HIGH CYCLE FATIGUE DELAMINATION MEASUREMENT AND GROWTH PREDICTION

1
HIGH CYCLE FATIGUE DELAMINATION MEASUREMENT AND
GROWTH PREDICTION
Adam Pickard and Stephen Hallett
www.bris.ac.uk/composites
2
Overview

• Objective Investigation of high cycle fatigue
  damage through vibration excitation at large
  amplitudes.
• Sample design and artificial delamination
  production.
• Experimental testing method and results.
• Delamination growth prediction through FEM.

3
Sample Design
60mm

• Small rectangular samples (300mm x 60mm).
• Simple symmetric cross-ply lay-up (0/90)30s.
• IM7/8552 material.
• Full width artificial delamination.

150mm
300mm
4
FEM Predicted behaviour

• Modelled in MSC Nastran with shell elements,
  delamination used double layer of shells with
  offsets
• Well separated modes, little incidence of
  harmonics mixing.
• Frequency drop with damage indicates odd bending
  modes are most sensitive.
• 2nd mode for driving damage, 3rd mode for damage
  detection.

5
Delamination Production

• Delamination in initial samples produced with two
  layers of 15µm thick ETFE release film.
• Great difficulty with propagating delamination
  Not true delamination?
• Force open delamination with feeler gauge
  ruffle release film.
• Attempted to use Steel Shim, 25µm thick, in place
  of release film.
• Shim debonded easily but was not removable.
• Improved initial condition allowed reproducible
  delamination behaviour and easier propagation.

6
Delamination Comparison

• Both specimens produce good quality interface.
• Using more shim metal to separate the shim insert
  produced an expanded delamination region.
• Definite separation, although size of delaminated
  region is near 2/3 of a ply.
• No evidence of fibre nesting with release film
  inserts.

Shim
ETFE
7
Set-up

• The sample is clamped at the base and attached
  directly to the shaker armature.
• A Polytech Scanning Laser Doppler Vibrometer
  (SLDV) measures the specimens velocity.
• In house developed control system identifies and
  tracks the resonant frequency.
• Automated periodic capture of frequency response
  and thermal imagery.

CFRP Specimen
E.M. Shaker
Clamp
8
Vibration Characteristics

• Initial vibration characteristics were
  investigated and showed a softening effect with
  increasing amplitude at the 2nd bending mode.
• Conversely, a stiffening effect was observed at
  the 3rd bending mode.
• ETFE insert samples were fatigued for 34 x 106
  cycles with only 3 measurement occurences.
• Shim insert samples were fatigued for 40 x 106
  cycles with measurements every 5 x 105 cycles.

9
Vibration Characteristics
Shim Insert
ETFE Insert
MODE 2
10
Vibration Characteristics
Shim Insert
ETFE Insert
MODE 3
11
Endurance Measurements - ETFE

• Little variation was observed in the damping
  factor as the delamination grew.
• Frequency drops indicate material softening.
• Amplitude drops indicate material hardening.
• Issue with nodal position?

12
Endurance Measurements - Shim

• Different variations observed at each mode.
• Mode 2 demonstrated clear increases in the
  amplitude of response at each voltage.
• Mode 3 showed clear and consistent drops in
  frequency although small.

13
Endurance Measurements - ETFE

• C-scan clearly shows delamination growth.
• Thermal measurements potentially highlight
  delaminated region.
• Damage shown here was after 34x106 cycles.

14
Endurance Measurements - Shim

• C-scan suggests separation and minimal growth.
• Thermal measurements again highlight delaminated
  region.

Before
After
15
Endurance Measurements

• A line scan for sample displacement was measured
  to identify possible movement of the vibration
  nodal position.
• Shift of the node could cause erroneous
  displacement measurements.
• Curvatures will be calculated to potentially
  identify damaged region.

16
Virtual Crack Closure Technique (VCCT)

• VCCT is commonly used to predict the growth of
  damage in composite samples.
• Sample is constructed with 2 surfaces of shell
  elements, nodes are connected with stiff springs.
• Springs constants are calculated from resin
  properties.
• The Strain Energy Release Rate can be calculated
  using the forces in the springs and the nodal
  displacements.

Uz2
Fz
Uz1
Measure displacement between nodes
Measure force between nodes at delamination front.
17
Virtual Crack Closure Technique (VCCT)

• The calculated SERR can be used along with a
  Fatigue Delamination Growth Law.
• This allows the calculation of the number of
  cycles to failure.
• Mode ratio is predominantly mode II (95 top
  edge, 90 bottom edge).
• No R-1 data for IM7/8552.
• Initial analyses used GIImax (R0.1) GIImax
  (R-1).
• Tanaka and Tanaka found a 3 order of magnitude
  difference between R-1 and R0.1 for Toray
  T800H/3631 material.

Tanaka, K. and Tanaka, H., "Stress-Ratio Effect
on Mode II Propagation of Interlaminar Fatigue
Cracks in Graphite/Epoxy Composites," ASTM STP
1285, 1997, pp. 126-142.
18
VCCT Matlab Code

• Allows user to build model with delaminated
  region.
• Creates Nastran input files and interprets
  results files.
• Calculates SERR and determines which node reaches
  life limit first.
• This node is released whilst other nodes store
  percentage of life reached.
• Program iterates to identify next nodes to
  release.
• Hence, the code predicts delamination growth.

19
VCCT
Converge on frequency value
Run Modal Analysis
Run Frequency Analysis
Build Model
Identify spring Forces and nodal Displacements
Calculate Gmax
Calculate da/dN and thus dN
Calculate proportion of life reached at every node
Release critical node/nodes
Re-run from Modal Analysis
20
VCCT
21
Comparison

• The values were predicted with R0.1 data. Using
  the 3 orders of magnitude variation, these values
  predict a detectable delamination growth of 3mm
  within 10 x 106 cycles.
• From experimental results, damage has become
  detectable after 25 x 106 cycles.
• Delamination growth pattern appears accurate.
• Issues remain with true delamination condition.
• R-1 data required to improve routine accuracy.

22
Conclusions

• Excitation to high amplitudes extremely
  difficult.
• Response is non-linear with high amplitudes and
  varies with excitation mode.
• Artificial delamination production technique
  influences results.
• Delamination growth only plausible for
  experimentation time at high amplitude with high
  cycles.
• VCCT can be used to predict delamination growth
  in dynamic testing.
• IM7/8552 data for R-1 required to validate
  method.
• Further testing required to gather more data.

23
Acknowledgement

• This work has been supported by Rolls-Royce plc.