6 views

Uploaded by khayat

Composites Handouts 2008-1

- 10.5923.j.textile.20120104.02
- FEM
- Design Standard - Section 3-3 - Steel Storage Tanks -12!31!12 - FINALWEBPOST_201301090858057238
- MIL-R-7705B
- Introduction to Composite Materials
- WEST SYSTEM Launches G Flex Epoxy
- Aci 351.1 R-99 Grouting Bet Foundn & Bases for Support of Equipment & Machinery (PDF.io)
- pro_fiberglass_GB_SCREEN.pdf
- Endura Resin Chip System
- Utility fuzzy multiobjective optimization
- Novolac Screed
- ch08.pdf
- 693B CF Orange Hardener Ratio Guide - Steel
- Molecule Design
- Nathan a Laurel
- ROM composite
- DS_066_Epilux_84_Clear
- 3M-460DP Joining Adhesive
- A Review On Mechanical Properties Of Natural Fiber Reinforced Hybrid Composites
- Coating Material for Tendon for Pre Stressed Concrete

You are on page 1of 91

Table of content:

2- Basic Micromechanics

7- Fabrication techniques

9- Interlaminar stresses

Laurent Warnet

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.1

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1.1. Introduction

A composite material is in its broadest definition a combination of two or more

constituents. In general, it combines the characteristics of these components in

order to obtain properties, which could not be obtained with the separate

constituents.

In this Composites Course we will restrict ourselves to continuous fibre reinforced

plastics. In these composite materials, the fibres are used for their high stiffness

and strength, while the matrix protects, binds the fibres together and transfers load

between the fibres.

The form in which fibres can be combined with a matrix is pretty infinite. Every

composite is tailored for a certain application. Even before a composite is used as

a component in a mechanical structure, it is already a structure. It would be

therefore more appropriate to talk about composite structure instead of composite

material.

Ply (a)

Laminate (b)

We will focus on composites that are made as a laminate form, i.e. a stacking of

(different) layers (or plies). This is illustrated in fig. 1.1 for layers made of

continuous fibre aligned in a unidirectional way. The laminate used in this

example is made of 3 layers where the fibres are oriented in different directions.

As far as constituents are concerned, we will focus on fibre-reinforced polymers,

which are used in laminate form.

The fibres can be organic as well as anorganic. They are manufactured by some

kind of drawing or spinning process, whereas the polymers are synthesised from

monomers.

1

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.2

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1.2. Fibres

A few types of fibre will be described next. The form in which it is used as a

reinforcement material will be dealt with in the manufacturing section of this

chapter.

Glass fibre is the most popular type of fibre in the composite world. Glass fibre is

stronger than metals in bulk form by virtue of the fact that its structure contains

fewer flaws or defects. The ultimate tensile strength of a freshly drawn single

glass filament (diameter 9-15 m) is about 3.5 GPa. Fibreglass filaments became

commercially available in 1935. Glass reinforced polyesters were used in the

construction of aircraft radomes in the early 1940s. In the 1950s these materials

were applied widely in plastic boat hulls, car bodies and truck cabs.

All glass fibres are based on Si-oxydes. Addition of other oxides, like, B, Al, Na,

K, Ca and Mg for different applications. Production of glass fibres starts from

melt around 1300-1500C (higher for S-glass). The molten glass is driven by

gravity through small orifices. It is rapidly cooled down and pulled at high

velocity (65m/s) to form a fibre. Surface micro-damages, which have a negative

effect on strength, are induced during processing. A rubber roller applies a

subsequent coating with a protective aqueous size solution in order to reduce the

effect of surface micro-damages. This treatment also contains a coupling agent,

which promotes the chemical compatibility of the matrix and the fibre. The

diameter of the obtained glass fibre (for reinforcement) is typically 9 to 15m.

The internal structure of glass is a 3D long network of silicon, oxygen and other

atoms arranged in random fashion. Glass fibres are therefore amorphous and

isotropic.

The most used fibre is E-glass, originally developed for electronic applications

(good isolation). Others are C-glass, which has better chemical resistance. S-glass

has a higher strength, but is costly to produce. D-glass is an example of a tailor

made fibre, and it finds applications in radars.

Summarising in a few words:

- The most popular

- Isotropic properties

- Elasticity modulus lower than carbon or aramid.

- Brittle fibre, its surface contains lots of microdefects.

Edison produced the first carbon fibres (1879) for electric lamps, by careful

carbonisation of cellulose strands. Bamboo or cotton was used as a starting

material, leading to extremely brittle filaments. The second-generation carbon

fibres (from 1963 onwards) are miles apart from their predecessors and have

exceptionally high strength and stiffness. Carbon fibres are based on

polyacrylonitrile (PAN) or pitch (by-product of petroleum refining). The PAN

precursor is stretched to obtain molecular and structural orientation.

2

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.3

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

letting the carbon-containing

precursor fibre going through

different thermal treatments:

oxidisation at around 200-

400C in air, carbonisation in

N2 around 1000C (carbon

fibre), and possibly

graphitisation in Ar from fig. 1.2: Microstructure of carbon fibres on basis of PAN

1500C (graphite fibre).

Carbon fibres have a carbon

content of 95%, graphite fibres at least 99%. The fibre can be subsequently

oxidised to promote adhesion to matrix resins, and coated with a sizing resin to

protect them during handling. The typical diameter of a carbon fibre is around 5 to

7m. It is important to note that the structure obtained is a mix of amorphous

carbon and graphitic carbon. This last form where the atoms are arranged in

parallel plan provides the high tensile modulus. A strong covalent bond exists

within these planes, but the planes are held together with weak van der Waals-type

forces (see fig. 1.2). This results in highly anisotropic properties.

Summarising in a few words:

- High specific elasticity modulus

- Anisotropic properties

- Negative coefficient of thermal expansion

- Electrically conductive

- Structures made with carbon are generally sensitive to impact and

therefore used in combination with aramid fibre.

The aramid fibre is one of the organic fibres. Aramids are highly crystalline

aromatic polyamides. The aramid fibre (Kevlar from DuPont as introduced in the

early 1970s, Twaron from Akzo, now Teijin) is obtained by spinning the polymer

into thin threads, which are then stretched to increase the stiffness. The stretching

results in a high molecular orientation in the longitudinal direction. Their

properties are therefore highly anisotropic. The fibre diameter is around 12m.

Summarising in a few words:

- Low density (1450kg/m3) and high specific strength.

- Not recommended in structures with high compressive loads.

(compression strength is 20% of tensile strength)

- Difficult to process

- Good thermal stability (decomposition at 450oC)

- Ductile fibre. Often used in applications where good energy

absorption is necessary.

The PE fibre is a second example of an organic fibre. Ultra-high molecular weight

polyethylene is dissolved in a solvent and then spun through small orifices

(spinneret). Successively, the spun solution is solidified by cooling, which fixes a

molecular structure which contains a very low entanglement density of molecular

3

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.4

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

chain. This structure gives an extremely high draw ratio and results in extremely

high strength. The highly drawn fibre (Dyneema from DSM, patented in 1979,

Spectra from Honeywell) contains an almost 100% crystalline structure with

perfectly arranged molecules, which promotes its extremely high strength,

modulus, and other excellent properties.

Summarising in a few words:

- Low density (950kg/m3). Further similar to aramid fibre but:

- Low upper-limit temperature (120oC)

- Low moisture absortion (1% compared to 5% for Kevlar)

The table below presents some typical properties of selected fibres. The properties

can vary from one type to another, depending on the actual manufacturing

process.

Diameter Modulus Strength Failure Thermal

Expansion

(m) (kg/m3) EfL/EfT (GPa) max(GPa) max (%) (10-6/C)

Glass

E-glass 10 2540 72 / 72 3.451 4.81 5

S-glass 10 2490 87 / 87 4.31 51 3

PAN Carbon

2

T-300 7 1760 230 / 20 3.53 1.5 -0.7

3

IM-7 5 1780 300/ ?? 5.31 1.31

Pitch Carbon

4

P-55 10 2000 380 / ?? 1.9 0.5 -1.3

4

P-100 10 2150 690 / ?? 2.4 0.32 -1.4

Aramid

5

Kevlar 49 12 1450 131 / ?? 3.6 2.8 -2

Polyethylene

Spectra 900 38 970 117 2.6 3.5

1

Literature value of the bare fibre under ideal circumstances. Contact with any substrate reduces the

value by 50%.

2

Torayca

3

Hercules

4

Amoco

5

DuPont

4

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.5

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

3.50

3.00

PAN

IM7

Specific strength (x10 m /s )

2

Polyethylene

2.50

2

Aramide

6

2.00

PAN

S-Glass T300

1.50

E-Glass

1.00 Pitch

Pitch P1004

P55

Steel wire

0.50

0.00

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50

8 2 2

Specific modulus (x10 m /s )

fig. 1.3: Specific stiffness vs. specific strength for various fibres.

No matter in what form the fibre is used, it is hardly possible to use it as a single

component in any structure. A matrix is necessary to

1) transfer the load to the fibres and

2) protect the fibre from the environment (chemical-mechanical

aggression).

The matrix plays a minor role in the load carrying capacity of the composite in the

fibre direction. A typical elasticity modulus of an average matrix material

compared with that of an average Carbon fibre is 3GPa against 230GPa. However,

the matrix plays a major role for the in-plane shear properties of the laminate

(important for torsion) as well as its interlaminar shear properties (important for

bending properties). In addition, the physical interaction between the fibre and the

matrix (mostly called interface) plays an important role in the transfer loading as

well as for the way the laminate absorbs energy.

In general three types of plastics are distinguished:

1. Thermoplastics are linear or branched macromolecules, which are held

together by weak secondary bonding forces. They soften when they are

heated.

2. Thermosets are cross linked by many valency bonds, leading to a rigid

(and unsoluble) three-dimensional network which will not be affected by

heat until the temperature is high enough to break up the molecule. Usually

the formation of such network polymers requires heat.

3. In elastomers the molecules are linked together by a small number of

valency bonds, leading to a loose network behaving as a rubber-like

material with more elastic properties than a simple thermoplastic.

The history of synthetic polymers starts with Alexander Parkes (UK), who made

an ivory-similar material in 1860 but did not succeed in making his Ivoride or

Parkesine a commercial success. John Hyatt (USA, 1870) was commercially

5

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.6

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

successful with this Celluloid, using it as a coating for billiard balls (motivated

by the threat of a shortage of elephants). Currently, this early thermoplastic

material is mainly used for table-tennis balls. The thermosetting family of

polymers is stemming largely from the work of Leo Baekeland, with his first

patent on phenol-aldehyde plastics in 1907. He formed his General Bakelite

Company in the USA in 1910. Urea-formaldehyde (UF) compounds were

introduced in 1926, followed by melamine-formaldehyde (MF) materials in 1935.

Well-known thermoplastics were developed in the same decade: polystyrene (PS)

in 1930, polyvinylchloride (PVC) and polyamide (PA) in 1938, low density

polyethylene (LDPE) in 1938. More thermosets were introduced in the 40s:

unsaturated polyester (UP) in 1941, epoxide resins starting from a patent by

Farben in 1934, leading to CIBAs Araldite in 1946. In the 50s further

thermoplastics were developed: high density polyethylene (HDPE) in 1955,

polycarbonate (PC) in 1956 and polypropylene (PP) in 1957. Technical

thermoplastics are available since the seventies, such as Polyphenylene Sulphide

(PPS) in 1973, Polyetherimide (PEI) in 1982, Polyetheretherketone (PEEK) in

1987.

groups. The functionality of a molecule is defined as the number of reactive

groups. A functionality of more than 1 is required to produce polymers. Appendix

A summarises a number of molecular groups in polymer matrix materials.

1. Polymerisation: a chemical reaction, generally carried out in the presence

of a catalyst, which combines small molecules (monomers), containing a

double bond, into long chain molecules. The double bond is opened up,

thereby making valency bonds available for linking with its neighbouring

monomer molecule. No by-products are produced.

Example: polyethylene

H H H H

n C C C C

H H H H

n

basic units which have at least two functional groups. It gives rise to the

elimination of small, low molecular weight by-products such as water,

hydrochloric acid, etc.

6

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.7

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

OH OH

O O

O O n

brought together. No by-products are eliminated, but hydrogen atoms

migrate from their positions in the functional group leaving combinable

free valencies.

R OH

O

R H + C C C C

(we will leave elastomers aside, although car tyres and reinforced rubber tubing

meet the definition of a polymer matrix composite). Thermosets are mostly

synthesized in a multistep process of subsequent condensation and addition

reactions. The formation of a cross-linked 3D molecular network requires more

than two functional groups in at least one of the reactants. A major difference with

thermoplastics is that once solidified, subsequent heating cannot reshape the

thermosetting matrix. Thermoplastics can be repeatedly softened by increasing the

temperature and hardened by decreasing the temperature. The next points put the

emphasis on some advantages/disadvantages of the two types of matrix.

Melt viscosity: Very low for the thermosets, high for most thermoplastics.

Low viscosity is important for a good fibre impregnation. The use of

thermoplastics therefore requires fabrication methods designed specifically for

this purpose.

Processing cycles: Longer for thermosets than for thermoplastics. The final

hardening of a polyester based composite can take days or weeks.

Fracture toughness: Higher for thermoplastics. This item was, together with a

potential reduction in processing cycle, the argument to investigate the use of

7

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.8

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

combination with other phases (rubber, thermoplastics).

Fabrication costs: Depends on the type of composites with thermosets (from

high in aerospace application to low in for example glass fibre pipes or tanks).

Are potentially low for thermoplastic. However, efficient production

techniques still need to be developed.

Chemistry: Complex for the thermosets. Reproducibility in material properties

is therefore difficult. Comparably straightforward for thermoplastics

Residual stresses: Residual stresses are inherent to any composite structures.

In thermosets, their origin is three-fold: thermally, hygroscopically (absorption

of water) or chemically (when cross-linking) induced. No chemical shrinkage

occurs when producing with thermoplastics, but crystallisation can play a role

in semi-crystalline materials.

1.3.1. Thermosets

Epoxy resins

Epoxy resin is a generic O O

term that covers a wide

C C CH2 CH CH2

range of crosslinked

polymers, which are based

on polymers containing

epoxide group and the glycidyl group

epoxide groups. Often,

epoxide resins are manufactured from glycidyl chloride (epichlorohydrin,

ECH). The epoxide resins will react, in general, with compounds containing

active hydrogen atoms. A typical example of the polymerisation of epoxy is

the reaction between a diepoxide (functionality 2) with a primary diamine

(functionality 4).

O O

O OH OH O

N R' N

O OH OH O

linked network, when the parent resin is such a diepoxide.

8

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.9

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

The nature of the epoxide resins and the reagents causes the molecular

structure of the crosslinked epoxy resins to differ. Consequently, the

mechanical properties and the useful temperature range of epoxies, as well as

processing viscosity can vary over a wide range. Curing conditions and

therefore the cross-link density also have an important effect on the

mechanical properties. For example, the complete reaction of the epoxide

groups can only be achieved after long time at high temperature. This is

mostly achieved by post-curing and is important for an increase in Youngs

modulus, chemical resistance and useful temperature range.

Polyester resins

The base for this thermoset is a linear unsaturated polyester resin dissolved in

styrene monomer with added catalyst (often a peroxide), accelerators and

inhibitors. Usually, the process of cross-linking is very slow at room

C CH CH C O R O + CH CH2

O O styrene

unsaturated polyester

CH

CH2

C CH CH C O R O

O O

up starts the process of cross-linking. During this curing reaction the styrene

reacts with the carbon-carbon double bonds of the polyesters to build up a

three-dimensional (cross-linked) polyester network. Accelerators can also

induce cure at room temperature. As for the epoxy resins, the properties of

polyesters depend on the cross-link density; they are available in a wide range.

Unsaturated polyesters are often applied in hand lay-up, sheet and bulk

moulding compounds (SMC and BMC respectively). Advantages compared to

epoxies are their low viscosity, fast curing time and lower cost. However, their

material properties are generally lower. The major drawback is their chemical

shrinkage. The volumetric shrinkage of polyester during the curing process

can be up to 8%. Apart from this chemical shrinkage, thermal shrinkage can

occur when heat is used during curing. It is worth adding that even for cold-

cure resin, the curing process can produce heat and therefore can induce extra

local shrinkage of the matrix. In recent years, the emission of styrenes during

processing puts a heavy restraint on processing polyesters.

