tows. This is expected as in this case the load must be carried by thefiber-matrix interface, whereas in the previous case the load is basically carried bythe fibers in the loading direction only.

Under compressive loading, the critical load against buckling is decreasedsharply if the cross section of a column is partitioned into several smaller columns.Imperfect wetting could in effect reduce the integrity of the specimen cross section.Splitting tensile stress perpendicular to the compressive load could strengthen thedisintegration of the specimen under a compressive load. This appears to be thecase for EBX 4800 and ETLX 7000 panels with imperfect wetting. The compres-sive strength for the two panels along a fiber direction is lower than the tensilestrength. For the well consolidated panels, by contrast, the compressive strengthalong a fiber direction is higher than the respective tensile strength. The loweredcompressive strength for EBX 4800 and ETLX 7000 panels does not seem to havean adverse effect on the flexural strength. This is probably because in the flexuraltest the compressive stress is at a maximum only at the center of the specimen, andcompressive ply buckling does not play a major role. However, when subjectedto pure bending or when a much thicker composite part is under a three-pointflexural load, compressive failure could become the limiting factor, as the effectivelength under high compressive stress is longer than that of a thin beam subjectedto three-point bending.

Since some of the tests do not properly reflect the consolidation quality ofthe specimens, and in applications a composite part may be subjected to complexloading conditions, it is important to obtain all the necessary test data for designingcritical composite structures.

7. Summary and Conclusions

We have studied the properties of composite laminates reinforced with stitchedmulti-layer, multi-axial non-crimp fabrics (NCFs). NCFs exhibit excellent con-formability and therefore are well suited for composite parts ranging from flatpanels to complicated shapes. A single layer of NCF may contain several pliesof straight fibers oriented in different directions, and therefore the behavior ofan NCF composite can be easily tailored to meet specific needs. NCF fabrics aredensely packed structures, and reasonably high fiber volume fraction (about 50%)in the final composite part has been obtained from the wet manual layup process.This can result in good mechanical properties as well as significant savings inresin consumption. A method to characterize the bulk of preform fabrics has beendiscussed.

Composite panels with six NCF fabrics have been fabricated using the manuallayup process, a method often used to make NCF composites. The panels are tested


in tension, flexure, and compression along three directions: 0, 90, and 45 degrees.The experimental results provide useful design data for applications. The failuremode is discussed in relation to the panel layup, testing condition, and the direc-tion of loading. The strength of the laminates is also found to be affected by thetesting condition. Generally the highest strength is observed in flexure, followedby compression and tension. Consolidation quality has a pronounced effect on thecompressive strength and off-axial tensile strength of composite laminate, but lessof an effect on the tensile and flexural strength along a fiber direction. Care mustbe taken in experimental evaluation to reflect the actual loading condition of thecomposite part.

References

1. Vinson, J. R. and Sierakowski, R. L., The Behavior of Structures Composed of CompositeMaterials, Martinus Nijhoff Publishers, Dordrecht, The Netherlands, 1987.

2. Chou, T. W. and Ko, F. K., Textile Structural Composites, Elsevier, New York, 1989.3. Kaufmann, J. R., ‘Industrial Applications of Multiaxial Warp Knit Composites’, in Proceedings

of Fiber Tex Conference, Raleigh, NC, 1991, pp. 77–86.4. Ko, F. K., ‘Three Dimensional Fabrics for Composites’, in Ref. 54, pp. 129–171.5. Horsting, K., Wulhorst, B., Franzke, G., and Offermann, P., ‘New Types of Textile Fabrics for

Fiber Composites’, SAMPE Journal 29, 1993, 7–12.6. Lindsay, K. F., ‘Stitched and Braided Fabrics: Engineered Reinforcements’, Composites Design

& Applications, Spring 1996, 12–14.7. Wang, Y., Li, J., and Do, P. B., ‘Properties of Composite Laminates Reinforced With Knitted

Multi-layer Non-Crimp Fabrics’, Journal of Composite Materials 29(10), 1995, 2317–2333.8. Wang, Y., ‘Effect of Consolidation Method on the Mechanical Properties of Nonwoven Fabric

Reinforced Composites’, Applied Composite Materials 6, 1999, 19–34.9. Halpin, J. C., Primer on Composite Materials Analysis, Technomic Publishing Co., Lancaster,

1992.10. Wang, Y. and Li, J., ‘Properties of Composites Reinforced With E-Glass Nonwoven Fabrics’,

Journal of Advanced Materials 26(3), 1995, 28–34.11. Kozey, V., Kumar, S., Adnur, S., and Mohamed, M. H., ‘Compressive Failure in 2D and 3D

Woven and Laminated Glass/Epoxy Composites’, in Proceedings of the 26th InternationalSAMPE Technical Conference, October 17–20, 1994, Atlanta.

12. Bullock, R. E., ‘Strength Ratios of Composite Materials in Flexure and in Tension’, Journal ofComposite Materials 8, 1974, 200.

13. Whitney, J. M. and Knight, M., ‘The Relationship Between Tensile Strength and FlexureStrength in Fiber-Reinforced Composites’, Experimental Mechanics 20, 1980, 211.

mechanical properties of stitched multiaxial fabric reinforced composites from mannual layup process

Documents