*

~~~~

-'

*V----

W.

AFWAL-TR-83-3027

r/

('4

DURABILITY DESIGN HANDBOOK:

"GUIDELINES FOR THE ANALYSIS AND
DESIGN OF DURABLE AIRCRAFT STRUCTURES

"USAF

S. D. Manning
General Dynamics Corporation
Fort Worth Division
P.O. Box 748
Fort Worth, Texas 76101
and

J. N. Yang
The George Washington University
Washington, D. C. 20052
q1

DTIC

"January 1984

JUN 2 5 1984

:"
FINAL REPORT JULY 1981 - JANUARY 1984
Approved for public release; distribution unlimited

"1..
"LLJ AIR FORCE FLIGHT DYNAMICS LABORATORY

..

AIR FORCE WRIGHT AERONAUTICAL LABORATORIES

LZ

AIR FORCE SYSTEMS COMMAND

WRIGHT-PATTERSON AIR FORCE BASE, OHIO 45433

"I
,-

- -

. - - * -

--

*-**

* *

* *

*.r%

* .

II
NOTICE
data are used for any purtlost:
When Government drawings, specifications, or othor
Srelated Government procuremert oporation..
incurs no responsibilityi nor any obligat on
eby
ernment ther
the United States Gov
formulated, furnI-shed, or in
whatsoever; and the fact that the government may have other data
is , not to be reor
any way supplied the said drawings, specifications,
or any
holder
the
garded by implication or otherwise as in aay manner licensing
to manufacture
other person or corporation, or conveying any righ6s or permission
thereto.
related
be
way
any
in
n~ay
that
invention
any patented
use, or sell
(ASD/PA) and is
This report has been reviewed by the Office of Public Affairs
At NTIS, it will
(NTIS).
.rervice
releasable to the National Technical Information
nacions.
gn
be available to the general public, including fol

This technical report has been reviewed and is

ams L.

approved for publication.

dRud

7;a:

Project Engineer

esDavey

,...

L. Smith
Structural Integrity Branch

Structu re s & Dy

nam
ics Div ision

FOR THE COMMANDER:

Ralph L. Kuster, Jr., Colonel, USAF

Chief, Structures & Dynamics Division

.',

or
list,
"If wour address has changed, i.: you wish to be removed from our mailing
notify AF%4AL//1IB),
if the addressee is no longer employed by your organization please
list".
mailing
W-PAFB, OH 45433 to help us maintain a current
is requirad b security
Copies of this report should not be returned unless return
document.
specific
a
on
contractual obligations, or notice
,
considerations,

N.0

Unclassifiqd
SECURITY CLASSIFICATION OF THIS PACE ('lt7en Data.Entered)

REAL) INSTRUCTIONS

REPORT DOCUMENTATION PAGE

2. GOVT ACCESSION NO.1

REFORT NUMBER

1.

2/:2.

A-WAL-TR-83-3027/AY
4.

AUTHOR(,)

Manning. and J.

CONTRACT OR GRANT NUMBER(a)

PROGRAM ELEMEN-, PROJECT, TASK

AREA & WORK UNIT NUMBERS

24010118

76101

Flight Dynamics Laboratory

Januaryj194

(FIBE)

Wright-Patterson Air Force Base,

.'__

_-_,.--__

REPORT DATE

12.

Air Force Wright Aeronautical Laboratories

OH

13.

NUMBER OF PAGES

213

45433

MON:TORING AGENCY NAME & ADDRESS(If different from C-rntrollnig Office)

14.

8.

P.E. 62201F

CONTROLLING OFFICE NAME AND ADDRESS

II.

PERFORMING O1G. REPORT NUMBER

10.

General Dynamics
Fort Worth Division
Texas

6.

F33615-77-C-3123

N; Yang

PERFORMING ORGANIZATION NAME AND ADDRESS

Fort Worth,

TYPE OF REPO'RT & PERIOD COVERED

July 1981 to Ja-nary 1984

Aircraft Structures

9.

__/_____._"_..

Final Report for period

Guidelines

fir the Analysis and Design of Durable

S. D.

RECIPIENT'S CATALOG NUMBER

3.

5,

TITLE (and Subtitle)

USAF Durability Design Handbook:

7.

BEFORE COMPLETING FORM

15.

SECURITY CLASS. (of thls report)

Unclassified
ISa.

16.

