FatigueTesting(1)

1

Mechanical characterization

Fatigue testing

28/11/2013 1CompositesIves De Baere and Joris Degrieck – 2013‐2014

What is fatigue?

• A periodically changing load (or deformation) on a structure is a fatigue load

D t titi th t t• Due to numerous repetitions, the structure will fail at (significantly) lower stresses or deformations than would be the case under the (quasi‐)static loading

understanding of this behaviour is important


2

What is fatigue?

Realistic loads (see example) are quite random, although in most cases (rotating structures) one or some frequencies are more important than others

Normal stress in an airplane wing during ground‐air‐ground manouvre


Fatigue tests usually consist of imposing a sinusoidal load (or deformation) on the structure and observing its behaviour

Sinusoidal loading

• Realistic loading is often not possible with a servo‐hydraulic test machine

• Realistic loading is even more expensive and time consuming g p gthan sinusoidal ones

• Often, the realistic loading conditions are not know.


3

Sinusoidal loading

Some parameters

Mean stress

Stress amplitude

Stress ratio

max min

max min

min

max

2

2

mean

amp

R


Parameters for fatigue

• Sample– Material– Lay‐upy p– Dimensions and geometry

• Loading– Type of loading (uni‐axial, bi‐axial, bending, …)– Orientation of the loading with respect to the fibre lay‐up– Frequentie of the loading– Stresses and stress ratio


• Environment– Temperature– humidity

4

Sinusoidal block loading – Miner’s rule

Block loading:To simulate realistic loading, tests are conducted, consisting of different periods each with its own mean stressperiods, each with its own mean stress and stress amplitude.

To predict damage, usually Miner’s rule is applied:During a constant amplitude fatigue loading, damage D increases linearly with the lifetime, going from 0 at start to 1 at


failure:

For different loading‐blocks:

f

ND

N

1 21 2

,1 ,2 ,

... ... ii

f f f i

NN ND D D D

N N N

Sinusoidal block loading – Miner’s rule

• The rule is commonly applied, but the predicted lifetime often exceeds the actual lifetime (unsafe!!!)

• Miner assumes linear damage growth which is (quite• Miner assumes linear damage growth, which is (quite often) not the case

• Miner does not take the order of the blocks into account, but for fibre reinforced composites, the fatigue life for low‐high amplitude significantly differs from high‐low amplitude

Mi i t ti b t d


• Miner assumes no interaction between damage corresponding to different load levels. Damage growth, however, is definitely influenced by the damage, already present

5

Representation – S‐N curves

• Typical representation of fatigue results

• For a given load case (e.g. maximum stress and R‐ratio), the b f l f il i l d


number of cycles to failure is plotted.

• Due to inherent scatter (which is even larger for composites compared to metals), a lot of experiments are necessary for 1 S‐N curve (meaning 1 load case)

S‐N curves: not suited for composites

• S‐N curve only valid for one specific combination of sample, loading case, environmental conditionsenvironmental conditions.– Results from literature are often not comparable

• Damage mechanics in composites:– Each layer has its own stress state, depending on its on mechanical properties (and orientation)


– Damage lower mechanical properties stress redistribution damaged layer is subjected to a lower loading and other layers carry the load damage will first grow in other layers …

6

Why composites for fatigue?


Different damage behaviour in composites

• Stress concentrations present between fibre and matrix, resulting in micro‐cracks

• Other anomalies present (voids foreign particlesOther anomalies present (voids, foreign particles, thermal cracks, …)At the beginning of the fatigue, damage is

already present• However, micro cracks will lower the local stress concentrations, so damage will grow very gradually


gradually• Metals: most of the fatigue life is spent on the initiation of cracks. Crack growth till final failure is only a small part.

7


• Due to the local cracks, the mechanical properties deteriorate

• Usually, this results in a decrease in stiffness properties and an increase in permanent deformationincrease in permanent deformation




• As such, fatigue life for composites will be a function of the desired application for example:– maintaining stiffness for a composite spring– maintaining strength– Limitation in permanent deformation (geometrical stability)– Avoiding surface cracks (corrosion related)

8

Influence of some important parameters‐specimen

• Material– Most fibres have a good

(glass) to very good (carbon,(glass) to very good (carbon, aramid) fatigue resistance.

– UD carbon or UD aramid have an extra increase in fatigue resistance in tension compared to glass, since they carry more load due to their higher stiffness


– The fibre volume fraction


• Material– Matrices with higher failure strain

(typically thermoplastics) tend to have fewer crack initiationhave fewer crack initiation, resulting in longer fatigue life

– Crack density= number of cracks per surface area

For PEEK (thermoplastic)

For EPOXY

Cycles x 105


For EPOXY

(usually counted by hand under a microscope quite time‐consuming and inherent human error/scatter

Cycles x 104

9


• Geometry– With increasing dimensions, the

change of having area’s withchange of having area s with defects increases, so a lower fatigue life is usually the result

– Too small width, however, results in the fact that cracks grow to fast from one side to the other

– Dumbbell shape for CETEX: only


p ya small area (middle of the curvature) sees the highest stress, so less chances of defects, while width is still sufficient

Influence of some important parameters‐environment

Temperature

similar to the static properties, also the f i b h i ifatigue behaviour is influenced

Humidity


HumidityMoisture absorption usually results in softening and swelling of the matrix and a degradation of the fibre‐matrix interface

10

Influence of some important parameters‐loading

• Loading direction– Direction of loading

compared to the orthotropiccompared to the orthotropic directions determines whether the behaviour is fibre‐dominated or matrix‐dominated.