9

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.10

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

EVERs (or Vinyl Esters) consist of an epoxy backbone with vinylester

end groups.

epoxy backbone

ester group

vinyl group O CH3 O

C C

The vinyl esters have a very good chemical resistance and a higher

temperature range than their polyester counterparts. They also offer a low

viscosity and fast curing, better than the epoxy systems. However, the

chemical shrinkage is higher than the epoxies, in the order of 5-10%.

Phenolics were developed in the early 20th century. These brittle materials are

suitable for an elevated temparture range (up to 200C) and offer excellent

flame retardance. Bismaleimides and other thermoset polyimides offer again a

higher thermal resistance: up to 370C continuous temperature for certain

Polymerised Monomer Reactants. Cyanate esters also offer a high temperature

resistance, in addition to low moisture absorption and excellent hot-wet

performance.

1.3.2. Thermoplastics

The molecules in a polymer are of varying sizes. In a thermoplastic, these are so

tangled together that instead of a sharp melting point to an easy flowing liquid,

they soften over a much wider range of temperatures to form a melt which will

flow with difficulty. The macromolecules can be linear or branched. If the

polymeric molecules have few branches which increase the distance between the

main chains, then the parallel lengths of the chains can form crystalline regions.

Due to the limited mobility of the polymer chains, a thermoplastic will never

reach a fully crystalline state: the material is at most semi-crystalline. Multi-

branched molecules cannot be close together so there will be less crystallinity and

the density will be lower. Polymers with low crystallinity are called amorphous.

Only some of the thermoplastic matrices, which are used as a basis for continuous

fibre reinforced composites (high performance), are shortly reviewed here.

with long (50mm) or continuous glass fibres, in Glass Mat Thermoplastic

(GMT) and commingled composites respectively. The PP/glass composites

can mostly be found in automotive applications.

10

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.11

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

H H

C C

H CH3

n

Polypropylene is a semicrystalline thermoplastic, with a glass transition at 0C

and a melting temperature range from 158-168C. PP has a density of 900

kg/m3. It should not be used in air above 110C. Suitable forming

temperatures of PP/glass composites are around 200C.

outstanding thermal stability. This thermoplastic, reinforced with carbon, can

be regarded as the prime high performance thermoplastic composite,

introduced as APC-2 (Aromatic Polymer Composite) by ICI-Fiberite in the

late 1980s.

O

O O C

143C and a melting temperature of 334C. It can be safely used up to 250C.

Suitable forming temperatures are around 400C. The strength and solvent

resistance are excellent. However, the material is notoriously expensive.

materials. Reinforced with glass or carbon it is applied in aerospace

applications such as the Airbus A340 and the new A380 for structural parts.

With a glass transition at 90C and a melt temperature range from 280-288C

PPS is another high temperature thermoplastic. It can be used safely up to

240C. Suitable forming temperatures are around 350C. The material

crystallises rapidly and has an excellent solvent resistance. PPS is more brittle

than PEEK but also significantly cheaper.

other polyimide resins). Due to its low flammability with low smoke emission

it is very suitable for aerospace interior parts, reinforced with glass or carbon.

Being an amorphous material, PEI is more susceptible to solvents than its

semicrystalline counterparts. On the other hand, the fracture toughness is very

good, certainly when compared to PPS.

11

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.12

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

O O

CH3

C C

N N

C O C O C

O CH3 O n

3

PEI has a glass transition temperature of 215C and a density of 1270 kg/m .

Typical forming temperatures are around 350C.

1.3.3. Properties

The table below presents some typical properties of selected matrices.

Properties of thermosets are indications, as they very much can vary according

to their compositions. The properties of the semi-crystalline thermoplastics

will vary with their degree of crystallisation.

Modulus Thermal

Matrix max(MPa) max (%) Expansion

(kg/m3) E (GPa) (10-6/C)

Yield Failure Yield Failure

80 + 1406 1150-1200 3.5 60 - 80 35 60

120 + 160 1200-1300 3.5 80 - 90 45 65

PPS 1370 3.7 80 50

PEI 1270 3 105 85 6 60 55

PES 1370

6

Curing and post curing temperature

12

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.13

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

With any material, the manufacturing process can greatly influence the

performance of the obtained structure. Think for example about molecular

orientation in injection moulded plastic parts. For composites, the fabrication

technique will influence the fibre content, entrapment of air bubbles (voids),

degree of crystallinity for semi-crystalline thermoplastics, positioning of the

reinforcement

Different basic manufacturing techniques will be explained briefly. As pointed out

earlier, the method used depends to a great extent on the type of matrix used, with

in general a large difference in viscosity between the uncured thermosetting resin

and the (highly viscous) thermoplastic melt. The choice of a fabrication technique

will also depend on the shape complexity, size and amount of product to be

fabricated, and the performance required.

The form in which the reinforcement is applied can vary depending on the

performance required. Only the reinforcement forms used for laminates will be

described here. Most reinforcements are based on rovings (bundles) of continuous

fibres. They can be aligned in an unidirectional way, or weaved to form a cloth

which can have different patterns as shown in fig. 1.4. Weaving techniques also

allow 3D types of reinforcement. Continuous fibres can also be randomly aligned

(think of a plate of spaghetti). Chopped fibres are generally used in high-volume

applications.

weave Weave

performed before the composite is put into shape. For unidirectional or woven

13

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.14

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

quality-high fibre content applications. A similar technique is used for chopped

fibre with unsaturated polyester: Sheet Moulding Compound and Bulk Moulding

Compound (SMC and BMC). Preimpregnation reduces manual procedures and

improves for example the fibre distribution. In the case of thermosets, the resin is

partially cured and the prepreg needs to be stored at low temperature and used

fairly rapidly.

This easy and moreover flexible way uses

an open mould, and is used for example for

prototype fabrication or when producing large

components (boat). Both resin and

reinforcement are applied by hand using

simple tools as shown in fig. 1.5. Cold cure

thermosets are mostly used with this

technique. Reinforcement forms are woven,

random and chopped. Stiffening ribs, inserts,

and core material can be fitted during the fig. 1.5: Hand laminating

laminate build-up. An alternative to the hand

lay-up is to spray a mix of resin and chopped fibres. Emission of styrenes or other

toxic gases more and more limit the use of hand lamination in the composite

industry.

Another open mould technique to

produce simple hollow shapes

(tubes, pressure vessels). The

negative mould is rotating while a

fibre guide moves parallel to the

mandrel longitudinal axis. The

fibre is impregnated in

thermosetting resin before it is

guided to the mandrel through a

guide. The product is mostly cured

in an autoclave after winding.

It is worth adding that some

applications are based on

thermoplastic matrix. In this case, a

prepreg tape is used.

14

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.15

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

autoclave

Two closed mould techniques,

which are aimed to improve the

hand lay-up technique. Resin and

reinforcement are still applied by

hand in a simple open mould. Then

vacuum is applied via a rubber bag.

This induces a processing pressure

of 1 bar, whilst reducing air

entrapment. This technique is also

used for bonding sandwich cores to

existing laminates.

The pressure bag technique is

similar, but allows to apply

pressures up to 3,5 bar. This is fig. 1.7: Autoclave technique

favourable for the consolidation

and for fibre content. This puts higher demands on the stiffness of the mould. The

combination of pressure with the possibility of adding temperature to the process

makes it possible to use prepregs. This reduces the hand lay-up step.

The autoclave technique combines earlier named bag techniques. The vaccum bag

assembly is then used within a pressurised vessel (up to 10bar). Both vacuum and

pressure ensure a high quality product.

Rubber press

On basis of prepreg / SMC,

hot pressing uses higher Rubber mould

to 100bar) to increase

productivity and to ensure

good consolidation of the

laminate. Thermoplastic

matrices require a high

Flat laminate Infrared oven Steel mould

temperature to enable forming

or draping (up to 400C,

depending on the matrix fig. 1.8: Rubber pressing

material), thermosets require

acceleration of the curing process. It is clear that different thermoset composition

with a higher viscosity is required. Reorientation of the reinforcement is inherent

to the high pressure. This induces variability in the thermo-mechanical properties

and makes it difficult to obtain good dimensional accuracy. Another consequence

is high tooling cost due to the pressure.

For automotive applications of thermoplastic composites two steel mould halves

are used, as in conventional deep-drawing of metals. The prepreg is preheated

above Tmelt (for example in an infra-red oven), transported to a high-speed press

and rapidly formed. This stamping process dramatically reduces the cycle time (to

15

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.16

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

composite forming, now between a metal mould and a rubber mould. When

compressed, the rubber mould generates a nearly hydrostatic pressure, enabling

good consolidation throughout the laminate, also in products with sharp

curvatures. The lifetime of the rubber mould is of course significantly shorter than

the steel moulds used for stamping.

LCM is a technique where the

reinforcement is pre-formed in the

general shape of the final product.

These pre-forms can be

manufactured by e.g. stitching

layers of fabric, compaction of

fabrics with a binder powder or

braiding around a mandrel. The

preform is then placed in a mould,

where a low-viscosity thermoset

resin is injected under pressure.

fig. 1.9: Resin Transfer Moulding

LCM can be applied in many

forms. A variant is called Resin Transfer Moulding, where a rigid mould (matched

metal tooling) is used (see figure). Other variant is the Resin Infusion in Flexible

Tooling, where the resin is impregnated by vacuum. In this technique, the preform

is layed onto a rigid mould. The matched metal tool found in RTM is replaced by

a formable vacuum bag. Due to the low pressure applied to the preform, the rigid

mould can be made of cheap and easy to form material, like PS foam. A know

application of RIFT are wind mill blades.

1.4.7. Pultrusion

Pultrusion is a continuous process, which makes it possible to produce infinite (in

principle) length of composite having a certain profile. Bundles of fibres

impregnated with a thermosetting matrix are pulled through a (heated) die having

the required profile cross-section form. Layers of random mat or woven cloth can

also be added before the die, as shown in fig. 1.10.

16

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 1.17

R

M

OU

CO

1. Constituents and manufacturing techniques - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

materials

C O C

CH3 methyl ether

(R O R')

R C O R'

CH3 CH2 ethyl ester

O

CH2 CH vinyl CH2 O formaldehyde

phenylene OH phenol

O

epoxide CH CH2 styrene

CH2 CH

O

glycidyl

CH2 CH CH2

O

C

CH

N imide

CH

C

17

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.1

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

2. Micromechanics

2.1. Introduction

In the field of composite material, micromechanics

is about mechanics at the level of the constituents,

i.e. the fibre, the matrix and possibly the interface.

This can be used for example to study the stress-

strain situation around a fibre under different

loading conditions in order to understand the failure

behaviour of composites.

In our case, we will concentrate on fig. 2.1: Fibre-matrix scale

micromechanical models, which are built in order

to find a relation between the properties of the constituents and that of the

composite layer. These models allow the designer to evaluate the combination of

different constituents. Another important aspect is that of understanding

composites.

The first part of this chapter is dedicated to the relation between the constituents

volume and mass fraction. Then, two methodologies giving expressions for in-

plane mechanical properties are discussed. A numerical example is also given.

This chapter shows that although convenient and helpful, these models should be

used in conjunction with some experimental work. Experimental data of the

constituents is necessary as input to the models, but also to validate the model

results.

The relative content of the constituents is mostly quoted as a volume content or

fraction (of fibres for example). Whatever the amount of constituents, we have:

n

Vi

v

i =1

i = 1 with vi =

Vc

(2.1)

Vi is the volume of the ith constituent,

Vc is the total volume of the composite.

The constituents considered are fibres (subscript f), matrix (m) and voids (v).

Although micromechanics equations are mostly based on volume content,

measurement of constituent content is often based on mass fractions. The mass

content of matrix and fibres (weight of voids is neglegted) follow a similar rule as

the volume fractions:

n

M

i =1

mi = 1 with mi = i

Mc

(2.2)

Substituting the product of density and volume for mass for each constituents

gives:

n

c = i vi (2.3)

i =1

The density can also be obtained as a function of mass by substituting the quotient

of mass by density for volume in equation (2.1), this gives:

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.2

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1

c = n

(m

(2.4)

i / i )

i =1

Using the same substitution, it is possible to calculate the void fraction from

equation (2.1):

M f Mm

+

f m

vv = 1 (2.5)

Mc

c

Typical values:

The packing (or distribution) of fibres within a

composite cannot be described. As shown in fig.

2.2, the distribution of fibres is unhomogeneous. In

order to build micromechanical models, simplifying

assumptions are made on the packing of fibres. The

most simple packing is the square packing as shown

in fig. 2.3.

fig. 2.2: Typical composite

cross-section micrograph

vf of such a square packing, the fibre diameter d, and the distance between fibre s:

d

2

vf = (2.6)

4s

An other typical packing is the triangular (fig. 2.4)

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.3

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

d

2

vf = (2.7)

2 3s

Maximum packing is obtained in both packing models for d=s. It gives in the case

of square packing vf-max=0.79, and for the triangular packing 0.91.

In the practice, fibre volume fraction for composite based on unidirectional layers

can be found in the range 0.5 to 0.8.

An other remark concerns the void content. Typical autoclave (pressure+vacuum)

cured composite products have voids content varying from 0.1 to 1%. Pressure

bag (no vacuum) cured composites can have voids content in the order of 5%.

As was quoted in the introduction, this insight into micromechanics for

composites will be restricted to unidirectional layers. The mechanics of material

approach provides the simplest micromechanical equations for the in-plane

elasticity moduli E1, E2 and G12. No attention is paid to the type of fibre packing.

The composite is therefore considered as homogeneous. A block of composite

containing fibre and matrix as show in fig. 2.5 is simplified to block containing

two volumes. These two

volumes are connected

together and represent f

m

the matrix (m) and the

3

fibre (f) with their

2

respective properties

1

and volume fractions.

An elasticity modulus is fig. 2.5: Representative matrix and fibre volume elements

then obtained by

performing a simple experiment, where the two representative volumes are

subjected to an average stress. Poisson effects are neglected.

The longitudinal modulus is obtained by

considering the matrix (m) and fibre (f) 1 f 1

representative element in parallel, and by m

applying an average stress along the 1-axis as

shown in fig. 2.6 (model of Voigt). It can be 3

2

shown from strain energy approach that the strain

induced 1 is uniform for both fibre and matrix 1

volume elements. If the areas of the fibre and fig. 2.6: Basic experiment for the

matrix volumes on which the stress is applied are longitudinal modulus.

(Voigts model)

called Af and Am respectively, the following

relation can be written:

1 A = m Am + f A f (2.8)

Using equation (2.1) applied to areas, we obtain:

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.4

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1 = m vm + f v f (2.9)

It is now assumed that both fibre and matrix behave in a linear elastic way, and

that the matrix is isotropic and is characterised by the elasticity modulus Em. The

fibre is assumed orthotropic (important in case carbon or aramid fibre are

considered) and is therefore characterised by a longitudinal modulus Ef1 and a

transverse modulus Ef2. Poisson strains are neglected. With these assumptions, the

equation (2.9) can further be developed as:

E1 1 = Em m vm + E f 1 f v f (2.10)

It was earlier assumed that the strain was equal for both volumes and therefore:

E1 = Em vm + E f 1v f (2.11)

This relation is called the rule of mixture for the longitudinal modulus. As will be

shown later on basis of comparison with experimental data, the Voigts model

provides, and is generally known as an adequate prediction for predicting the

longitudinal modulus.

The transverse modulus is obtained by 2

considering the matrix (m) and fibre (f)

representative element in series. An average f

stress is applied along the 2-axis as shown in

m

fig. 2.7 (model of Reuss). The total 3 2

deformation is the sum of the deformations 2

occurring on the two volumes. This can be 1

expressed in terms of strain: fig. 2.7: Basic experiment for the

transverse modulus.