DECLA.41FICATION/DOWNGRADING
SCHEDULE

DISTRIBUTION STATEMENT (of thie Report,

Approved

for public release;

distribution

unlimited.

17,

DISTRIBUTION STATEMENT (of the abetract entered In Block 20, if different from Report)

IS.

SUPPLEMENTARY NOTES

The associate
University,

investigator

Washington,

for ':his report was J.

D.C.

N.

Yang of George Washington

.N

2,0052

KEY WORDS (Cwitlnue on reverse side itf neceeary and identify by block number)

19.

deterministic
crack size distribution,
Durability, fatigue, economic. life,
flaw size
equivalent initial
crack growth, time-tc-crack-initiation (TTCI),
fracture mechanics,
fatigue quality (IFQ), probabilistic
(EIFS), initial
probability of crack exceedance,
N<

extent of damage,

small crack sizes,

Weibull

0 distribution.
ABSTR
S

"This

CT (Continue on reverse aide If neceseary and Identity by block number)

is

the first

edition of the Durability Design Handbook.

U.S. Air Force

provide durability

the essential
the handbook are: (1) summarize and interpret
Durability Design requirements for metallic airframes, (2)
analysis criteria
state-of-the-art

and durability critical

for economic life
durability analysis 'oncepts and methods

the extent of damage for a population of details

(Continuted)

DD

-IJAN

1473

EDITION OF 1 NOV6 6 IS OBSOLETE

(e.g.,

Objectives of

parts,

(3)

provide

for quantifying

fastener holes,

"

Unclassified
S.CURITY CLASSIFICATION OF THIS PAGE (When Data Enttred)

""

rrV
Unclaesified
SECURITY CLASSIFICATION OF THISI PAGE(Whan Data Entered)

20.

Abstract (CQntinued)

-- Afillets, cutouts, lugs, etc.) as a function of time, (4) provide guidelines

and design data for implementing the durability methodology and for assisting
%
contractor and USAF personnel in complying with the intent of the durability
This document, loosely called a
specifications for metallic airframes.
..
provides guidelines, concepts, analytical tools and the framework
% ."Handbook",
for incorporating future durability methodology advancements and design data.,,_
The durability analysis methodology, based on'a probabi.istic fracture
fatigue quality, fatigue crack growth
mechanics approach, accounts for initial
accumulation in a population of structural details, load s !ctra and structural

"properties. A statistical

flaw sizes is
distribution of equivalent initial
fatigue quality of structural details (e.g.,
used to represent the initial
flaw
The equivalent initial
cutouts, lugs, etc.).
fastener holes, fillets,
to
conditions,
design
size distribution is grown forward, using the applicable
a selected service time using a deterministic crack growth relationship,
The durability analysis methodology
Procedures are described and illustrated.
has been demonstrated for quantifying the extent of damage due to relatively
Further research
"small cracks in clearance-fit fastener holes (e.g.,!_ 0.10").

,'..',of

is needed to extend the methodology to larger crack sizes and to vertfy the
methodology for other details, such as, fillets,
cutouts and lugs.
The effects
fretting,

faying surface sealant,

ence-fit fasteners,
L_...

An evaluation is
--

etc., on the initial

fastener clamp-up,

environment,

interfer-

fatigue quality need to be investigated

made of the accuracy of the durplility analysis by correlating

analytical predic,:ions with test

data for a fighuer full-.scale

test

article

and complex splice specimens subjected to a bomber load spectrum.

Unclassified
5'CURITY CLASSIFICATIOt4 OF -- I- PAGE(WhOFn Date Entered)

Ni

i.
v.

." ' " - .

"'-

-."''
.'

'."

'

"

.v'. ;

';*'

-'

V'

.U71

.- -r ,-' -. C-i 7-

'V

7'

FOREWORD

This

handbook

was

prepared

by General Dynamics,

Fort

Worth Division,

and by George

Phase

the "Durability Methods Development" program

III

of

(Air Force Contract

Wright

Washington

F33615-77-C-3123)

Aer-onautical

Laboratories

Rudd was the Air Force Project

University

for

the

Air

(AFWAL/FIBEC).

Engineer

and

under

Force

James L.

Dr.

Jack

W.

L4ncoln of ASD/ENFS was a technical advisor for the program.

Dr.

B. G. W. Yee of the General Dynamics'

Laboratory

was

the

Program

Manager

Materials Research
and

Manning was the Principal Investigator.