– Small deviations from the (intended) fibre directions result in significant reduction


of properties

– Chopped fibre composites mainly show a matrix dominated behaviour


• Loading type:tension versus compression

Types

min

max

R

– Types

Tension‐tension:

Tension‐compression:

Compression‐compression:

– Due to lower tensile strength of the matrix (compared to

its compressive strength), fatigue life with tension loading

max

max

max

0 and 0 1

0 and R 0

0 and 1

R

R


is mostly lower than when only compression is applied

– Due to tension, quite often ‘debris’ is formed within the matrix cracks, causing increase in permanent deformation, increase in temperature and increase in degradation

11


• Loading type:tension versus compression– Be careful when interpreting fatigue life: tension‐tension

fatigue may yield high fatigue life, even when the specimen is completely damaged. Residual compressive strength, however, will be negligible.

– Most real life components are loaded both in tension and compression



• Type of loading– Due to the orthotropic

nature, the type of loadingnature, the type of loading (uni‐axial, bending, shear, …) will have an important influence.

– E.g. a bending test will result in a combination of normal stresses and shear stresses (both inter‐ and intralaminar).


Therefore, there is no such thing as ‘bending fatigue life’

12


• Loading frequency– General preference: keep the loading frequency as high as possible,

to shorten the duration of the testto shorten the duration of the test.

– However, the fatigue behaviour of (especially, but not limited to) matrix‐dominated composites is highly frequency dependent, due to the visco‐elasticity of the matrix.

– Higher frequencies will result in heating of the specimen (especially with shear loads)

– Lower frequencies give existing cracks more time to grow within one cycle reducing fatigue life


cycle, reducing fatigue life

Influence of some important parameters‐loadingHeating of the composite due to shear load: three rail shear test on CETEX

T of PPSTG of PPS is 90°C!


13


Heating of the composite due to shear load, [±45°]4s (CETEX)

• Temperature exceeds TG allowing the fibresTG, allowing the fibres to align themselves more along the loading‐direction.

• Thus, a new loading regime with less shear is present.

• Temperature drops


Temperature drops and fatigue continues under different conditions

Ultra high cycle fatigue

• Low cycle fatigue – up to 106 cycles– 1Hz 11.6 days

• High cycle fatigue – 106 tot 108 cycles• High cycle fatigue – 10 tot 10 cycles– 1Hz 1157 days !!!

• Very (or ultra) high cycle fatigue – more than 108

cycles– Application in aircraft, automobile, railway…– Gas turbine disk : 1010 cycles


– High speed trains: 109 cycles

• No longer achievable with standard tensile machines

14


• Usually done ultrasonic (f=20kHz) to limit test time

Number of cycles Ulstrasonic (20kHz) Conventional (100Hz)

107 9 minutes 1 day

109 14 hours 4 months

1010 6 days 3 years



• For composites: specimen is usually larger (a number of unit cells necessary to have an accurate representation

• Usually done with a shaker, in bending

Main problems

• Highest load occur at the clamping clampingfailure due to stressconcentration

• Mechanical properties (E


• Mechanical properties (Eii, Gij, …) change significantly compared to static values

• Interpretation/modelling/prediction is difficult

15

Thermal fatigue

• Instead of applying cyclic loading (forces or deformationg), the temperature has a cyclictemperature has a cyclic evolution.

• Very time‐consuming• Mostly done for Ceramic

matrix composites and metal‐matrix composites

• Is also important for fibre reinforced


fibre reinforced polymers, due to the degradation of the polymer.

Lifetime expectancy for real structures


16

Full scale testing: Windturbine blade

• Before a new design of Wind turbine blade can be sold, it needs to be certified prove that it will survive

• Data on certification tests: case dependentBesides static failure loads, a fatigue life of 108

cycles is desired (lifetime about 20 years).• To shorten test time, stresses are usually higher, but given the low test frequency, certification may take up at least 1 year


may take up at least 1 year• Two test methods available

– Using hydraulic actuators– Using the eigenfrequency of the blade


Using hydraulic actuators


http://www.youtube.com/watch?v=e8ePrX1GKck

17


Using the eigenfrequency of the blade


http://www.youtube.com/watch?v=XdsC73Q‐E2s after 1min

Full scale fatigue test on Boeing 747


more details: http://www.youtube.com/watch?v=TH9k9fWaFrs

18

Full scale testing on bicycle framesFatigue test simulating pedal forces: go or no‐go test according to the standard


FatigueTesting(1)

Documents

t ti b t d28112013

theb f

linear damage growth

mean stressperiods