(Reuss model)

2 L = m Lm + f L f (2.12)

where Lm and Lf are the lengths of the matrix and fibre volumes. Using equation

(2.1) applied to the volume lengths, we obtain:

2 = m vm + f v f (2.13)

Here again we assume that the 1D Hookes law applies. Equation (2.13) can

further be developed as:

2 m vm f v f

= + (2.14)

E2 Em Ef2

From the series volume arrangement, the stress in matrix, fibre and the composite

are equal, an expression for the transverse elasticity modulus E2 is obtained:

1 v v

= m + f (2.15)

E2 Em E f 2

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.5

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

This relation is often called the inverse rule of mixture. Note again that the fibre

transverse modulus should be used. This equation is generally known as being

inadequate for predicting the transverse modulus. This is due to the fact the

assumption made on the equality of the stress in the matrix and the fibre in the

volume-in-series model is not valid in a real composite. This can be shown on

basis of strain energy approach. A second reason for the inaccuracy of models for

the transverse modulus composite based on orthotropic fibre (carbon and aramid)

is that the fibre transverse modulus is difficult to measure (and has actually never

directly been measured). Quoted values for the transverse modulus of fibres are

actually derived from the comparison between micromechanical model results and

experiments. The same actually applies for the shear modulus of these orthotropic

fibres.

adding a matrix volume in parallel to the series

model as shown in fig. 2.8. This geometrical

model is an approximation of a square fibre m

packing. This result in the following equation: f

3 2

2

1

fig. 2.8: Parallel-series model for

the transverse modulus.

v f Em E f 2

E 2 = (1 v f ) Em + (2.16)

v f E m + v f (1 v f ) E f 2

For the shear modulus, a similar approach similar to that, which led to the

expression of the transverse modulus, is used. Assuming equal shear stresses in

fibres and matrix, the following equations is obtained.

1 v v

= m + f (2.17)

G12 Gm G f 12

For the same reasons as for the transverse modulus, this equation is not very

accurate. The Poissons ratio can be derived form the following rules of mixture:

12 = m vm + f 12 v f (2.18)

This equation is known to be accurate enough for design purposes.

These models are mostly based on a mechanics of material solution. However

they contain curve-fitting parameters in order to match experimental results or

elasticity solution. The Tsai-Hahn equations for example, use similar assumptions

as the inverse rule of mixture (2.15), but adds a "stress partitioning factor " in

order to take into account the mismatch in stress in the fibre and in the matrix. The

Tsai-Hahn equation for the transverse modulus is:

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.6

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1 1 v f vm

= + (2.19)

E2 v f + v m E f 2 Em

This equation gives for =1 the inverse rule of mixture. A similar equation can be

written for the shear modulus.

Other widely used equations are the Halpin-Tsai equations, which for the

transverse modulus are:

E 2 1 + v f E f 2 Em

= , with = (2.20)

Em 1 v f E f 2 + Em

where is the curve fitting parameter. It is worth noting that giving the value 0

gives the inverse rule of mixture (2.15). It was shown that using =2 gives similar

results as more complex elasticity solutions. Replacing Ei by Gi in equation (2.20)

gives the Halpin-Tsai equation for the shear modulus. A curve fitting parameter

of 1 is mostly used.

calculating the thermo-mechanical propoerties of unidirectional layers, as well as

different fabric types. Not only mechanics of materials based procedures are

included, but also elasticity solutions (CCA). A document reviewing this method

is also available on this site.

http://www.pt.ctw.utwente.nl/organisation/tools/

2.5 Example:

An example is given, based on a unidirectional carbon reinforced Polyetherimide

(PEI) having the following constituents properties:

Carbon

(GPa) (GPa) (GPa) (GPa) (m/m.oC) (m/m.oC) (kg/m3)

PEI Em Gm

m m m

(GPa) (GPa) (/oC) (kg/m3)

table 2.1: Linear elastic thermomechanical properties of the Torayca T300 carbon fibre, and the

Ultem1000 PEI.

The matrix mass content of the composite obtained with these components is

mm = 41.4%, as measured by desolving the matrix of several composite specimen

in Choroform. The corresponding fibre volume fraction is vf = 51%, The measured

in-plane properties can be found in table 2.2. The measured longitudinal modulus

compares very well with the rule of mixture (119 GPa). The transverse modulus

according to different models are given in table 2.3.

1

Comes as well as Gf12 from R.F. Gibson's Principle of Composite Material Mechanics book.

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.7

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

Experimentally: 120 7.8 3.5

table 2.2: Measured unidirectional Carbon-PEI in-plane properties

carbon-PEI E2 (GPa)

Experimentally: 7.8

Inverse rule of mixture (2.15) 5.0

Parallel-series model (2.16) 7.7

Halpin-Tsai =2 6.5

Halpin-Tsai =10 7.85

table 2.3: Different models for the transverse modulus

The results obtained from the different models are set in graph form as a function

of the fibre volume fraction. The experimental value is also marked as a reference.

1.6E+10

1.4E+10

1.2E+10

experimental value

1.0E+10

E 2 (Pa)

8.0E+09

6.0E+09

4.0E+09

2.0E+09

0.0E+00

0 0.2 0.4 0.6 0.8 1

vf

Inverse rule of mixture Parallel-series model

Halpin-Tsai x=2 Halpin-Tsai, x=10

fig. 2.9: Composite transverse modulus as a function of the fibre volume fraction.

It is clear from fig. 2.9 that the inverse rule of mixture does not apply to the

carbon-PEI considered. In a less extent, a similar conclusion can be drawn for the

Halpin-Tsai having =2. The parallel-series model and Halpin-Tsai with =10

both give a good approximation of the composite's transverse modulus. As

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.8

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

discussed earlier, these models are dependent on the value chosen for the fibre's

transverse modulus. One should therefore be careful when drawing conclusion.

An other remark is that the parallel-series model and Halpin-Tsai with =10 do

not follow the same trend as a function of the fibre volume fraction.

SI T ES G

O

Composites Course 2008 University of Twente, Eng. & Tech. 2.9

R

M

OU

CO

2. Micromechanics - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

2.6 Problems:

2.1 Derive equation (2.16)

2.2 Find in the literature expressions for the longitudinal and transverse

coefficients of thermal expansion 1 and 2 as a function of de components

properties.

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.1

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

3.1 Introduction

In the first chapter, it was shown that

the properties of the basic

constituents of a composite material

fibre and matrix- are extremely

different. This implies that a

composite made of these two

constituents is heterogeneous

(properties vary from point to point)

when considered at a fibre matrix

scale as shown in fig. 3.1. When

coming to a composite layer, it can

be assumed that the anisotropic fig. 3.1: Fibre-matrix scale, ply scale.

properties are independent of the

position from which they are looked at. It makes it possible to assume a composite

as a plate having averaged anisotropic material properties. The stress-strain

relations of a composite layer in general will be developed next.

The stresses at a point in a material 3

can be described by 9 components ij

as shown in fig. 3.2, where i,j = 1,2,3. 33

According to this notation, ij are 32

normal stresses when i=j, shear 31 23

stresses when ij (also noted ij). It 22 2

13 12 21

can be shown that only 6 of these

stress components are independent, 11

i.e. ij = ji. Similarly, there are 6 1

strain components ij to describe the

deformation at this point. The fig. 3.2: State of stress in a material

principle is similar as for the stress

components, although the engineering shear components ij are twice the shear

deformation tensor components.

Tensor

11 22 33 23=23 31=31 12=12 11 22 33 23=223 31=231 12=212

notation

Contracted

1 2 3 4 5 6 1 2 3 4 5 6

Notation

Assuming small linear elastic deformations, the strain components are related to

the deformations by the following relations:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.2

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

u1 u u

1 = ; 2 = 2 ; 3 = 3 ;

x1 x2 x3

(3.1)

u u u u u u

4 = 2 + 3 ; 5 = 3 + 1 ; 6 = 1 + 2 ;

x3 x2 x1 x3 x2 x1

where u1, u2 and u3 are the displacement along the 1, 2, 3 axis respectively.

The relation between stress and strain components describes the behaviour of the

material when loading is applied. We will consider linear elastic materials, and in

this case, each strain component i is a linear function of the stress components

j (j = 1,,6):

6

i = Sij j (3.2)

j =1

{ } = [S ] { } (3.3)

where [S] is the compliance matrix. There are 36 (6 x 6) compliance constants.

The inverse of the compliance matrix is called the stiffness matrix ([C]=[S]-1).

The stress-strain relationship just described is valid for any anisotropic elastic

material. It will be shown that the amount of compliance (or stiffness) constants

can be reduced.

Firstly, it can be shown that the compliance (and stiffness) matrix is symmetric.

The demonstration of this symmetry is based on the fact that the energy necessary

to induce deformation to the elastic (and hence reversible) material is independent

from the way the energy is produced. This leads to 15 equations of the form

i j

= (3.4)

j i

This means that the compliance matrix is symmetric (and therefore the stiffness

matrix), i.e.:

Sij = S ji ; Cij = C ji (3.5)

This leaves 21 of the 36 compliance constants to describe an anisotropic material.

3

Further simplifications can be made

by considering geometrical 2

symmetries in the composite layer

considered. Take as an example a

layer of unidirectional material as 1

shown in fig. 3.3. Let us define a

Cartesian coordinate system where fig. 3.3: A unidirectional layer and its material

the first axis (1) coincides with the coordinate system

direction along the fibres. The two

other directions are then perpendicular to the fibres (2 & 3), with the third axis

perpendicular to the layer. This coordinate system is called the material coordinate

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.3

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

system. It can be seen that there are 3 planes of material property symmetry (i.e.,

the planes defined by the coordinate system axis 1&2, 2&3, 1&3. A material

agreeing to these symmetry assumptions is called orthotropic. It can be shown

that this leaves a compliance matrix with 9 independent components.

1 S11 S12 S13 0 0 0 1

S 22 S 23 0 0 0 2

2

3 S 33 0 0 0 3

= (3.6)

4 Sym S 44 0 0 4

5 S 55 0 5

6 S 66 6

As an example, apply a stress 1 along the first axis (all other stress

components are zero). This means that the relation for the in-plane shear

reduces to

6 = S 61 1 (3.7)

It can be seen from fig. 3.3 that a stress applied along the fibres cannot

generate any in-plane shear deformation. Therefore the compliance matrix

component S61=0. Such simple experiments can be applied for the other zero-

components.

It is important to note that the concept of orthotropy is dependent on the

coordinate system chosen. For example, a normal stress applied in an arbitrary

direction will induce shear strains. This will be shown later on when rotating the

compliance (or stiffness) matrix.

A particular case of orthotropy is transverse isotropy. This assumption is taken

for unidirectional layers, where the properties are assumed isotropic in the 2-3

plane, i.e. perpendicular to the fibres. The compliance matrix contains now 5

independent constants and can be further reduced to the following set of relations:

1 S11 S12 S12 0 0 0 1

S 22 S 23 0 0 0 2

2

3 S 22 0 0 0 3

= (3.8)

4 Sym 2( S 22 S 23 ) 0 0 4

5 S 66 0 5

6 S 66 6

following set of equations:

1 S11 S12 S12 0 0 0 1

S11 S12 0 0 0

2 2

3 S11 0 0 0 3

= (3.9)

4 Sym 2( S11 S12 ) 0 0 4

5 2( S11 S12 ) 0 5

6 2( S11 S12 ) 6

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.4

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

stiffness) matrix is also due to the relation between the engineering constants E

and G:

E = 2 G (1 + ) (3.10)

The relation between the compliance coefficients Si and the engineering constants

will be developed in the next alinea.

It is convenient to express the compliance (or stiffness) matrix components into

engineering constants such as the Youngs modulus E. These engineering

constants are defined by considering simple experiments. A stress situation is

defined where only one stress component is different from zero, i.e.:

k 0; i = 0 for i k ; k = 1,...,6 (3.11)

possible to define a modulus as:

Rk = k / kk (3.12)

where the superscript k in the strain component kk reminds that it finds its origin

in a k -stress.

For k=1,2,3: Rk is the modulus of elasticity along the k-axis and will be

given the notation Ek.

For k=4,5,6: Rk is the shear modulus relating respectively to the 2-3, 3-1 and

1-2 planes and along the k-axis and will be given the notation

Gk.

If we now consider the ik k -strain component, a deformation ratio is defined as:

kl = lk / kk (3.13)

For k and l = 1,2,3, kl are the Poisson's ratios with a negative sign (kl). In the

case of an orthotropic layer, all other deformation ratios are zero, which is not true

for an anisotropic material in general. In this case, it is worth mentioning the

shear-extension coupling coefficients kl (Lekhnitskii) and the shear-shear

coupling coefficients kl (Chentsov). The earlier applies for k = 1,2,3 and l = 4,5,6

or k = 4,5,6 and l = 1,2,3. The later one concerns k and l = 4,5,6.

These definitions make it possible to fill the compliance matrix. For an orthotropic

layer, this gives:

Skk = 1 / Rk for k = 1,..,6;

lk lk kk kl (3.14)

Skl = = = for k = 1,..,3, l k

k kk k Rk

The equations are:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.5

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1 21 31

E 0 0 0

E2 E3

1

12 1 32

1 0 0 0 1

E1 E2 E3

2 13 23 2

0

1

0 0

3 E1 E2 E3 3

= (3.15)

4 0

0 4

1

0 0 0

5

5

G23

1

6 0 0 0 0 0 6

G13

1

0 0 0 0 0

G12

kl lk

= for k , l = 1,..,3; l k (3.16)

E k El

A simplification of the stress-strain relations for an orthotropic layer can be made

in case of plane state of stress, i.e. 3=4=5=0. If only the in-plane deformations

are considered, equation (3.15) reduces to:

1 21

0

1 E1 E2 1

21

2 =

1

0 2 or

{ } = [S ] { } (3.17)

E 2 E2

6 1 6

0 0

G12

The other strain components can be non-zero, but we will focus on the behaviour

of the in-plane strains. Inverting the compliance matrix gives:

E1 21 E1

0

1 1 12 21 1 12 21

1

21 E1

2 =

E2

0 2 or { } = [C ] { } (3.18)

1 12 21 1 12 21

6 0 0 G12 6

Up to now, the stress-strain relationships x2

x2

for a composite layer with the material x1

coordinate corresponding with the plate

coordinate system were considered. The

expressions for a rotated coordinate

x1

system will be derived here for a rotation x3= x3

in the 1-2 plane (axis 3 is fixed).

fig. 3.4: Rotation from (x1, x2, x3) to (x1, x2, x3)

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.6

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

In general, the transformation of the coordinates of a point from the (x1, x2, x3)

system to the (x1, x2, x3) system (fig. 3.4) can be written as:

3

xi' = ij x j with ij = cos( xi' , x j )

j =1

(3.19)

i.e. 11 = 22 = cos ; 12 = 21 = sin

13 = 31 = 23 = 32 = 0; 33 = 1

The transformation of the Cartesian components of any tensor tij is then given by:

tij' = ik jl tkl (3.20)

In matrix form, the stress transformation gives:

1' m 2 n2 0 0 0 2mn 1

'

2 n

2

m2 0 0 0 2mn 2

3 0 0 3 m = cos

'

0 1 0 0

'= with

4 0 0 0 m n 0 4 n = sin

5' 0 0 0 n m 0 5

' 2

6 mn mn 0 0 0

m n 6

2

(3.21)

This transformation matrix will be called 3

2*

[T] in the following. 2

1

1*,2*,3: layer CS

1,2,3: material CS (at )

fig. 3.5 shows the chosen notations for the coordinate system: (1, 2, 3) for the

material CS as in fig. 3.3, (1*, 2*, 3) for the CS of the layer. The properties in the

(1, 2, 3) coordinate system are known, and need to be transformed into the layer

CS. Compared to the situation in fig. 3.4, it therefore implies to rotate with an

angle -, which means using [T]-1. In case of plane stress, the stress transformation

can be written:

1* cos2 sin 2 2 cos sin 1

2 cos sin 2 or { * }= [T ] { }

* 1

2 = sin cos2

2

(3.22)

* cos sin cos sin cos2 sin 2 6

6

The transformation of the strain components is performed similarly, but the

transformation described above based on tensors is valid for 12, and therefore 6/2

(see table 3.1), i.e:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.7

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1* 1

* 1

2 = [T ] 2 (3.23)