George

Washington

Dr.

University (Washington,

Shinozuka of Modern Analysis

Dr.
J.

Sherrell D.
N.

D.C.)

Incorporated

Yang

of

and Dr.

M.

(Ridgewood,

New

Jersey) were associate investigators.

This

program was supported by several General Dynamics'

personnel as follows: All tests were

Dynamics'

Metallurgy

Laboratory

direction

of

Nordquist.

F.

C.

responsible

for

Fractographic

readings were made by

Kaarlela,
Speaker

A.

the

Meder,

acquired

and

by

R.

W.

0.

the

supported

testing
the

E.

Nay and S.

initial

and

General

Kaarlela

data

D.

in

Nay tinder the

0.
T.

fractographic

R.

coordinated

performed

was

acquisition.

Gordon,

M. Speaker.

W.

T.

S.

fractographic

M.

data
Li

fatigue quality model

SDi stributIon/
Availability Codes

Avail and/or
Dist

Special

-61
S"

''

'

+ ,

'

"

.- ,'

,"''V7

,"-"

,"

-"%,

""-

'r

-.-

-. -.

-. .-.

.-

. "

..

calibration/e,,aluation studies.
computer

software

for

storing

calibration/evaluation

and

Henslee

survey.

Dr.

analyzing

initial

fatigue

studies

preliminary version of the handbook.

P.

J. W. Norris developed

fractographic data, supported the

model

"

and

the

Y. H. Kim,

Dr. W. R. Garver and M. A.

contributed

to

assessment.

F-16 durability

the

durability

and

a
S.

structural durability

analysis

test

quality

B. J. Pendley

aircraft

the

worked on

conducted

the

Flanders

state-of-the-art

results

and

supporting

data for the durability analysis demonstration were provided

by J. W. Morrow, V. Juarez, D. R. McSwain,
Dr.

V.

D.

Smith

analysis effort.
"by T.

E.

Ernestine

supported

the

and P. D. Hudson.

modeling and statistical

Photoela.tic investigations were conducted

Love.
Bruner.

Typing was performed by Peggy Thomas and

Ron

Jordan

prepared

many

of

the

illustrations and Joe Conde" provided printing and editorial

"support.

This

handbook

i'i

the final product of the "Durability

Development" program.

SMethods

design

requirements

are

The U.S. Air Force durability

reviewed

and

methodology (i.e.,

economic life criteria, analytical tools, guidelines,

data,

etc.)

for satisfying these requirements are described

and discussed.

:4

design

.,

The

following

reports

(AFFDL-TR-79-3V18)

were

also

prepared under the "Durability Methods Development" program:

Phase I Reports

* Vol.

* Vol.

II

Vol.

III

- Phase i Summary

Durability Analysis: State-of-the-art Assessment

Structural Durability Survey:

btate-of-the-art

Assessment
* Vol.

IV

Vol.

- Durability Analysis Methodology Development

Vol.

VI

- Documentation of Computer Programs for Initial

Initial Fatigue Quality Representation

Quality Representation

Phase II

(Vol.

IV)

Reports

Vol.

VII - Phase II Documentation

* Vol.

VIII-

Test and Fractography Data

* Vol.

IX

Documentation of Durability Analysis

Computer Program

This handbook covers work accomplished during the period

July1981 through
This

report

January
was

1984.

r:eleased

1984.

