3/28/2017

FRANC3D Training Workshop:

Part VII
April - 2017

Drs. Bruce Carter, Paul Wash

Wawrzynek, Tony Ingraffea, and
Omar Ibrahim

Fracture Analysis
Consultants, Inc.

Workshop Agenda

Part I: Introduction to Fracture Mechanics Analysis

Part II: Introduction to FRANC3D
Part III: Finite Element (FE) Model Import
Part IV: Crack Insertion
Part V: Static Crack Analysis & SIF Computation
Part VI: Crack Growth
Part VII: Multiple/Variable DOF Approach to Fatigue Life
Part VIII: SIF History & Fatigue Life
Part IX: Session Log, Command Line and Python Interface
Part X: Miscellaneous

2
Part I

1
 3/28/2017

Example Simulation: 22 Steps of Growth

Crack mesh at step 22

Portion of model
contains crack
Minidisk
crack fronts

How does one make a fatigue life

prediction based on this simulation?

3
Part V

Example Simulation

4
Part V

2
 3/28/2017

Reducing SIF (K) Data to a Single DOF

Path through the

crack fronts

Crack growth simulation yields

SIF distributions along each
crack front for all crack steps.

Single degree of freedom

(DOF) K vs a curve
5
Part V

Conventional Lifing with FRANC3D Results

A path is defined through the crack fronts that effectively
reduces the full 3D results to a single DOF model.
log da/dN

da
N
f (DK , R, DKth ,)

The single DOF SIF (ranges) are integrated

log DK with a material crack growth rate model to
determine cycles (N). 6
Part V

3
 3/28/2017

Determining a Single DOF Path(s)

?
A path
through the Or some
mid-points! other path?
?

How should one pick the single DOF path, and

how does this affect the predicted fatigue life?

7
Part V

Life Predictions For Different Paths

3.5% Different paths provide

different number of
cycles; could extract an
average or mean and
standard deviation.

Large Crack Steps 2.7%

Example:
corner crack 16

in a plate Small Crack Steps

under
tension 0.25 8
2
Part V

4
 3/28/2017

Single DOF Life Predictions

The least accurate SIFs

are computed on the Comparison of life predictions using SIFs
specimen surface; so we evaluated along six different paths shows
choose a path where SIFs a significant variation.
are more accurate.

path

DK along path

9
Part V

Problems With the Single DOF Approach

Defining an appropriate
single path becomes difficult
for a more complex crack
geometry ?? ??
Defining a single path requires subjective engineering
judgment leading to non-objective life predictions
Difficult to automate path selection
Uses only a small amount of the available SIF information
Can not easily correlate cycles with the observed or
measured crack lengths on the specimen surface
For many real world models, a single path cannot intersect all
crack fronts so multiple paths need to be evaluated.
10
Part V

5
 3/28/2017

Problems With the Single DOF Approach

11
Part V

Problems With the Single DOF Approach

12
Part V

6
 3/28/2017

Multiple/Variable DOF Approach

nodei 1,k 1 For all nodes on crack front i
DKi 1,k 1 1. Project perpendicular to the
crack-front to find the
DKi 1, j intersection with the next crack
front i+1

z y nodei 1,k 2. Interpolate to find DK at the

projected intersection
x DKi 1,k
DKi , j 3. Assume a linear variation in the
nodei, j
DKs going from step i to step i+1
stepi 1 and integrate to find the cycles
(Ni,j)
stepi 4. Average the computed cycles for
all crack front nodes to obtain one
Da value for cycles required to grow

da
Ni , j from crack front i to crack
n front
f (DK , R,)

1
0 i+1 Nii 1 Ni, j
n
where DK DKi, j
a
Da

DKi 1, j DKi , j j 1

13
Part V

Multiple/Variable DOF Approach

Compute the number of cycles

for all nodes along a crack front
Compute the average
Store average cycles versus step
of crack growth

node j

crack front i+1 crack front i

14
Part V

7
 3/28/2017

Multiple/Variable DOF Approach

Life curves can be generated for different paths, and the number
of cycles will be consistent.
Can even plot cycles versus
crack length for a path on the
surface where computed SIFs
are not usually accurate.

15
Part V

Compare to Single DOF Predictions

2.7%

multiple DOF
based prediction

16
Part V

8
 3/28/2017

Benefits of Multiple/Variable DOF Approach

Ignores cases where a projected point cannot be found
stepi 1
stepi 1

stepi
stepi

Generates a unique cycle count for each crack front.

?? ?? Part V
17

Benefits of Multiple/Variable DOF Approach

Avoids SIFs computed at surface, which are unreliable because:
- plane strain conditions do not hold locally, and
- crack front elements can be highly distorted
distorted elements used in
the crack-front integrals
Cycles for each crack step include
information from the entire crack
front.

Generate accurate predictions of

cycles versus surface breaking crack front
crack length despite the fact that
we cannot compute accurate
stress intensity factors at surface
breaking points.

18
Part V

9
 3/28/2017

Fatigue Crack Growth Rate Model Library

Separate growth rate and R-ratio models (except NASGRO eqn.)
All model can be specified as temperature dependent
Growth rate models:
Paris Paris
Sigmoidal
Bi-Linear Paris
Bi-Linear Paris
Sigmoidal
Hyperbolic Sine
Tabular
NASGRO

Hyperbolic Sine Tabular

R-ratio models:
NASGRO
None
Walker
Closure
Tabular

19
Part V

Fatigue Crack Growth Rate Model

Properties can be specified for various temperatures.

Various options are available for interpolating among temperatures

Units for material properties specified separately from the analysis

model units, conversions are done automatically.
20
Part V

10
 3/28/2017

Crack Growth Stopping Criteria

Formerly the user had to specify a fixed number of FRANC3D

crack growth steps.
Now automatic crack growth stops when the first of the
following is satisfied:
N N max K max Kc DK DKthreshold step max step

We would like to add a stopping criterion related to the

maximum crack size, but we have not yet determined the best
way to do this.
The new criteria require that FRANC3D maintain a running
estimate of the applied cycles.

21
Part V

11