1 * 1

2 6 2 6

This is usually corrected by introducing the Reuter matrix [R]. In plane stress

form, it can be written as:

1 0 0

[ ] 0 1 0

R = (3.24)

0 0 2

becomes:

1* 1* 1 1

* * 1 1

2 = [R ] 2 = [R ] [T ] 2 = [R ] [T ] [R ] 2

1

(3.25)

* 1 *' 1

6 2 6 2 6 6

i.e.:

1* cos2 sin 2 cos sin 1

*

2 = sin cos2 cos sin 2

2

(3.26)

* 2 cos sin 2 cos sin cos2 sin 2 6

6

It is now possible to formulate a transformed stiffness matrix [C*]:

1* 1 1 1*

* 1 1 *

2 = [T ] 2 = [T ] [C ] 2 = [T ] [C ] [R] [T ] [R ] 2

1 1

* * (3.27)

6 6 6 6

=[C*]

This gives for an orthotropic material with its stiffness matrix written in its

material coordinate system (i.e. C16=C26=0):

C11* = C11 cos 4 + 2(C12 + 2C66 ) sin 2 cos 2 + C22 sin 4

*

C 22 = C11 sin 4 + 2(C12 + 2C66 ) sin 2 cos 2 + C22 cos4

C12* = (C11 + C22 4C66 ) sin 2 cos 2 + C12 (sin 4 + cos 4 )

(3.28)

C16* = (C11 C12 2C66 ) sin cos3 + (C12 C 22 + 2C66 ) sin 3 cos

*

C 26 = (C11 C12 2C66 ) sin 3 cos + (C12 C22 + 2C66 ) sin cos3

*

C66 = (C11 + C22 2C12 2C66 ) sin 2 cos 2 + C66 (sin 4 + cos 4 )

matrix [S*]:

1* 1 1 1*

* *

2 = [R ] [T ] [R ] 2 = [R ] [T ] [R] [S ] 2 = [R ] [T ] [R ] [S ] [T ] 2

1 1 1 1 1 1

* *

6 6 6 6

=[S*]

(3.29)

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.8

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

any coordinate system can be written. For example, apply a stress 1* along the axis

1*. The strain along the same axis is then:

1* = S11* 1* (3.30)

Which means that E1* = 1 / S11* . Expanding (3.29) gives:

1 1 1 2 12 1

= cos 4 + ( ) sin 2 cos 2 + sin 4 (3.31)

E1* E1 G12 E1 E2

effects

The anisotropic behaviour of the deformation of a layer is not only evident when a

mechanical stress is applied, but is also subject to changes in environmental

conditions. The most important are change in temperature and moisture content.

This can affect the composite both during fabrication as during its end-use. It is

therefore important to include them in the analysis.

The strain induced by a change in temperature iT or in moisture content iH can be

expressed as:

T i = 1, 2, 3 c i = 1, 2, 3

iT = i for ; iH = i for ; (3.32)

0 i = 4, 5, 6 0 i = 4, 5, 6

where: i is the coefficient of thermal expansion (CTE)

T is the change in temperature (T-T0), T0 the temperature for which

iT =0.

i is the coefficient of hygroscopic expansion (CHE)

c is the moisture concentration, with a moisture free-state as a

reference.

These hygrothermal strain terms are valid for temperature and moisture content

ranges where the coefficient of thermal expansion and the coefficient of

hygroscopic expansion can reasonably be assumed constant. For an orthotropic

material, the stress-strain relationship in the material coordinate system can be

written as:

1 S11 S12 0 1 1 1

2 = S 21 S 22 0 2 + 2 T + 2 c

(3.33)

0 S 66 6 6

6 0 6

or:

{ } = [C ] ({ } { } T { } c ) (3.34)

The transformation of the hygrothermal strain from the material coordinate system

to an arbitrary coordinate system is performed as for the strain components, i.e.

for the thermal expansion components:

1* 1* 1 1

* * 1 1

2 = [R ] 2 = [R ] [T ] 2 = [R ] [T ] [R ] 2

1

(3.35)

* 1 *' 0 0

6 2 6

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 3.9

R

M

OU

CO

3. Stress-strain relationships of a composite layer - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

3.7 Problems

3.1 Write expressions for E2* , G12* , 12* , 16* and 26

*

.

*

as a function of for a

Carbon-PEI having orthotropic properties in its material coordinate system

with:

E1 = 120GPa, E2 = 8GPa, G12 = 3.5GPa, 12 = 0.3.

3.3 A filament wound cylindrical pressure vessel of mean diameter d=1m and

wall thickness t=0.02m is subjected to an internal pressure p. The vessel is

wound at an angle of 54o from its longitudinal axis and the glass-epoxy

material used has the following mechanical properties:

E1 = 40GPa, E2 = 10GPa, G12 = 3.5GPa, 12 = 0.25.

By the use of a strain gauge, the strain along the fibre direction is measured

and has at the pressure a value of 1=0.001m/m. Determine the internal

pressure in the vessel.

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.1

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

4.1. Introduction

Chapter 2 and 3 focused on the behaviour of a single layer. The thermo-

mechanical properties of such a layer can be predicted, and the stress-strain

relationship is known. This will now be extended to the more general case of a

laminated plate. The most common type of analysis will be developed next, and is

known as the classical lamination theory. This plane stress theory makes it

possible to relate external loads (in-plane forces and moments) to the composite

plate deformations.

The analysis of laminates will first be introduced by considering the behaviour of

simple laminated beam under pure bending.

Considering laminated beams under pure flexural loading gives an introduction to

the more general laminated plate theory. Elementary principles of mechanics of

material applied for beam calculations will be used here. We will focus on the

main differences occurring with isotropic beams.

A rectangular laminated beam of thickness h and width b is subjected to a bending

moment M, which is constant over the whole length of the beam (pure bending).

The beam is made of N layers. As

shown in fig. 4.1, the position of a

layer k in the beam is defined by

the distances zk-1 and zk. It is worth

emphasising that these distances

are taken from the middle axis of

the beam and not from the neutral

d

axis.

The following assumptions are

taken:

z0 1

a. The plies are perfectly bonded z

together. M 1 2 h/2 -p l a

ne

m id

b. Plane cross-sections which are zk zk-1 M

initially perpendicular to the h

x

longitudinal axis of the beam k

remain plane and normal n

during flexure. z

c. Each ply behaves linear

elastically with no in-plane

dx

shear coupling ([C16]=0)

We will add for simplicity that the

beam has geometrical and property fig. 4.1: Composite beam bending with layer

numbering system

symmetry about the middle axis.

therefore having a radius of curvature and an angle between the normals to the

beam axis d. From assumption (b), an expression for the longitudinal strain at a

distance z from the middle axis is:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.2

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

x =

( + z ) d d

z

= (4.1)

d

According to assumption (c), the longitudinal stress at a distance z from the

middle axis becomes:

z

x = Ex (4.2)

Static equilibrium gives an expression for the bending moment:

h

2

M= x b z dz (4.3)

h

2

By taking into account the stress xk and elasticity modulus E xk in each layer k,

we obtain:

b z2

E x (zk 1 z k3 )

N zk

b N k 3

M = E xk dz = (4.4)

k =1 zk 1 3 k =1

The bending moment can also be expressed as a function of the elasticity modulus

of the laminated beam E xl :

E xl I y

M= with I y = b z 2 dz (4.5)

With Iy the moment of inertia with respect to the y-axis. From (4.4) and (4.5), an

expression for the elasticity modulus of the beam can be obtained:

E xl =

b N k 3

3I y k =1

E x z k 1 z k3 ( ) (4.6)

laminated beams from elementary mechanics of material.

Expression for the stress in the kth layer xk can be written by eliminating in

relations (4.2) and (4.5):

M z E xk

x =k

(4.7)

I y E xl

This relation for the stress is similar to the one used for isotropic 1

beam, corrected by the dimensionless term in bracket. The stress 2

is therefore a discontinuous function of the beam depth, in 3

contrast to the stress in an isotropic beam. An example of a stress x

profile of a symmetric 6 layers beam subjected to a bending

moment, where E x2 > E 1x = E x3 is given in fig. 4.2.

fig. 4.2: Example of a

stress profile of a beam

beam subjected to a

bending moment.

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.3

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

The more general case of a z0=-1/2 h 1

laminated plate under plane

stress condition will now be zk-1 1*

analysed. In-plane loading (axial

and shear) will be considered as

loading, as well as moments 2*

k x

0

(bending and torsion). As for the 2

laminated beam, the layers are y

assumed bonded together. No

restriction is set on the lay-up zn=1/2 h

used, which means that various

coupling effects will be present. n

in-plane stress applied to a 3* z

laminated plate may result in a

complex combination of

extensional, flexural and x,y,z: Laminate CS

torsional deformations. The (at mid-plane)

1*,2*,3: Layer CS

different notations are defined in

1,2,3: Material CS (at )

fig. 4.3. This figure also defines

the different coordinate systems.

fig. 4.3: The different coordinate systems used in the

The position of the layers in the laminated plate theory

normal direction is defined with

the mid-plane as a reference (and

not the neutral plane).

The assumptions relevant for the analysis are similar to the one used for the beam

analysis and can be translated to:

a. The displacements u, v, w corresponding to the directions x, y and z are small

compared to the plate thickness h. The displacement u and v are a linear

function of the depth z (plane cross-sections which are initially perpendicular

to the longitudinal axis of the plate remain plane and normal during

deformation)

b. Transverse shear strains and normal strain are negligible.

According to assumption (a), in-plane displacements are a linear function of the

depth z. Both can be expressed with the displacement in the middle surface (u0

and v0) as reference:

u = u 0 ( x, y ) + z f u ( x, y )

v = v 0 ( x , y ) + z f v ( x, y ) (4.8)

w = w 0 ( x, y ) = w( x, y )

Normal strain is negligible (assumption (b)) and the normal (or transverse)

displacement w is constant for any coordinate (x,y). According to the same

assumption, transverse shear strains are negligible and can be used in the strain-

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.4

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

displacement relations in order to find an expression for fu and fv. The strain in a

layer and in its coordinate system (1*, 2*, 3) can be written as:

v w w w

4* = + = f v ( x, y ) + = 0 f v ( x, y ) =

z y y y

u w w w (4.9)

5* = + = f u ( x, y ) + = 0 f u ( x, y ) =

z x x x

Using the expressions for the in-plane displacements (4.8), the in-plane strains

are:

u u 0 2w 2w

1* = = z 2 = x0 z 2

x x x x

v v 0

w

2

2w

2* = = z 2 = y0 z 2 (4.10)

y y y y

u v u 0 v 0 2w 2w

6* = + = + 2z = xy0 2 z

y x y x xy xy

With x0 , 0y , xy0 the in-plane strains in the mid-plane. It can be shown from

geometrical considerations that:

2w 2w 2w

x = ; = ; = 2 (4.11)

x 2 y 2 xy

y xy

with the plate curvature (1/). The equation (4.10) relates the strain in a layer

with quantities related to the laminate: the mid-plane strain and the curvature.

An expression for the stress in the kth layer as a function of the mid-plane strain

and the plate curvature is obtained by combining (4.10) with the relation (3.27)

obtained in chapter 3. In matrix form, this can be written as:

{ *}k = [C*]k { 0 } + z[C*]k { } (4.12)

External forces and moments acting on a laminated plate can be related to the

stress in the layer, and then to the laminate deformation. For example, the axial

forces Nx per unit width can be obtained by summing the axial stresses x acting

on each layer:

N zk

Nx = ( 1* ) k dz

z

(4.13)

k =1

k 1

where (1*)k is the stress in the kth layer in the 1* direction (layer coordinate

system). A similar expression can be written for the normal force in the

y-direction as well as for the in-plane shear force Nxy. Substituting (4.12) in the

force resultants gives in matrix form:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.5

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

N zk

{N } = ([ C *] k { 0 } + z [C *] k { }) dz

z

k =1 k 1

N zk zk

{N } = [C *] k { 0 } dz + [C *] k { } zdz

(4.14)

k =1 z k 1 z k 1

N 1 N

{ N } = [C *] k ( z k z k 1 ) { 0 } + [C *] k ( z k2 z k21 ) { }

k =1 2 k =1

This is mostly rewritten in the following way:

Nx A11 A12 A16 x0 B11 B12 B16 x

Ny = A22 A26 y0 + B 22 B 26 y (4.15)

N sym A66 xy0 sym B66 xy

xy

The A-matrix is also called the laminate extensional stiffness matrix, is symmetric

and its components are defined as:

N

Aij =

k =1

( C ij *) k ( z k z k 1 ) (4.16)

The B-matrix is called the laminate coupling stiffness matrix, is symmetric and its

components are defined as:

1 N

Bij =

2 k =1

( C ij *) k ( z k2 z k2 1 ) (4.17)

A similar development can be performed for the moment resultants. This gives as

end result the following relations:

Mx B11 B12 B16 x0 D11 D12 D16 x

My = B 22 B 26 y0 + D 22 D 26 y (4.18)

M sym B66 xy0 sym D 66 xy

xy

The B-matrix is also present here. The D-matrix is called the laminate bending

stiffness matrix, is symmetric and its components are defined as:

1 N

Dij = (Cij *) k ( zk3 zk3 1 )

3 k =1

(4.19)

The relations (4.15) and (4.18) are often written in partitioned form:

N A B 0

=

D

(4.20)

M B

Although the three components of the 'ABD' matrix have similar appellations

(stiffness), they have distinct units. As loading is mostly expressed "per unit

width" (force resultants N in N/m and moments M in N), the A-components have

for unit N/m, the B-components in N and the D-components in Nm.

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.6

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

Inversion of relation (4.20) is often necessary, as the loading is generally known,

and not the other way around. This can be done by simply inverting the [ABD]

matrix as a whole. However, performing the inversion by inverting the sub

matrices A, B and D can be interesting for particular laminate lay-up. The result is

then of the form:

0 a b N

= b T

d M

(4.21)

For example, it can be shown that the extensional compliance matrix [a] can be

written:

[ a ] = [ A] 1 + [ A]1 [ B ] ([ D ] [ B ] [ A] 1 [ B ]) [ B ] [ A] 1

1

(4.22)

This relation reduces for to [A]-1 when the components of the [B]-matrix are zero

(laminate is symmetric). Components of the [b]- and [d]-matrix can be evaluated

as follow:

[ b ] = [ A] 1 [ B ] ([ D ] [ B ] [ A] 1 [ B ])

1

(4.23)

[ d ] = ([ D ] [ B ] [ A] [ B ])

1 1

Similarly to the [A]-matrix, the inverted bending stiffness [d] components can be

calculated by directly inverting the [D]-matrix only if the components of the [B]-

matrix are zero. It is worth adding that the coupling compliance matrix [b] is not

symmetric. The relation between the curvature vector {} and the force vector

{N} is defined by the transposed of the [b]-matrix.

4.4. Problems

4.1 Derive for the [ABD]-matrix for 2 laminates [-45/45]s (symmetric) and

[-45/45/-45/45] (antisymmetric) based on 1mm thick layers of Carbon-PEI

having the following layer properties:

E1=120GPa, E2=8GPa, G12=3,5GPa, 12=0.3

In which [ABD] terms do these two lay-up differ, and what does this mean?

4.2 Give on the next figure the physical signification of the following stiffness-

matrix terms:

A16 and A26; B11, B22 and B12, B16 and B26, B66, D16 and D26

SI T ES G

O

Composites Course 2007 University of Twente, Eng. & Tech. 4.7

R

M

OU

CO

4. Classical lamination theory - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

0

y

a a 22 a 26 b21 b22 b26 N y

21

xy0 a 61 a 62 a 66 b61 b62 b66 N xy

=

x b11 b21 b61 d 11 d 12 d 16 M x

y b12 b22 b62 d 21 d 22 d 26 M y

xy b16 b26 b66 d 61 d 62 d 66 M xy

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 5.1

R

M

OU

CO

5. Classical lamination theory-2 - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

5.1. Introduction

The classical lamination theory as summarised in chapter 4 allows us to find a

relation between the laminated plate deformation on one side, and the loading on

the plate on the other side. The deformations considered were the mid-plane in-

plane strains ( x0 , y0 , xy0 ) and the plate curvatures ( x , y , xy ). The loading

studied were in-plane force (Nx, Ny, Nxy) and moments (Mx, My, Mxy).