for publication in February

TABLE OF CONTENTS

Section
I
,

II

Page
INTRODUCTION

1.1

1.1

General

1.1

1.2

Bat'

1.2

--. ound

DURABILITY DESIGN REQUIRENENTS AND ANALYSIS

2. 1

CRITERIA/GUIDELINES
2.1

Introduction

2.1

2.2

Durability Design Requirements

2.1

"2.2.1 Objective and Scope

2.1

2.2.2

General Requirements

2.2

2.2.3

Analytical Requirements

2.2

2.2.4

Experimental Requirements

2.4

2.3

Durability Analysis Criteria

2.3.1

Durability Damage Modes

2.5

2.3.2

Durability Critical Parts Criteria

2.5

2.3.3

Economic Life Criteria/Guidelines

2.6

2.3.3.1

Economic Life Definition

2.3.3.2

Economic Repair Limit

2.10

2.3.3.3

Extent of Damage

2.11

2.3.3.4

Formats for Economic Life Cri.teria

2.12

vii

2.8

TABLE OF CONTENrS

(Continued)

Section
III

3.1

Introduction

3.1

3.2

General Description of the M.ethod

3.1

3.3

Assumptions and Limitations

3.4

Initial

Fatigue Qv.lity Model

3.6

3.4.1

TPQ Model Equations for Case I

3.4.2

IFQ Model Equations for Case II

3.5
IV

3.1

SUMMARY OF THE DURABILITY ANALYSIS METHOD

(L,> 1)

3.10

1)

3.13

(b

3.15

Durability Analysis Procedures

INITIAL FATIGUE QUALITY DETAILS

4.1

4.1

Introduction

4.1

4.2
4.3

EIFS Distribution
Test/Fractcgraphy Guidelines

4.2

4.3.1

Test Guidelines

4.3.2

Guidelines

4.4

Procedures
Parameters

4.4.1

4.5
4.6

for Fractographic

for C-1.ibrating

Data

the IFQ Model

Generic Nature of IFQ and Data Pooling

4.11
4.12
4.13

Concepts
4.4.2

Calibration and Data Pooling Procedures

4.19

4.4.2.1

Determination of Qi

4.4.2.2

Determination of a.i

4.4.2.3

Determination of a and QB

viii

4.19

and c

4.25
4.27

TABLE OF CONTENTS (Continued)

SectionPage

4.4.2.4

Optimization of Parameters

4.30

4.4.2,5

General Steps

4.30

4.5

Statistical Scalii-g of

4.6

Illustration of Procedures for Determining

IFQ

4.37

4.6.1

Determination of TTCI and Qi Values

4.39

4.6.2

Determination of a and Q3

4.43

4.6.3

Goodness-of-Fit of IFQ Model

4.49

4.5.4

Discussion of EIFS Distribution

4.61

4.6.5

Practical Asr.ects

4.63

DETAILS FOR PERFORMINSI\

3 for Multiple Details

DURABILITY ANALYSIS

',.33

5.1

5.1

Introduction

5.1

5.2

Service Crack Gr.'-owth Master Curve

5.1

5.2.1

5.2

Guidelines

"5.2.2 Illustratic

5.5

Case 1
Case 2

5.6

5,2.2.1
5.2.2.2

5.7
I'redictions

5.3

Crack Exceedanc,

5.4

Extent of Damagz Tormats/Illustrations

5.10
5.11

5.4.1

Extent of Dr,:ge Formats

5.11

5.4.2

Extent of Dar-.ge Illustrations

5.11.

(Continued)

TABLE OF CONTENTS

Section
5.5

VI

Additional

5.19

Considerations

5.5.1

Details Other Than Fastener Holes

5.19

5.5.2

Large Crack Sizes

5.20

5.5.3

Effects of Scale-Up and Hole Interactions

5.20
5.22

Functional Impairment
5.5.4
COMPARISON OF DETERMINISTIC AND PROBAPILI.STIC

6.1

APPROACHES FOR DURABILITY ANALYSIS

6.1
.2
,.3
6.4

VII

Introduction

6.1

F-16 Durability Analysis Approach

6.2

Probabilistic Fracture Mechanics Approach

6.6

Exampl-! Durability Problem

6.6

6.4.1

Durability Analysis

Based on the DCGA

6.8

6.4.2

Durability Analysis

Based on the PFMA

6.10

6.4.3

Conclusions

1.12

DURABILITY ANALYSIS DEMONSTRATION

7.1

7.1

Introduction

7.1

7.2

F-16 Lower Wing Skin

7.1

'/.3

Complex Splice Specimens

7.21

Subjected to a

B-I Bomber Spectrum

7.4

Conclusions and Recommendations

7.27
R. 1

REFERENCES

x
S-'.,

,4,.,,.,".""Y

/ ,

./

"".

