Life Prediction Techniques for Composite Materials

1
Life Prediction Techniques for Composite Materials

• Scott Case, Joseph South, Jeremy Duthoit, and Ken
  Reifsnider
• Materials Response Group
• Department of Engineering Science and Mechanics
• Blacksburg, VA 24061

2
Necessity for Life Prediction (or Why go to all
this trouble?)

• To certify structures for service
• To reduce the need for experimental testing
• To design new components or structures (what if
  studies)
• To warranty existing or new products

3
Outline

• Philosophy of life prediction based upon residual
  strength
• Application to composite materials for fiber
  controlled failures (for cases in which two
  mechanisms act concurrently)
• Extension to matrix controlled failures

4
Elements of the Problem

• Physical behavior damage and failure modes
• Modeling discrete events, multiple processes
• Measurements independent observables as inputs
  to the models
• Life prediction extensions, generalizations,
  accelerations of laboratory experience

5
Key Features

• Strength (and, as a result, life) of composite
  material systems is controlled by
• Statistical accumulation of flaws subsequent
  interaction of flaws
• Large changes in material states and stress
  states
• Highly local level behavior (often on the
  fiber/matrix scale) controls failure

6
Remaining Strength Predictions

• Track remaining strength during the fatigue
  process
• Define a scalar failure function based upon
  tensor strength and stresses use this failure
  function for calculations
• May include the effects of changing loading
  conditions
• May be directly validated experimentally, unlike
  Miners rule

Residual Strength
Sult
Stress or Strength
Life Curve
N1
Cycles
7
Remaining Strength Predictions

• Track remaining strength during the fatigue
  process
• Define a scalar failure function based upon
  tensor strength and stresses use this failure
  function for calculations
• May include the effects of changing loading
  conditions
• May be directly validated experimentally, unlike
  Miners rule

Residual Strength
Sult
Stress or Strength
Life Curve
N2
Cycles
8
Remaining Strength Predictions

• Track remaining strength during the fatigue
  process
• Define a scalar failure function based upon
  tensor strength and stresses use this failure
  function for calculations
• May include the effects of changing loading
  conditions
• May be directly validated experimentally, unlike
  Miners rule

Residual Strength
Sult
Stress or Strength
Life Curve
N1
N2
Cycles
9
Remaining Strength Predictions

• Track remaining strength during the fatigue
  process
• Define a scalar failure function based upon
  tensor strength and stresses use this failure
  function for calculations
• May include the effects of changing loading
  conditions
• May be directly validated experimentally, unlike
  Miners rule

Residual Strength
Sult
Stress or Strength
Miners rule
N1
N2
Cycles
10
Mathematical Representation

• Some possible choices for failure criteria
• Maximum stress/strain
• Tsai-Hill/Tsai-Wu
• Micromechanics

• Define a failure criterion, Fa, and a remaining
  strength in terms of that failure criterion, Fr
• Define a generalized time (for example n/N or
  t/trupture)
• Postuate change in remaining strength over the
  interval
• Fa is constant over
  
• For the special case in which is equal to
  zero

11
Mathematical Representation

• Calculate change in remaining strength over the
  interval
• Calculate number of cycles required for failure

High-low n2 gt Miners rule Low-high n2 lt
Miners rule
12
Example Broutman-Sahu (1972)
Fatigue (R0.05) of E-glass/epoxy (a study of
sequence effects)
Fa
One-Stress Level Results
13
Example Broutman-Sahu (1972)
Two-Stress Level Results
14
Example Broutman-Sahu (1972)
Two-Stress Level Results
15
Life Prediction MRLife

• Modeling combined, interactive effects

16
Approach for Variable Loading with Rupture and
Fatigue Acting

• Divide each step of loading into time increments
• Treat each increment as a stress rupture problem
  (constant applied stress and temperature)
• Reduce residual strength due to time dependent
  damage accumulation
• Refine number of intervals until residual
  strength converges
• Input next load level
• Check for load reversal. If load reversal,
  increment by 1/2 cycle and reduce residual
  strength due to fatigue damage accumulation

17
Prediction of Combined Rupture and Fatigue on
Coupon Level

• Characterize fatigue effect with room temperature
  fatigue tests
• Characterize elevated temperature effect with
  tensile rupture tests at temperature
• Combine effects using analysis and compare to
  experimental results

18
Characterize Rupture Effect

• Tensile rupture tests at 90 C
• Fit data with Kachanov-type curve

19
Characterize Fatigue Effect

• Fatigue Tests at 25C
• R 0.1
• f 10 Hz
• Fit data with S-N Curve

20
Predict Elevated Temperature Fatigue Behavior

• Fatigue behavior accurately predicted at 90C, R
  0.1
• Validates the life prediction technique for this
  case

21
Predict Failure Under Combined Conditions for
Matrix-Controlled Failures

• Characterize matrix rupture and fatigue behavior
  as a function of temperature and stress level
• Time/cycles to failure
• Residual strength versus time/cycles
• Define a failure criterion for the composite
  behavior (non-linear for elastomers)
• Apply the residual strength approach to life
  prediction based upon failure criterion

22
Elastomer Stress Rupture Testing
23
Elastomer Residual Strength Tests
24
Elastomer Fatigue Data
Rupture Data
Fatigue Data?
25
Prediction of Off-Axis Failure
Need Mechanics-based means to predicting
failure in the matrix

• Phenomenological approaches (such as Tsai-Wu)
• Micromechanics based approaches (such as Aboudi,
  Carman and Averill, Subramanian)

For elastomers, some type of non-linear criterion
is required
Assume
26
Does it Work?

• Begin by validating with static strength for
  linear-elastic materials

27
Does it Work?

• Extend to steel-reinforced elastomers (in
  progress)

28
Summary

• Developed a method for predicting the lifetime of
  composites in which multiple damage mechanisms
  act
• Compared predictions based upon residual strength
  to experimental data for fiber-controlled failure
• Discussed extending approach to matrix-controlled
  failure using micromechanics