From this relation, in-plane stresses and strains can be deduced in any layer of the

laminate.

From this displacement-loading relation, engineering constants of the laminate can

also be derived. As for the single layers in chapter 3, this will be done by

performing simple experiments.

As shown in chapter 3, anisotropy of composites is also present in hygrothermal

properties. It means stresses and strains are sensitive to changes in temperature as

well as in moisture contents. The influence of temperature is already noticeable

during fabrication, as most composites are consolidated above room temperature.

Taking hygrothermal loading into account in the deformation-loading relation

will be developed in the second part of this chapter. An example on the

calculation of thermal residual stresses in a laminated cross-ply beam ([0/90]s)

will also be given.

Laminate engineering constants can be obtained by performing simple

experiments on the laminate. For example, the longitudinal elasticity modulus of

the laminate Ex can be obtained by applying a force per unit width Nx. From the

laminate force-deformation relation:

x0 a11 a12 a16 N x

0

y = a 21 a22

a26 N y or

{ }= [a ]{N }

0

(5.1)

0 a a61 N xy

xy 61 a61

x Nx /h 1

Ex = = = (5.2)

x a11 N x h a11

0

Using the same loading case, an expression for the in-plane Poisson's ratio can be

written:

0y a N a

xy = 0 = 21 x = 21 (5.3)

x a11 N x a11

Similarly, the transverse elasticity modulus of the laminate Ey is obtained by

applying a force Ny and is then:

1

Ey = (5.4)

h a 22

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 5.2

R

M

OU

CO

5. Classical lamination theory-2 - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1

Gxy = (5.5)

h a66

The stress-strain relationship of an orthotropic layer with hygrothermal effects

was developed in 3.6. It was based on the assumption that strain having

mechanical, thermal or moisture origins can be treated separately and then added

using superposition. The strain in a layer k subjected to a mechanical stress, a

temperature change T and a moisture concentration c is therefore written as

(3.33):

{ }k = [S ]k { }k + { }k T + { }k c (5.6)

With the thermal expansion vector {} and the moisture swelling vector {} as

defined in (3.32). The resulting stresses are therefore:

{ }k = [C ]k ({ }k { }k T { }k c ) (5.7)

In the global laminate coordinate system, relation (5.7) becomes:

{ *}k = [C *]k ({ *}k { *}k T { *}k c ) (5.8)

According to (4.12), this expression can be written as a function of the mid-plane

strain 0 and the plate curvature :

{ *}k = [C *]k ({ 0 }+ z { } { *}k T { *}k c ) (5.9)

Following a similar procedure as in 4.3.3, external forces and moments acting on a

laminated plate can be related to the stress in the layer, and then to the laminate

deformation. For example, the axial forces Nx per unit width can be obtained by

summing the axial stresses x acting on each layer:

N zk

Nx = ( 1* ) k dz

z

(5.10)

k =1

k 1

where (1*)k is the stress in the kth layer in the 1* direction (global coordinate

system). A similar expression can be written for the normal force in the y

direction as well as for the in-plane shear force Nxy. Substituting (5.9) in the force

resultants gives in matrix form:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 5.3

R

M

OU

CO

5. Classical lamination theory-2 - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

N zk

{N } = ([ C *] { 0 } + z[C *] { } T [C *] { *} c.[ C *] { *} ) dz

z k k k k k k

k =1 k 1

N 1 N

{ N } = [C *] k ( z k z k 1 ) { 0 } + [C *] k ( z k2 z k21 ) { }

k =1 2 k =1

N N

T [C *] k ( z k z k 1 ) { *} k c [C *] k ( z k z k 1 ) { *} k

k =1 k =1

(5.11)

This is mostly rewritten in the following way:

{N } = [A] { 0 }+ [B] { } {N Th } {N H } (5.12)

Where the [A]- and [B]-matrix are defined in (4.16) and (4.17). The components

of vectors {NTh} and {NH} are fictive thermal and hygroscopic forces respectively

defined as:

N

N iTh = T

k =1

( C ij *) k ( j *) k ( z k z k 1 ) (5.13)

And :

N

N iH = c ( C ij *) k ( j *) k ( z k z k 1 ) (5.14)

k =1

The term fictive emphasises that the components in {NTh} and {NH} are not

actual forces. It only means that for the deformations to be zero, {0}={0} and

{}={0}, a force vector {N} equal and opposite to {NTh} must be applied to the

system.

A similar development can be performed for the moment resultants. This gives as

end result the following relations:

{M } = [B ] { 0 } + [D ] { } {M Th } {M H } (5.15)

Where the D-matrix is defined in (4.19). The components of vectors {MT} and

{MH} are fictive thermal and hygroscopic moments respectively defined as:

N

1

M iTh =

2

T

k =1

( C ij *) k ( j *) k ( z k2 z k21 ) (5.16)

And:

N

1

M iH =

2

c

k =1

( C ij *) k ( j *) k ( z k2 z k2 1 ) (5.17)

Now the laminate stiffness matrix and the fictive force and moment vector are

known ([A], [B], [D], {NTh} ,{NH} ,{MTh} & {MH}), it is possible to express the

laminate deformations:

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 5.4

R

M

OU

CO

5. Classical lamination theory-2 - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

0 a b N N Th N H

= T + +

d M M Th M H

(5.18)

b

5.4. Example:

Thermal residual stress in a cross-ply laminate:

1D solution.

Roughly said, residual stresses are stresses, which are present in a material even

before it is loaded. These stresses will build up during fabrication, when the

matrix is consolidated to form the composite. The following example is meant to

illustrate the formation of thermal residual stresses in a simple laminate on a

macroscopic scale. The matrix used consolidates at a temperature higher than the

temperature at use. A cross-ply [90/0/90] beam is shown in fig. 5.1, where three

situations are depicted:

The situation at a temperature where the matrix starts to consolidate. In

the case of an amorphous thermoplastic material at the glass transition

temperature Tg. It is assumed that the layers have the same initial length.

The situation at a lower temperature, in the case the layers can freely

contract, i.e. with a strain i.T.

The situation at room temperature, in the most common case that the

layers were not able to freely contract due to the temperature change.

u90T

u0T

90o

: at Tg

0o

: i T 90o

u90m u0m

90o

: at RT

0o t0

90o t90/2

The residual stress profile on the macroscopic scale can be obtained from a simple

strain compatibility equation:

T

u90 m

u90 u0T u0m

+ = +

L L L L

T

where ui (with i being 0 or 90 stands for the ply orientation) are the thermal

displacements due to the ply free expansion only, and uim are the mechanical

displacements necessary to compensate the free expansion of the different plies.

This equation can then be rewritten as:

90

r

r

90 T + = 0 T + 0

E90 E0

SI T ES G

O

Composites Course 2007 University of Twente, Eng. &Tech. 5.5

R

M

OU

CO

5. Classical lamination theory-2 - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

where i and Ei are respectively the coefficients of thermal expansion and the

Youngs moduli of the unidirectional 0 and 90 layers. The stress equilibrium

equation is:

90

r

t90 + 0r t0 = 0

with ti the layer thickness. This equation then leads to the following relation for

the thermal stress in 90 layer:

E t E ( 0 )T

90r = 90 0 0 90

t 0 E0 + t 90 E90

laminated plate.

k-ply engineering constants k-ply stiffness in ply local CS

(E1)k, (E2)k, (G12) k, (12) k, [C]k, []k , []k (3.18, 3.32)

(1)k, (2) k, (1) k, (2) k

k [C*]k , []k , []k (3.28, 3.35)

[A], [B], [D] (4.16-19)

Laminate compliance

[a], [b], [d] (4.22-23)

(mechanical, thermal, moisture) [0], [] (5.18)

[*]k (4.10)

[*]k (3.27, 5.9)

[]k (3.21)

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.1

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

6. Laminate failure

6.1. Introduction

It has been shown in the first lectures that composite laminates have anisotropic

thermo-mechanical properties. The behaviour of laminates at failure is also highly

directional in nature. The strength of the constituents plays an important role in this

anisotropy. For example, the strength of an E-glass fibre is around 1750 MPa,

compared to a value of about 50MPa for an epoxy matrix. This means for example

that a unidirectional layer will sustain a higher stress when loaded parallel to the

fibres than perpendicular to the fibres.

When studying the (potential) failure behaviour of laminates, the term strength

alone is not appropriate. It suggests that the problem is limited to a simple change in

state from non-failed to failed, and would not provide a realistic representation

of the complexity of laminate failure.

The presence of fibres makes a composite a structure having an infinite amount of

potential stress concentrations for a crack to initiate. More than this, the presence of

fibres and layers makes a composite a sum of a nearly infinite amount of surfaces,

all of which are a potential site for a crack to grow1. This means that even if a

certain amount of cracks occurs within a composite, it does not necessary mean the

complete fracture of the structure. There might still be enough cohesion between

fibre and matrix to carry load. This makes the qualitative description, i.e. the way

the laminate chronologically fails an important part of a composite failure study.

The quantitative prediction of failure (ultimate stress, strain or energy) is of use for

the designer. However, it will be seen that the prediction of failure is mostly

phenomenological (only the occurrence of failure is predicted, not the actual mode

of failure). They only represent the level of stress at which failure occurs.

Both qualitative and quantitative approaches can be performed at different scales

(microscopically: the interface between fibre and matrix fails, macroscopically: a

delamination separating two adjacent layers).

These lecture notes will first present an overview of in the literature often

mentioned failure mechanisms. Then, simple mechanics of material failure criteria

aiming at predicting the initiation of a failure history will be described. A typical

failure analysis of a laminate is then presented. A short insight in fracture

mechanics is then given.

1

This typical characteristic of composite is used in structures made to absorb impact energy: Such a

structure is designed as such that as many crack surfaces as possible are created.

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.2

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

A short list of failure mechanisms will be given here.

This is often the first damage mechanism to occur. This type of crack grows parallel

to the fibre and in the thickness direction of the laminate. The formation of

transverse cracks does rarely mean the total fracture of a laminate, as it does not

affect the load carrying capacity of the fibres. However, transverse cracking

influences the mechanical and thermal properties of the laminate. Most importantly,

this type of cracking forms a trigger for further damage mechanisms, and provide

paths for the diffusion of surrounding fluids leading to accelerated environmental

degradation.

A simple illustration of a transverse crack is given in fig. 6.1, where a [90/0]s beam

is subjected to a three point bending test. In this case the crack initiates at the free

surface of the 90o layer, and grows in the thickness direction until it meets the fibres

of the 0o layer. On a macroscopic scale, the crack can be approximated by a plane,

which runs perpendicular to the principal stress. Microscopically, an example of a

micrograph (fig. 6.2) shows that the crack runs between fibre interfaces, also

through matrix region when locally the matrix fraction is high.

b

h t0

y

45o t90 /2

x

z d

L

(a) (b)

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.3

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

fig. 6.1: Transverse cracking in a [90/0]s laminated beam in a three point bending loading situation.

fig. 6.2: Micrograph of cross-sectioned 90o layer in the region of a transverse crack..

mechanics of material failure criteria, as described in 6.3.

6.2.2. Delamination

The term delamination is directly related to the process of laminating. Just in

the same way as producing a laminate is about bonding different layers to each

other, delaminating is about debonding the layers. It is therefore a crack, which

again runs in a plane parallel to the fibres, but at the interface between two layers

(can also within a layer).

Chronologically, it is recognised that a delamination mostly initiates from the tip of

a transverse crack. Using a similar illustration as for the transverse crack, the

transverse crack is shown to grow at the interface between the 0o and the 90o layer.

Another example of the combination transverse crack delamination is given on a

micrograph of a cross-section of a glass-reinforced epoxy tube subjected to an

impact test. Delamination will further affect the thermo-mechanical properties of

the laminate, for example its bending stiffness. However, the presence of

delaminations does not generally lead to the complete fracture of the laminate.

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.4

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

90o

0o

y 90o

x

Transverse crack Delamination

z

fig. 6.3: Transverse cracking in a [90/0]s laminated beam in a three point bending loading

situation.

fig. 6.4: Cross-section of an impacted composite tube illustrating transverse cracks and

delaminations.

of transverse cracking. The growth of a delamination is preferably described using

fracture mechanics energy criteria, as discussed in 6.5.

It is obvious that the fracture of fibres decreases significantly the load carrying

capacity of a laminate. Chronologically, matrix related cracks (transverse crack +

delamination) are mostly present in the laminate, as shown in fig. 6.5.

fig. 6.5: SEM (Scanning Electron Microscope)picture showing delaminations and fibre fracture

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.5

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

Fibre related failure can also involve local buckling of fibre, or fibre pull-out, as

shown in fig. 6.6.

fig. 6.6: Micrograph showing a fibre bundle buckling (left), and a SEM showing fibre pull-out

on a fracture surface

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.6

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

It should first be made clear that failure criteria are not based on any physical theory

of failure. They are not concerned with the details of micromechanical failure

modes like the growth of a defect, fibre pull-out or interfacial failure. A mechanics

of material strength criterion merely defines a stress (or strain) limit upon which

failure occurs. This can also be done in a biaxial stress space (or triaxial), where

failure envelopes are defined. The stress limit or the envelope are related to basic

lamina strength values. These lamina strength are obtained by measurements on

unidirectional lamina, and in such a way that a uniform stress state is imposed on

the specimen. The envelopes describing a multiaxial state of stress are then defined

by "curve fitting" the basic experimental failure data.

Most of these criteria are based on existing criteria used for isotropic material. For

example, the Tsai-Hill criterion discussed later is based on the von Mises criterion,

which was originally meant for describing yielding.

In the following, we will use the 5 lamina strength:

s1t, s1c, s2t, s2c, s6,

where the subscript 1, 2, and 6 relate to the loading directions parallel to the fibres,

perpendicular to the fibres and to the in-plane shear respectively. The second

subscript t and c relates respectively to the tensile and the compression loading. In

most criteria found in the literature, all numerical values are assumed positive.

The Maximum Stress Criterion compares the stress occurring in a layer (in the layer

local coordinate system) to the corresponding lamina strength. No failure occurs if

the following set of inequalities is satisfied:

s1c < 1 < s1t

s2 c < 2 < s 2 t (6.1)

s6 > 6

No interaction between the stress or strain components are taken into account.

The Maximum Strain Criterion is an alternative to the stress criterion. The

inequalities are:

e1c < 1 < e1t

(6.2)

e2 c < 2 < e2 t

e6 > 6

The limits described by these criteria (failure surface) are shown in a 1-2 space in

the following figure:

2

1

Maximum Strain Criterion

fig. 6.7: Maximum stress and strain criterion in 1-2 space

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.7

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

Quadratic interaction criteria include terms which allow for the interaction between

the different stress components. For example, the Tsai-Hill criterion in 2D plane-

stress predict failure if this inequality is fulfilled:

12 1 2 22 62

2 + 2 + 2 >1 (6.3)

s12 s1 s2 s6

In this form, the difference in tensile and compression strength properties of the

basic lamina is not taken into account. The failure surface is therefore a symmetric

ellipse. Taking into account the difference in tensile and compressive strength is

done with Tsai-Hill by using the appropriate values of s1 or s2 according to the

considered quadrant of the failure surface. This is shown in fig. 6.8.

2

1

Tsai-Hill Criterion

fig. 6.8: Maximum stress and strain criterion in 1-2 space

failure criterion, which takes more cross-terms into account.

A typical laminate analysis will be developed here, where the initiation of failure is

predicted using the mechanics of material failure criteria presented earlier. A stress

analysis will first be performed. This will summarise some of the previous chapters.

A Carbon reinforced Polyetherimide laminate of total thickness h is made of

unidirectional layers. The lay-up considered is quasi-isotropic [0/45/-45/90]s.