''''''',

'",",/

,"-

-"-

-" ",

"""""""-

.:'

,.

,","

LIST OF ILLUSTRATIONS

Figure

Pg

1.1

Wing Box Assembly

1.3

2.1

U.

Air Force Durability Design Requirements

2.3

2.2

Flow Diagram for Selecting Durability Critical

Parts

2.7

2.3

Analytical Format for Economic Life

2.9

2.4

Durability Analysis Approach

3.1

Essential Elements of the Durability Analysis

Methodology

3.3

3.2

Initial

3.8

3.3

Growth of EIFS Distribution as Function of Time

3.17

4.1

No-Load Transfer Specimen with Multiple Details

4.8

4.2

Two-for-One No-Load Transfer Specimen Design

4.8

4.3

Reverse Double Dog-Bone Specimens

T.ond Transfer)

4.4

Illustration
of Concept of Transforming Individual
TTCI Distributionsfor Different Data Sets into a
"Generic" Ei"' Distribution

4.16

4.5

Illustration
Showing TTCI's and EIFS Master Curve
for the ith Fractographic Dura Set

4.17

4.6

Notational Scheme for Detemnination of Qi

4.22

4.7

Conceptual Description of Fractographic Data uscd

to Determine Qt for the ith Fractographic Data Set

4.24

4.8

Illustration
of A Scaling Concept
Fractographic Data Set

4.36

4.9

EIFS Cumulative Distribution Parameters for Pooled

Fractographic Data Sets (AFXLR4, AFXMR4, AFXHR4)

S.

2.13

Fatigue Quality Model

xi

(15% Bolt

for the ith

4.10

4.52

LIST OF ILLUSTRATIONS

(Continued)

Figure

.1

Page
TTCI Goodness-of-Fit Plot for Pooled Fractographic
Data Sets (AFXTR4, AFXMR4, AFXHR4); a 0 = 0.035"

4.11

TTCI Goodness-of-Fit Plot for AFXLR4 Data Set Based 4.55

on IFQ Model Parameters for Pooled Data Setsj
s = 0.035"

4.12

TTCI Goodness-of-Fit Plot for AFXMR4 Data Set

Based on IFQ Model Parameters for Pooled Data Sets;
a 0 = 0.035"

4.56

4413

TTCI Goodness-of-Fit Plot for AFXHR4 Data Set

Based on IFQ Model Parameters for Pooled Data Sets;
a 0 = 0.035"

4.57

4.14

TTCI Goodness-of-Fit Plot for AFXLR4 Data Set

4.58

.. ased on IFQ Model Parameters

a0 = 0.050'
%%
"

4.54

4.10

for Pooled Data Sets;

4.15

TTCI Goodness-of-Fit Plot for AFXMR4 Data Set

Based on IFQ Model Parameters for Pooled Data Sets;
= 0.050"

4.59

4.16

TTCI Goodness-of-Fit Plot for AFXHR4 Data Set

Based on IFQ Model Parameters for Pooled Data Sets;
a 0 = 0.050"

4.60

4.17

Plot of Qk i versus a to Determine Applicable

4.62

"5.1

Service Crack Growth Master Curve

5.2

Formats for Presenting Extent of Damage Results

6.1

F-16 Durability Analysis Approach

6.3

6.2

Probabilistic Fracture Mechanics Approach

6.7

7.1

Stress Zones for F-16 Lower Wing Skin

7.4

7.2

Q- Versus

7.9

"7.3

Qi Versus Gross Stress for the F-16 400-Hour

Spectrum

-.

k. for the F-16 400-Hour Spectrum

xii

SCGMC

5.3
5.12

7.11

LIST OF ILLUSTRATIONS

(Continued)

Figure

Page

7.4

Percentage of Crack Exceedance Versus Crack Size

at 16000 Hours for 3 Probability Levels (F-16
Fighter)

7.14

7.5

Service Crack Growth Ma t,Crack Size Ranges

7.16

,>rves for Different

4

7.6

Number of Holes with Crac

e
0 03
Versus
Flight Hours - Exceedance Probability Format
(F-16 Fighter)

7.18

7.7

Average Percentage of Holes with Crack Size

7.20

0.03"

Versus Flight Hours

Stress Level

Format (F-16 Fighter)

7.8

Complex Splice Specimen

7.22

7.9

Q Versus Gross Stress for B-I Bomber Load

Spectrum

7.25

7.10

Average Percentage of Holes with Crack Size

7.26

L0.05" Versus Flight Hours - Stress Level

Format (B-i Bomber)
7.11

7.12

Percentage of Crack Exceedance Versus Crack Size

and Exceedance Probability at 13500 Hours (B-i
Bomber)

7.28

Number of Holes (Outer Row) with Crack Size

L 0.05" Versus Flight Hours - Exceedance
Probability Format (B-I Bomber)

7.29

.1.