Thermal stresses are not taken into account. The laminate is loaded by a tensile

force in the x-direction inducing a global stress x. The following material data is

available:

E1 E2 G12 12 1 2

Stiffness data

(x109 Pa) (x109 Pa) (x109 Pa) (m/m.oC) (m/m.oC)

Carbon-PEI

(x106 Pa) (x106 Pa) (x106 Pa) (x106 Pa) (x106 Pa)

1400 1200 35 150 60

table 6.1: Measured unidirectional Carbon-PEI in-plane properties

Calculate the force (or the stress) at which the first layer (and which one) fails. Use

for this purpose both simple stress and Tsai-hill criteria.

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.8

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

Stress analysis.

The stress analysis follows the procedure described in 5.4.

The laminate is built up from the same type of layer. The stiffness matrix of all

layers in their local coordinate system is therefore:

E1 21 E1

1 0

1 12 21

12 21

120.8 2.5 0

0 = 2.5 7.8 0 ( x 10 9 Pa )

E E2

[C ] = 21 1 (6.4)

1 12 21 1 12 21

0 0 G12

0 0 3 . 5

The stiffness matrix of each layer can then be rotated in their global coordinate

system. The values obtained for the different layer orientations are given in the

following table:

k *

C11 *

C 22 *

C12 *

C16 *

C 26 *

C66

0o 120.8 7.8 2.5 0 0 3.5

45o 36.9 36.9 29.9 28.2 28.2 30.9

90o 7.8 120.8 2.5 0 0 3.5

table 6. 2: Stiffness matrix components in the global coordinate system

The thickness of each layer is h/8. The laminate extensional stiffness matrix can be

written relative to the laminate thickness h:

0o 45 o 45 o 90 o

A11

= 14 120.8 + 14 36.9 + 14 36.9 + 14 7.8 = 50.6 x 10 9 Pa

h

A22

= 14 7.8 + 14 36.9 + 14 36.9 + 14 120.8 = 50.6 x 10 9 Pa

h

A12

= 14 2.5 + 14 29.9 + 14 29.9 + 14 2.5 = 16.2 x 10 9 Pa

h (6.5)

A16

= 14 0 + 14 28.2 + 14 (28.2) + 14 0 = 0 x 10 9 Pa

h

A26

= 14 0 + 14 28.2 + 14 (28.2) + 14 0 = 0 x 10 9 Pa

h

A66

= 14 3.5 + 14 30.9 + 14 30.9 + 14 3.5 = 17.2 x 10 9 Pa

h

Only in-plane forces are applied to the laminate and the bending stiffness matrix

[D] does not need to be evaluated. The laminate considered is symmetric, which

implies that the components of the coupling [B] stiffness matrix are all zero. The

[A] matrix can therefore be inverted directly to give an expression for the mid-plane

strain vector {0}.

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.9

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1

1 50.6 16.2 0 0.022 0.007 0

[ A]

= h [a ] = 16.2 50.6 0

= 0.007 0.022 0 x 10 9 m 2 N

h (6.6)

0 0 17.2 0 0 0.058

The laminate strain vector is {0}=[a]{N} with {N}={hx 0 0}T, which gives:

x0 = a11 x h = 0.022 x 10 9 x

y0 = a 21 x h = 0.007 x 10 9 x (6.7)

0

xy = a 61 x h = 0

In-plane stresses can now be calculated for each layers. Expressions for these

stresses relative to the loading stress x are first given in their global coordinate

system:

1* 0 y0

= C11* x + C12* = C11* 0.022 x 10 9 C12* 0.007 x 10 9

x x x

2* * x

0

y0

= C 21 + C 22

*

= C 21

*

0.022 x 10 9 C 22

*

0.007 x 10 9 (6.8)

x x x

6* * x

0

y0

= C 61 + C 62

*

= C 61

*

0.022 x 10 9 C 62

*

0.007 x 10 9

x x x

This gives for the different layers:

0o 45 o 45 o 90 o

1*

x* 2.64 0.602 0.602 0.154

2 (6.9)

= 0.0004; 0.399; 0.399 ; 0.7906

x 0 0.423 0.423 0

6*

x

A useful check at this stage is to calculate the average of the stress vector

components over the thickness. For the loading case studied, the average of the 2*

and 6* components should be zero. The average of the 1* component should give

.

The stresses must now be rotated in their local coordinate system in order to apply

the failure criterion, i.e. {}k=[T]k{*}k

The transformation matrix are for the lay-up considered:

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.10

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

1 0 0 12 1

2 1 12 1

2 1 0 1 0

0 1 0; 1 1

1; 1 1

1 ; 1 0 0 (6.10)

2 2 2 2

0 0 1 12 1

2 0 12 12 0 0 0 1

0o 45 o 45 o 90 o

1

x 2.64 0.924 0.924 0.7906

2 (6.11)

= 0.0004; 0.078 ; 0.078 ; 0.1541

x 0 0.101 0.101 0

6

x

Failure criteria.

The Tsai-Hill criterion (6.3) can be written in relation to x:

12 1 2 22 s12 62 s12 s12

+ 2 2 + 2 2 = 2 (6.12)

x x x s2 x s6 x

2 2

And therefore:

12

12 1 2 22 s12 62 s12

x = s1 2 + 2 2 + 2 2 (6.13)

x s 2 x s6

2

x x

In the 0o layer, relation (6.11) shows tension along the fibres, hardly some tension

transverse to the fibres. In the modified Tsai-Hill relation, s1 is therefore replaced by

s1t and s2 by s2t:

12

2 2 s2 2 s2

x = s1t 12 1 2 2 + 22 12t + 62 12t

x x x s 2t x s6

12

(6.14)

1400 x 10 6

2

x = s1t (2.64) 2 (2.64)(0.0004) + (0.0004) 2

35 x 10

6

In the 45o layer, s1t and s2t can be filled in the equation:

2 2

2 1400 x 10 1400 x 10 6

6

x = s1t (0.924) (0.924)(0.078) + (0.078)

2

+ (0.1015) 2

35 x 10

6

60 x 10

6

(6.15)

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.11

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

The Tsai-Hill equation for the -45o layer is similar to the latest one.

In the 90o layer, the results given in (6.11) show compression in the 1-direction,

tension in the 2-direction requiring the use of s1c and s2t:

12

2

2 1200 x 10

6

x = s1c (0.7906) (0.7906)(0.1541) + (0.1541)

2

(6.15)

35 x 10

6

This gives:

0o 45 o 90 o

x x (6.16)

= 0.38; 0.25; = 0.19

s1t s1c

This means that failure will first occur in the 90o layer. This will occur at an

external stress x of:

x = 0.19 s1c = 0.19 (1200) = 228MPa (6.17)

Filling this value in relation (6.11) for the 90o layer shows that the failure occurs

transverse to the fibre.

Application of the maximum stress criterion shows that the first failure occurs in the

90o layer, transverse to the fibre direction (2-direction), at an external stress x of

2.27 x 108 N/m2. It is most plausible that the type of failure that occurs here is a

series of transverse cracks, which form planes perpendicular to the loading

direction. The stress profiles in the layers at this level of external stress are given in

the global coordinate system and the local coordinate system in respectively fig. 6.9

and fig. 6.10.

0o

45 o

45o 45o

-45o

90o 1 (x 108 N/m2) -45o

90

o 2 (x 108 N/m2) -45o

90o 6 (x 107 N/m2)

-2 0 2 4 6 -2 0 2 4 -10 -5 0 5 10

fig. 6.9: Stress profiles in the layer global coordinate system 1*, 2*, 3

0

45 o

45

o 45o

-45o

1 (x 108 N/m2) 2 (x 107 N/m 2) -45o

6 (x 107 N/m2)

o

-45

90o 90o 90o

-2 0 2 4 6 -2 0 2 4 -2 0 2 4

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.12

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

As pointed out in the introduction of this alinea, mechanics of material strength or

strain criteria are not based on any physical theory of failure. Besides this, it is

worth adding that the 5 lamina strength mentioned in 6.3 are no material properties.

This will be illustrated with two examples. The first one illustrates the difficulty to

obtain values for the critical strength (or strain) values used in the different criteria.

The second will show the dependency of geometry on the results obtained with

mechanics of material criteria.

Let us consider the ultimate stress transverse to the fibres and in tension s2t. The

usual approach to measure this value is a tensile test. Typically, the force-strain

diagram in such a test will grow linearly until a crack initiates in a direction

transverse to the fibres. At this point the crack mostly grows through the whole

laminate as no obstacles are present (like a 0o layer) and fully fractures it. As an

example, tests performed on five [906c]s carbon-polyetherimide tensile coupons

(150 mm x 25 mm x 2 mm) gave an average axial stress at fracture s2t of 35 MPa.

As shown earlier, this value can be used in any of the stress-based criteria for

predicting the failure of a laminate. However, testing the same [906c]s under three

point bending conditions already shows the dependency of s2t on the type of test

used. Under these conditions, a crack will initiate and fracture the specimen at aa

average stress of 70MPa. This shows that s2t certainly cannot be considered as a

material property.

An explanation for the difference found between these two tests can be suggested

when assuming that the formation of a crack is due to the presence of a defect,

mostly due to fabrication. The problem becomes statistical, and it can be assumed

that in any material there exists a certain defect distribution. In a tensile test, the

axial stress is constant over the whole cross-section and the full length of the

laminate. The worst defect present in the laminate will therefore act as the

weakest link and will lead to the formation of the crack. In a three point bending

test, the bending stress is not constant. It varies from compressive to tensile over the

cross-section, and is maximum at half the beam length. The probability that the

location of the weakest defect leading to the crack coincides with the location

where the stress is maximum is small. This can explain why the first crack occurs at

a lower average stress in a tensile test than in a bending test.

An other example is a well-studied subject in the seventies. Several authors have

considered the (multiple) failure of the 90o layer within a laminate ([0/90n/0])

subjected to a tensile force. The strain at which a 90o lamina ([90n]) fails is

independent of the amount of layers n used. In the laminate however, they found

that the strain at which cracking occurs is inversely proportional to the thickness of

the 90 layer. This means that in this case the use of a maximum strain criteria is no

valid. The behaviour described is illustrated in fig. 6.11, where the strain at the

onset of cracking is set against the thickness of the 90o layer. This effect is referred

to as the constraining effect of the 0 layer, but can also be explained in terms of

statistics as in the last example.

SI T ES G

O

Composites course 2008 University of Twente, MechEng 6.13

R

M

OU

CO

6. Laminate failure - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

[0/90n/0] laminate

Fracture mechanics based criteria mainly differ from mechanics of materials criteria

in one major aspect. It is assumed that materials contain defects, which are mostly

introduced during manufacturing. A fracture mechanics based failure criterion

assumes the presence of a defect/crack and studies the growth/no growth of the

considered crack. A crack will grow when a certain amount of energy per unit area

is available. This critical amount of energy is called the critical energy release rate

Gc, in J/m2 and is determined experimentally. It is well recognised that a crack can

grow following a combination of the three crack propagation modes illustrated in

fig. 6.12.

The mode I is the opening mode, mode II the sliding mode, mode III the tearing

mode.

6.5. Problems

6.1 Use the example derived in 6.4. The stress at which the first ply fails has been

predicted at a level of 2.27 x 108 N/m2. This concerns the 90o plies. Calculate

the stress at which the next ply fails according to two different assumption:

a. All 90o ply stiffnesses are set to zero.

b. Since transverse cracking was the failure assumed to occur for this first

ply failure, set the elasticity moduli of the 90o layer related to the direction

perpendicular to the fibres to zero, i.e. E2 = G12 = 21 = 0

For this purpose, you need to re-evaluate the laminate extensional stiffness

matrix [A].

S I T ES G

Composites 2007

O

RO

M

CO

UP

Composite Production Methods

TE

U NI

EN

V

R

W

SI T

TY OF

Remko Akkerman

Laurent Warnet

Composites 2007

glass thermoplastic

carbon

...

thermoset

MEASURE heater

pump

water

cooling

CFRP

impregnation continuous fibre thermally induced distortions of single

curvature composites

reinforced plastic

distortion

prepreg measurement

MODEL

al

c

bri

on

fa

c ti

residual stresses

en

ire

production

ov

id

un

thermomechanical properties of woven

autoclave fabric reinforced composites

RTM

hot pressing

rubber moulding lightweight part

... low density

high stiffness spring forward

CONTROL

high strength optimise process by inverse analysis

45o usually thin walled

good corrosion resistance

damage development in

cross-ply beams assembly

MEASURE

lightweight structure

micromechanical

inservice loading

stress analysis

of curved panels

MODEL mechanically induced stresses structural deformation

crack initiation & crack growth and failure

compression after impact

CONTROL damage resistant & damage tolerant designs

optimise for impact energy absorption

Thats about Impregnating the fibres

Wetting

Composites 2007

1

Composite fabrication techniques:

Impregnating a porous medium

Uncured Thermoset

+ pressure

Thermoplast above Tm

Composites 2007

Direct methods:

Direct impregnation

Resin

Product

Layer after layer

Composites 2007

Direct method:

Hand lay-up

(Polyester)

Low tooling costs

Low performance

Emission

Small series

Prototyping

large parts

Low control!

Composites 2007

2

Direct method:

Filament winding

Thermoset

Low tooling costs

Medium performance

Low-Medium series

Large parts

Oven for curing

Composites 2007

Direct method:

Filament winding:

Composites 2007

Direct method:

Pultrusion / Pull-winding / Pull-braiding

Thermoset

Continuous process!

Medium tooling costs

Medium-high performance

Medium-Large series

Small-medium cross-sections

Composites 2007

3

Direct method:

Pultrusion / Pull-winding / Pull-braiding

Composites 2007

Semi-Direct methods:

Direct impregnation of a dry preform

Preform

Resin

Product

On basis of dry preform

Composites 2007

Semi-direct method:

Liquid Composite Moulding (LCM)

Several variants:

Composites 2007

4

Semi-direct method:

Resin Transfer Moulding (RTM)

High tooling costs

High performance

Medium series

Small medium size

Composites 2007

Semi-direct method:

Resin Transfer Moulding (RTM)

Composites 2007

Composites 2007

5

Semi-direct method:

Resin Infusion under Flexible Tooling (RIFT)

Vacuum outlet

Bagging film

Sealant tape

Resin inlet

Low tooling costs

Medium performance

Medium series

Composites 2007 Medium Large size

Semi-direct method:

Resin Infusion under Flexible Tooling (RIFT)

Composites 2007

Semi-direct method:

Liquid Composite Moulding (LCM)

Composites 2007

6

Semi-direct method:

Liquid Composite Moulding (LCM)

Viscosity

Permeability=f(Fibre structure /

orientation / concentration)

K P

Darcys law: Q= (Macro)

x

Composites 2007

Semi-direct method:

Liquid Composite Moulding (LCM)

Flow on a mesoscale:

Composites 2007

Semi-direct method:

Liquid Composite Moulding (LCM)

Composites 2007

7

Semi-direct method:

Liquid Composite Moulding (LCM)

Common characteristics:

Function integration:

Composites 2007

Multi-step methods:

Composite processing based on prepreg

Fibre Yarn

/ Fabric Prepreg

Product

+ Resin

Semi-Product

Product

On basis of preimpregnated layer

Composites 2007

Multi-step methods:

Autoclave curing

High labour costs

High tooling costs

High performance

Long cycle time

Small-large products

Function integration

Composites 2007

8

Multi-step methods:

Autoclave curing

Constraints on prepreg:

Flexibility

Drapability

Tack

Thermoset

with thermoplastics

Composites 2007

Multi-step methods:

Autoclave curing

Vacuum:

(Debulking)

Withdrawing volatiles

Composites 2007

Multi-step methods:

Rubber pressing

Needs semi-product

Thermoplastic matrix

Low-medium labour costs

High tooling costs

High performance

Short cycle time

Small-medium products

Composites 2007

9

Multi-step methods:

Tape placement

Tailor-made

Thermoplastic

Thermoset (+autoclave)

Low labour costs

Medium-high tooling costs

High performance

Medium cycle time

Medium Large products

Composites 2007

Deformation due to forming:

Fibre reorientation

S S

S

S

S S S

S S

S S

S

Composites 2007

Deformation due to forming :

Composites 2007

10

Choice of a fabrication techniques:

Parameters playing a role:

Material (Matrix)

Performance

Vf

Fibre distribution

Void content

Control of the process

Reproducibility

Shape complexity

Assembly

Composites 2007

Direct methods:

Hand lay-up

Hand lay-up Productivity

5

4

Series size 3 Part size / cost

0

Function Complex

integration geometry

Technical

Performance

Simplicity

Composites 2007

Direct methods:

HandWinding

Filament lay-up Productivity

5

4

Series size 3 Part size / cost

0

Function Complex

integration geometry

Technical

Performance

Simplicity

Composites 2007

11

Direct methods:

Hand lay-up

Pull-Processes Productivity

5

4

Series size 3 Part size / cost

0

Continuous Process Function Complex

Thermoset integration geometry

Tailor-made lay-up

Technical

Performance

Simplicity

Composites 2007

Semi-direct methods:

Hand lay-up

Resin Transfer Moulding Productivity

5

4

Series size 3 Part size / cost

0

Function Complex

integration geometry

Only for Thermosets

Technical

Performance

Simplicity

Composites 2007

Semi-direct methods:

Hand

Resin Infusion lay-up

Flexible Tooling Productivity

5

Vacuum outlet

4

Bagging film Series size 3 Part size / cost

Sealant tape

Resin inlet

2

integration geometry

Technical

Performance

Simplicity

Composites 2007

12

Multi-step methods:

Hand lay-up

Autoclave Productivity

5

4

Series size 3 Part size / cost

integration geometry

Technical

Performance

Simplicity

Composites 2007

Multi-step methods:

Hand

Rubber lay-up

Pressing Productivity

5

4

Series size 3 Part size / cost

integration geometry

Technical

Performance

Simplicity

Composites 2007

Multi-step methods:

Hand lay-up

Tape layer Productivity

5

4

Series size 3 Part size / cost

integration geometry

Technical

Performance

Simplicity

Composites 2007

13

Composites Course University of Twente, Eng&Tech 8.1

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

8.1 Introduction

The CLT (Classical Lamination Theory, see Ch.4) provides a means for quick analyses of

composite laminate stiffness and strength. For complicated geometries, three-dimensional

stress analyses or large deflections one often reverts to finite element simulations. This lecture

will give a survey of the basics and some applications.

The finite element method can be based on the minimum of potential energy [1]. For an

elastic medium, the potential energy can be represented by:

1

2

= ij ij d ui f j d v j t j d = (7.1)

,

1

2

ij Cijkl kl d ui f j d v j t j d

in which is the total volume and the boundary surface of the continuum. The first

volume integral of equation (7.1) contains the stiffness matrix and the displacement field of

the system, whereas the second represents the volume integral of the body forces per unit

mass, which is supposed to be zero. Finally, the integral over the boundary surface contains

the applied forces. In terms of virtual work / minimum potential energy this leads to

1

2

= ij ij d ui f j d v j t j d = 0 . (7.2)

kl =

1

( uk ,l + ul ,k ). (7.3)

2

or

in which kl are the strains which can be written as a function of the displacement gradient.

The finite element matrix [B] contains the first derivatives of the shape functions1.

Substituting equation (7.4) into (7.2), minimised with respect to the displacements results in:

T

(7.5)

14442444

3

K

1

N.B. Both FE and CLT literature use B as a de facto standard term. Here, the choice was

made to use [B] for the FE matrix with derivatives of the shape functions, and B for the CLT

(strain-curvature coupling) matrix.

Composites Course University of Twente, Eng&Tech 8.2

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

in which [D] is the material stiffness matrix and {f} contains the externally applied forces. In

simplified form this is the familiar elastic equilibrium equation for a single element:

[K ]e {u}e = {R}e . (7.6)

The stiffness matrix for the whole system can be obtained by recognising the displacements

are continuous over the element boundaries. Adding all element contributions for a node n

leads the nodal force equilibrium,

m (7.7)

{F} n = {R} ne ,

c =1

in which {F}n is the applied nodal force and {R}en is the contribution of element e to the

reaction force in node n, where m is the number of elements that belong to node n.

Substituting equation (7.7) into equation (7.6) results in:

m (7.8)

{F }n = [K ]ne {u}ne .

c =1

Assembly of equation (7.8) for every node (or, usually, for every element) results in the

equilibrium equation for the entire system,

Fibre reinforced composite products are often thin-walled, h/L 1, where h is the laminate

thickness and L is a characteristic length. Hence, very often layered plate or shell elements

can be used for finite element analyses of these types of structures. In general, plate elements

will employ the Kirchhoff theory, which we already know from the CLT. The midplane

strains and the curvatures completely describe the deformation within an element, leading to

stresses and, integrated over the thickness, to forces and moments. Extensions can be made to

include transverse shear (important for sandwich structures, for instance) as in Mindlin type

elements.

Composites Course University of Twente, Eng&Tech 8.3

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

Often, elements with 3 displacement and 3 rotation degrees of freedom per node will be used.

The inplane displacement components will be used to determine the midplane strains, the

inplane rotations (and sometimes the out-of-plane displacement) will be used to determine the

curvatures,

xx u0, x xx y , x

{ o } = yy = v0, y = yy = x , y (7.10)

u + v +

xy 0, x 0, y xy y , y x ,x

.

With those, the strain distribution follows from

{ (z ) } = { 0 }+ z { } , (7.11)

which can be substituted into the volume integral in (7.2), split into an integral over the

element surface and the element thickness,

=

1

2Ah {

0 + z } . {[Q ] { 0 + z }}dz dA vi ti d .

T

(7.12)

For the layered composite elements, [Q] is the stiffness matrix according to CLT, A the

elements surface, h the element thickness and z the thickness co-ordinate. Evaluating this

equation results in:

2 3 2

1

{ 0 }T [Q ] { 0 }dz + { 0 }T h [Q ] { }dz + { }T h

[Q ] { }dz + { } [Q ] { 0 }dz dA

h

2 A

= h{

T

B D B

v i t i d 0 ( ) ,

Composites Course University of Twente, Eng&Tech 8.4

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

where the virtual strains and curvatures are related to the virtual displacements and rotations

by means of the finite element [B]-matrix,

{ }= [B ] {u},

0

u

{ }= [B ] { }.

0

(7.14)

T A B d

[B ]

A B D

[B ] dA = {R } ,

d b

(7.15)

where we recognise the familiar ABD-matrix. Further, [B] contains both [Bu] and [B], and

{dd} and {d} contain the displacement and rotation degrees of freedom, respectively. Note

that in this way the resulting forces and moments over an element are obtained. Equation

(7.15) is usually solved numerically using Gauss quadrature. As in the CLT, the stiffnesses are

integrated over the thickness of an element. Hence only the mid-plane displacements and

rotations serve as degrees of freedom, keeping the FE-system matrix to a very moderate size,

regardless of the number of layers in the laminate.

When the stress analysis has to include three-dimensional aspects (for instance at free edges

or connections), plate theory may be an over-simplification. One can go into full three-

dimensional analyses with one or more elements per layers. These elements then have to be

capable of simulating orthotropic properties. In this case, the values for Ex, Ey, Ez, Gxy, Gxz,

Gyz, xy, xz, yz (in the principal material directions) need to be supplied by the user, as well as

the orientation of the principal material axes. This can be an orientation with respect to the

global coordinate system, or with respect to the element axes, or with respect to some user-

defined coordinate system. The amount of data required easily leads to mistakes in the input

file for FE-analyses. This requires step-by-step generation of any model with meticulous step-

by-step verification!

With multi-layered structures, of course, it is easy to create a massive three-dimensional FE-

model requiring corresponding CPU-time. To this end, layered solid elements were

developed, similar to the shell elements discussed earlier. In this case, the plane stress

assumption will be replaced by another approximation of the through-thickness stress-strain

relations. Inplane, the same logic is followed as in the CLT type theories: the inplane strains

are assumed to be continuous over the thickness, so the inplane stresses will be integrated

over the thickness to find the inplane reaction forces. Out-of-plane, however, a different logic

is more applicable. The transverse shear and normal stresses are more likely to be continuous,

and hence the layer strains will be integrated to find the total transverse strain components.

Consider the three-dimensional stress-strain relation for transversely isotropic plies in their

principal coordinate system (123-coordinates),

Composites Course University of Twente, Eng&Tech 8.5

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

1 12 12 0 0 0

E1 E1 E1

1 23 1

12 E 1 0 0 0

E2 E2

2 1

23 2

3 12 1 0 0 0 3

= E1 E2 E2

(7.16)

0

23 0 0 1 0 0 23

13 G23

13

0 0 0 0 1 0

12 G12

12

0 0 0 0 0 1

G12

or, symbolically,

{ 123 } = [S ] { 123 } { 123 } = [S ]1 { 123 } [Q ] { 123 } (7.17)

Transformed to the laminate xyz-coordinates (due to an inplane rotation ) we find

{ }= [T ] [Q ] [R] [T ] [R] { } [Q ] { },

xyz

1 1

xyz

*

xyz (7.18)

where the transformation matrix [T] is applied, given by

cos 2 ( ) sin 2 ( ) 0 0 0 2 sin ( ) cos( )

sin ( ) cos 2 ( ) 2 sin ( ) cos( )

2

0 0 0

[T ] = 0 0 1 0 0 0

(7.19)

0 0 0 cos( ) sin ( ) 0

0 0 0 sin ( ) cos( ) 0

sin ( ) cos( ) sin ( ) cos( ) 0 0 0 cos 2 ( ) sin 2 ( )

1 0 0 0 0 0

0 1 0 0 0 0

0 0 1 0 0 0 .

[R] = (7.20)

0 0 0 2 0 0

0 0 0 0 2 0

0 0 0 0 0 2

Hence, the three-dimensional off-axis stress-strain relation for a unidirectional layer is given

by

(i )

x (i ) Q11* Q12* Q13* 0 0 Q16* x

(i ) * *

y Q12 Q22

* *

Q23 0 0 Q26 (yi )

z Q13* Q23 * *

Q33 0 *

0 Q36 z

=

(i ) , (7.21)

yz 0 0 0 Q44*

Q45 0 yz

*

(i )

xz 0 0 0 Q45* *

Q55 0 xz

(i ) *

xy Q16 Q26 xy

* * *

Q36 0 0 Q66

Composites Course University of Twente, Eng&Tech 8.6

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

where the transformed stiffness matrix components Qij* are all a linear combination of sin(k),

cos(k) with k=2,4 and as the rotation with respect to the lamina principal axes. The

components denoted with superscript (i) can be discontinous functions across the layer

boundaries. The components z, xz, yz, x, y and xy will be continuous functions across the

thickness of the laminate.

A possible approach in a three-dimensional element can be to bring all layer variables to the

left hand side by matrix manipulation,

(i ) (i )

x M 11* M 12* M 13* 0 0 M 16* x

* * * *

y M 12 M 22 M 23 0 0 M 26 y

z M 31

* *

M 32 *

M 33 0 0 *

M 36 z

= (7.22)

yz 0 yz

* *

0 0 0 M 44 M 45

xz 0 0 0 *

M 54 *

M 55 0 xz

*

xy M 16

*

M 26 *

M 63 0 0 *

M 66 xy

and subsequently add (or integrate) the layer variables over the thickness in order to find the

element stiffness matrix. Alternatively, an effective transverse modulus can be computed by

inverting an averaged compliance, as in

H

E z = n (i )

eff

h

(7.23)

(i )

i =1 E z

which is then directly applied in a layer stiffness contribution as in (7.21), which can

subsequently simply be added to the full laminate stiffness [1]. A detailed discussion on

transverse strains can be found in [3].

The main conclusion is that certain approximations have to be made to find a reasonable

description of the transverse behaviour within acceptable CPU-time. Users should perform

patch tests to get an indication of the accuracy of the method employed in their specific

application.

8.2.3 Nonlinearit y

Sofar, only linear elastic behaviour has been discussed. The thin-walled composite structures

may, however, easily experience deflections far greater than the laminate thicknesses. In this

case, the relation between strains and displacements easily surpasses the linear region, and

geometric nonlinearity (GNL) has to be taken into account.

Consider the large deflections in the following beam due to bending:

x

x0

x

Composites Course University of Twente, Eng&Tech 8.7

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

u

x = (7.24)

x

is used in this case, this will lead to an incorrect large inplane strain, as the length of the strip

is practically unchanged under perfect bending. This can be corrected by adding nonlinear

terms to the strain definition,

u 1 w

2

x = + , (7.25)

x 2 x

which can be easily verified by using Pythagoras theorem and subsequent linearisation.

GNL also plays an important role in buckling simulations, where bending is induced by

inplane loads. Consider for instance a simulation of a compression test of a rectangular plate.

The first and second buckling modes are depicted in Figure 7.3

Figure 7.3 The first two buckling modes for a simply supported plate under inplane compressive load.

Both modes of buckling affect the inplane stiffness significantly. The inplane load-

displacement predictions are depicted in Figure 7.4.

Composites Course University of Twente, Eng&Tech 8.8

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

20

in plane displacement (mm)

0

-1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2

-20

Mode 1

load (kN)

-40 Mode 2

straight

-60

-80

-100

Figure 7.4 Inplane loads versus deflection for the inplane compression test.

Further nonlinearity can arise from the material behaviour. During laminate production, the

matrix will turn from a fairly viscous state to approximately an elastic solid. Special purpose

material models are under development to capture these types of (physical) nonlinearity.

Transverse to the fibres the material behaviour is matrix-dominated. For certain matrix

materials, this can result in creep, another physically nonlinear phenomenon. Further

nonlinearities are of course found during damage growth.

Having performed a stress analysis, it is fairly straightforward to assess failure criteria in

terms of stress or strain components, such as a maximum stress or strain criterion, a Tsai-Hill

or a Tsai-Wu criterion. An interesting review of failure criteria is given in [4]. Again, the

more complicated criteria require many input parameters, which, most likely, will not be

readily available. Further, the use of these criteria requires meticulous step-by-step

verification by the user to ensure reliable results.

Having performed a strength analysis, the finite element simulation can be carried a step

further. Continuum damage theories (such as Chang Chang [5]) assume some stiffness

degradation (softening) after a failure criterion has been satisfied. Typically, one or more

damage parameters are used, with which the current stiffness is determined as

E (d ) = (1 d ) E0 , (7.26)

with d as the damage parameter (zero for perfect material, one for complete failure) and E0 as

the stiffness of the undamaged material. Theories for anisotropic materials describe different

stiffnesses, with possibly different damage parameters for the different failure modes (e.g.,

matrix or fibre fracture). Simulations of softening are susceptible to mesh-dependent results: a

Composites Course University of Twente, Eng&Tech 8.9

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

proper analysis must, however, give convergent results with mesh refinement (to be checked

by the user).

As an alternative, fracture mechanics methods can be applied in FE simulations. Initially,

only the crack growth criteria were evaluated. Special crack tip elements effectively

incorporate the stress singularity around a crack tip. Using those, stress intensity factors can

be evaluated and compared to the corresponding critical values. More generally, energy

criteria can be evaluated. The Virtual Crack Extension method evaluates the elastic energy

stored under a certain loading condition with an initial crack length a and with a slightly

extended crack a+da, under the same loading conditions. Comparing both energies leads to a

G-value (energy release rate, in J/m2),

U (a + da ) U (a )

G , (7.27)

da

which can be compared to an experimentally determined critical energy release rate Gc. A

simulation of this type will only predict whether crack growth will occur under certain

loading conditions. Note, that two elastic analyses have to be performed for this prediction.

As an alternative, also the nodal forces around the crack tip can be evaluated. Sethuraman [6]

and Raju [7] provide a good reference for this Virtual Crack Closure method. In this case,

only one elastic analysis has to be performed, and the method automatically provides a

separation of the three basic crack growth modes. Analyses of cracks between dissimilar

materials (or dissimilar orientations) have to be interpreted with care, however.

Similar approaches are in use for fatigue analyses. For regular loading, the so-called Paris

Law can be applicable, relating the delamination growth rate da/dN [mm/cycle] to

da

cG n (7.28)

dN

with a as the crack length, N the number of load cycles, G as the maximum cyclic energy

release rate and c,n constants. An example of experimental results under mixed mode bending

is given below.

Composites Course University of Twente, Eng&Tech 8.10

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

Figure 7.5 Paris Law diagram for APC-2 from a Mixed Mode Bending test, log(c) = -3.59, n = 3.58

Knowing the appropriate critical energy relations, arbitrary load cases can be simulated. Up to

the year 2000, the applicability of these laws for random loading appears fairly limited,

however.

One step further is the use of special interface elements (with zero thickness). Generally, these

are inserted on a predetermined possible crack plane (e.g. the interface between two plies in a

laminate). Up to a certain load, the elements are very stiff, preventing separation of the solid

surfaces attached. After a maximum stress (the strength), the element opens, according to

some softening curve, meanwhile dissipating the critical energy release as found from fracture

mechanics experiments.

Composites Course University of Twente, Eng&Tech 8.11

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

Figure 7.6 Stress-displacement curve for the LUSAS interface elements for delamination growth simulations [8].

8.4 Conclusion

The Finite Element Method offers significantly more possibilities for stress analysis of

composite materials than simple laminate analysis. However, the complexity caused by

anisotropy and orientations, and the corresponding amount of input data required, imply that

such an analysis is by no means a trivial exercise and all results should be critically verified

before drawing conclusions.

8.5 References

[1] A.H. van den Boogaard, Aanvulling en handleiding bij Stijfheid en Sterkte 3 (115711), Universiteit

Twente, juli 1999.

[2] ANSYS Theory Reference, 000855 Eighth Ed. SAS IP, Inc

SOLID46 3-D Layered Structural Solid

[3] A.K. Noor, M. Malik, An assessment of five modelling approaches for thermo-mechanical stress

analyses of laminated composite panels, Computational Mechanics, vol.25, p.43-58 (2000).

[4] H. Thom, A review of the biaxial strength of fibre-reinforced plastics, Composites Part A, vol.29A,

p.869-886 (1998).

[5] Chang, F.K & Chang, K.Y., A progressive damage model for laminated composites containing stress

concentrations, Journal of Composite Materials, vol.21, p.834-855 (1987).

[6] Sethuraman, R. & Maiti, S.K., Finite element based computation of strain energy release rate by

modified crack closure integral, Engineering fracture mechanics vol.30/2, p.227-231 (1988).

[7] Raju, I.S., Calculation of strain-energy release rates with higher order and singular finite elements,

Engineering fracture mechanics, vol.28/3, p.251-274 (1987).

[8] FEA Ltd., LUSAS Modeller User Guide, v.13.0 (1998).

Composites Course University of Twente, Eng&Tech 8.12

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

A8.1 Approximation

A series expansion is a usual approximation of a function f(x):

f% ( x ) = ci H i ( x ) (A7.1)

i

with a set of constants {ci} and a set of basic functions {Hi(x)}, e.g. polynomials, sine series

etc.

In the special case of basic functions Ni(x) which are piecewise linear functions with a local

support, satisfying Ni(xj)=ij, the approximation is equivalent to linear interpolation:

xi +1 x x xi

f% ( x ) = f ( xi ) + f ( xi +1 )

xi +1 xi xi +1 xi xi x xi +1 (A7.2)

N i ( x ) f i + N i +1 ( x ) f i +1

In FE terms this approximation employs the shape functions Ni(x). As Ni(xj)=ij, this

representation automatically satisfies fi=f(xi).

f (x)

linear interpolation

continuous function

shape functions

1

Ni(x) Ni+1(x)

x

Figure A7.1 Piecewise linear approximation of a continuous function f(x).

Composites Course University of Twente, Eng&Tech 8.13

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

A8.2 Differentiation

Differentiating the piecewise linear approximation in space leads to a constant value per

element [xi,xi+1]:

d % f ( xi +1 ) f ( xi )

f ( x) =

dx xi +1 xi

xi x xi +1 (A7.3)

d Ni ( x) d N i +1 ( x )

= fi + f i +1

dx dx

or, in general, expressed in the derivatives of the basic functions,

d % d Hi ( x)

f ( x ) = ci (A7.4)

dx i dx

In words: if the derivatives of the shape functions (in the global coordinates) are known, the

(approximated) derivative of the function can be determined easily, with the same set of

constants {ci} as the function approximation itself.

N.B. A linear element with displacement degrees of freedom will have a constant strain per

element and hence also a constant stress.

A8.3 Integration

An integral over a full domain can be split into element contributions:

xn n xi

f ( x ) dx =

x0 i xi 1

f ( x ) dx (A7.5)

m

f ( x ) dx W j f ( x j ) (A7.6)

x j

using a set of weight factors {Wj}, such as Gauss quadrature for which

m

W p ( x ) = p ( x ) dx

j

j k j k (A7.7)

x

is satisfied exactly for polynomials pk(x) of the degree k=2m-1 (in words: with two integration

points, m=2, a third order polynomial p3(x) can be integrated exactly). The weight factors

{Wj} and the specific locations, the integration points {xj} follow from this requirement.

For our linear approximation one integration point is sufficient, which is located in the centre

of the element, leading to

( )

xi

f% ( x ) dx = ( xi xi 1 ) f% xi 1 (A7.8)

2

xi 1

Composites Course University of Twente, Eng&Tech 8.14

8. FEM for composites 2008 - Laurent Warnet & Remko Akkerman.

f (x)

integral of the

linear approximation

x

Figure A7.2 1-point Gauss quadrature of the linear approximation of a continuous function f(x).

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 9.1

R

M

OU

CO

9. Interlaminar stresses - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

9. Interlaminar stresses

9.1. Introduction

In the first lectures, the analysis of stresses and strains in a layer of a composite

laminate was performed assuming a state of plane stress (3= 4= 5=0). This

assumption is often valid, as long as the analysis is performed far away from

geometric discontinuities, like free-edges or holes. Close to these discontinuities, a

3D state of stress exists. Out-of-plane stresses, or interlaminar stresses, cause

delamination. Delamination is one of the ways composite laminates fail, where

different layers are separated.

It is not the intention to present here the complete 3D stress analysis, but to give

an insight into the mechanisms causing the interlaminar stresses. The interlaminar

stresses induced at the free edges of a laminate will be used as an illustration. This

will help critically considering the lay-up of a laminate.

Let us consider a [0/90]s laminated strip of finite width which is loaded in tension

as shown in fig. 8.1(a).

From a cross-section, first

Fx x

consider the free body

o

diagram of half of the top 0

layer (fig. 8.1(b)).

At the left side of the free (a)

body diagram, the CLT y

predicts the existence of a Fx z

tensile transverse stress y.

This stress is due to the Free-body diagram

mismatch in elasticity moduli ty

and Poissons ratio of the 0o 0o y

x tyz

and 90o layers. This stress (b) 90o tz

0o

cannot exist at the right side

of the free body diagram, as z

this side is free. Force

equilibrium in the y-direction ty

0o y

can only be satisfied with a x

(c) 90o

yz stress. Moment 0o tz

equilibrium provides a stress,

which equilibrate the moment z

of y on the left side of the

free-body diagram. It can be fig. 8.1: Free body diagrams at the interface of a 90 and a

shown [Pipes & Pagano 0 layer in a cross-ply laminated strip loaded under

1970] that the sign of z tension laminated beam in a three point bending

changes near the edge as loading situation.

shown in qualitatively in fig.

8.2. In fig. 8.1(c), a free-body

diagram including the right halves of both top 0o and 90o layers gives the situation

at the 0o/0o interface. In this case, only normal interlaminar stresses do occur. Note

also that z is zero far away from the edges. The region where the 3D stresses are

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 9.2

R

M

OU

CO

9. Interlaminar stresses - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

not any longer negligible is often called the boundary layer width. It was shown in

many publications that this width is in the order of magnitude of a laminate

thickness at most. A finite analysis result of the cross-ply laminated strip

considered in fig. 8.1. also shows the deformations involved at the edges.

y

y/b

0 1

fig. 8.2: Interlaminar normal stress near fig. 8. 3: Deformations at the edge of the cross-ply

the edge. strip considered in fig. 8.1

It has been reported in the literature that high values of z (tensile), xz (showing

up in angle-ply laminates) and yz cause free-edge delaminations. These premature

failure cannot be predicted using the Classical Lamination Theory. However,

taking interlaminar stresses into account can be important when designing a

laminate, especially when it comes to choose the stacking sequence.

Take as a simple example the [0/90]s cross-ply laminate subjected to a Nx tensile

force. It was shown in fig. 8.2 that at the 0o/0o interface, tensile normal

interlaminar stress z occurs, which can lead to an edge delamination mechanism.

Choosing for a [90/0]s layup does not affect the extensional stiffness properties

(the [A] matrix) but would change the sign of the z stresses at the free edges from

positive to negative, therefore reducing the tendency to delaminations. An other

classical example is the one given by Pagano and Pipes (1971), where the

behaviour of a [15/-15/45/-45]s is compared with that of a [15/45/-45/-15]s. The

qualitative results concerning the normal interlaminar stress in fig. 8.4 show that

the latter lay-up should be stronger than the former one. Similar observations can

be drawn for interlaminar shear stresses.

This latter example gives us a simple rule which can be used when choosing the

lay-up of a laminate.

Avoid when possible stacking up layers of similar orientations in order to

reduce interlaminar shear and normal stresses at the edges.

Choose for example [15/45/-45/-15]s instead of [15/-15/45/-45]s

It can be shown that large orientation difference between adjacent layers tends

to increase the level of interlaminar stresses.

SI T ES G

O

Composites Course 2008 University of Twente, Eng. &Tech. 9.3

R

M

OU

CO

9. Interlaminar stresses - Laurent Warnet & Remko Akkerman.

P

TE

UN I

EN

VE

R

W

SI T

TY OF

More generally, the lecture 4 dedicated to

the stiffness matrix [ABD] of the laminate z [15/45/-45/-15]s

makes us aware of the possible coupling [15/-15/45/-45]s

effect induced when choosing a certain

lay-up. An obvious unwritten rule is to

avoid non-symmetrical lay-up. Translated

mathematically in the stiffness coupling z

matrix [B], non-symmetry induces for

example curvatures when used at a

temperature different than the one at

which it was manufactured. One should

also be aware of other coupling effects,

which can lead to surprising effects, like fig. 8.4: Interlaminar normal stress for

bending-torsion coupling translated in D16. two lay-up having the same

extensional compliance matrix

- 10.5923.j.textile.20120104.02Uploaded byMaria Jose Biasotti
- FEMUploaded byron_dany
- Design Standard - Section 3-3 - Steel Storage Tanks -12!31!12 - FINALWEBPOST_201301090858057238Uploaded byராபர்ட் ஆன்றோ ரெனி
- MIL-R-7705BUploaded byArmando Castelló
- Introduction to Composite MaterialsUploaded byswordprinces
- WEST SYSTEM Launches G Flex EpoxyUploaded byMIlos Andric
- Aci 351.1 R-99 Grouting Bet Foundn & Bases for Support of Equipment & Machinery (PDF.io)Uploaded byPrapa Karan
- pro_fiberglass_GB_SCREEN.pdfUploaded byMohaab Nasr
- Endura Resin Chip SystemUploaded byvictorpachade
- Utility fuzzy multiobjective optimizationUploaded byNaresh Deshpande
- Novolac ScreedUploaded bymd_kureshi
- ch08.pdfUploaded byAnderson Gomez Castro
- 693B CF Orange Hardener Ratio Guide - SteelUploaded byCRISTHIAN PIERR PAUSIC BAZALAR
- Molecule DesignUploaded byUsama Osman
- Nathan a LaurelUploaded byn8laurel
- ROM compositeUploaded byIN Mauliza
- DS_066_Epilux_84_ClearUploaded byShodhan M Amin
- 3M-460DP Joining AdhesiveUploaded byAvdhesh Khaitan
- A Review On Mechanical Properties Of Natural Fiber Reinforced Hybrid CompositesUploaded byIRJET Journal
- Coating Material for Tendon for Pre Stressed ConcreteUploaded byinfant010
- When Good Floors Go BadUploaded bySergio Dominguez
- 2.IJMPERDOCT20172Uploaded byTJPRC Publications
- M one two.docxUploaded byadhurssh sriram
- Sigma Marine Coatings Manual_Part97Uploaded byTommy2020
- 1221.pdfUploaded byApurv
- Poly Meng SlidesUploaded byHuy Nguyen
- Sikafloor 2430Uploaded bythe pilot
- LaminatingUploaded byAnonymous JISy0Z
- SPE-143771-MSUploaded byVictor Villón
- Literature CitedUploaded byDe Leon Kai

- AVMM.Ch01Uploaded byHamed Memarian
- How to design and size a baghouseUploaded byWalter Mustafa
- Development About Composite Homogenization in Static and in Dynamic_application to UD Composite MaterialsUploaded bykhayat
- Comparative Review Study on Elastic Properties Modelling for UD Composite MaterialsUploaded bykhayat
- Toy TrainUploaded byDraveilois
- Piping stress analysisUploaded bysandygo
- SI_BlackBox_v2015_05_28Uploaded bykhayat
- GevinMcDaniel FRP CompositesUploaded bykhayat
- MECH370_S07_L1.pdfUploaded bykhayat
- learning simulinkUploaded bymarrakamasutras
- Excercises Dynamics and ControlUploaded byLeandro Silva
- Smart Industry IAM Overview PDF v02Uploaded bykhayat
- DIY Microscope - AllUploaded bykhayat
- Renewal NoticeUploaded bykhayat
- Basic Computational PlasticityUploaded byadnanhasanovic
- Joyful Deformation.pdfUploaded bykhayat
- Colorful BendingUploaded bykhayat
- Awesome ShearUploaded bykhayat
- Crude DraftUploaded bykhayat
- Wind Turbulence InvestigationUploaded bykhayat
- Woven Fabrics Paper v1 (1)Uploaded bykhayat
- Notes on Structural Analysis of Composite Shell StructuresUploaded bykhayat
- Notes on DynamicsUploaded bykhayat
- War of emotions.pdfUploaded bykhayat
- Physical Expressive Dynamics.pdfUploaded bykhayat

- Chemline Plastics Chemical Resistance GuideUploaded byzaza
- Download FileUploaded byChary
- Green chemistryUploaded byVijaykumar Marakatti
- Sintex Part 2Uploaded bypayalmalay
- Composites Basics_ Materials-Print VersionUploaded byUMA MAHESH
- Book 3Uploaded byapi-3826695
- Synthesis, Testing & Characterization of Epoxy-CNT-Bamboo Polymer Matrix Hybrid CompositeUploaded byIRJET Journal
- SILANE-GUIDE[1]Uploaded byian1000
- EPL 0001016 ArticleUploaded byVinay Kusumanchi
- Composite Definitions.pdfUploaded byAbdelaziz Abdo
- 011005_rsi2Uploaded byA Mahmood
- Manufacture of Synthetic RubberUploaded byHossamKandil
- Injection Molding Scrap ReductionUploaded bykgggani
- Structural Use of AdhesivesUploaded byEngr Hosam Eldin Ezzat
- Mechanical Characterization and Improvement of Weaveability for Glass/Polypropylene Commingled Hybrid YarnsUploaded bySEP-Publisher
- Fire fighting Clothes - Full.pdfUploaded byismailirfy
- Engineering Properties of Recycled Organic Aggregate for Controlled Low Streng MaterialsUploaded byzipshiraz
- Polymer Matrix Composites – Matrix Resins – Thermosetting resins, Thermoplastic resinsUploaded bygidlavinay
- Bmc Auto ElectUploaded byLâle Bulut
- ASTM D1599.pdfUploaded byChemical Engineer
- GRP PipeUploaded byAhmedHussainQureshi
- HDPE PipeUploaded byKhalid
- Green RocketUploaded byjohnny_cashed
- M&PII(Plastic-I)Uploaded byShachi Pandey
- Shenzhen Xinst Technology Co.,LtdUploaded byNicacio Lucena
- AdhesiveUploaded bySiva Chandhiran
- 10.5923.j.ajps.20140402.01Uploaded byBrian
- Polymerization of Acrylic MonomersUploaded byarsenalmct
- Polymer Technology MCQs PDFUploaded byZubair Malik
- D 0709 01 Laminados Termoplasticos ComponentesUploaded byhcmcrlho