4,

xiii

kI

".4

'++

+m+

+.

..

"

"

""

"

L.

-+

"

+"+"

"LIST OF TABLES
Table

Page

4-1

Description of Fractographic Data Sets

4.38

4-2

Illustration of Procedure for Determining TTCI

Valaes for Selected a(t)'s

4.40

4-3

Illustration of Proceduresfor Determining

for Data Set AFXMR4 (Specimen No. 567 HB)

4-4

Summary of TTCI Values for Data Set AFXLR4

(7475-T7351 Aluminum)

4.44

Summary of TTCI Values for Data Set AFXMR4

4.45

S4-5

Qi

4.42

(7475-T7351 Aluminum)
4-6

Summary cf TTCI Values for Data Set AFXHR4

(7475-T7351 Aluminum)

4.46

4-7

Summary of Qi Results Based on Eq.

4.47

4-8

Summary of IFQ Model Parameters and Results for

Individual Data Sets (a 0 = 0.035"; xu = 0.025")

4.48

"4-9

Illustration of Prucedures and Results for Determination of a for Given Qk (a 0 = 0.035",

xu = 0.025")

4.50

4-10

Summary of IFQ Model Parameter Results

4.51

5-1

Analytical Crack Growth Results for Spectrum "A"

S(v =42 ksi)

5-2

Extent of Damage Assessment for Wing Skin Containing Fastener Holes

5.16

5-3

Extent of Damage Assessment for a Component

Containing Different Detail Types

5.18

6-1

Illustration of the Deterministic Crack Growth

Approach and the Type of Information Obtained
from the Analysis

V'

4-6

5.9

6.9

xv

......................................
-

..

LIST OF TABLES (Continued)

Table

Page

6-2

Illustration
of the Probabilistic Fracture Mech-anics Approach and the Type of Information
Obtained from the Analyris

7-1

Stress Levels and Number of Fastener Holes for

F-16 Lower Wing Skin

7.5

7-2

Summary of IFQ Model Parameters for F-16 400-Hour

Spectrum

7.7

7-3

Durability Analysis Results for F-16 Lower Wing

Skin

*.%

N;

4',

xv

6.11

7.12

LIST OF SYMBOLS

Crack size

aDL

Durability limit flaw size

Economic repair size limit

aae
a0

Reference crack size for given TTCI's

a(O)

Crack size at t-O

S- Repair.

aRL
a(t

a(t),
SaUt

),

a(t

2)

limit flaw size

Crack size at time t,

t1 and t

respectively

aL

Upper and lower bound fractographic

crack size, respectively, used to

define the IFQ model parameters
a(T)

Crack growth parameters in the

Used
b
a
equation da(t)

b,

Crack size at service time r

dt .1-r
in
bi

Qi

(a(t0

conjunction with the IFQ model.

Crack growth constants in

for the ith
Q [a(t)] bi
da(t)
dt
dt

st'ress region when this equation

is used in conjunction with the
service crack growth master curve

(SCGMC).
*

b., Qi

.R data
*
Crack growth constants in dt -i(a(t)]
for the~ith fractographic data set,

where bi - 1.0.

Notation used in

conjunction with fractographic data

pooling procedures and the EIFS master
curve for the ith fractographic data set:.

',

xvii

-.

_,i

.-.

-bij'

",

...-.

QIJ

:r -.-

-.

,<

Crack growth constants in

b"
Qij [a(t)] bi,J for the

dt
i
ith fractographic data set and
the jth fractographic samkle from
the ith data set, where bi.J A 1.0.
Notation distinguishes parameters
for each sample in a given fractographic
data set.
b - 1; Used in

c=

conjunction with the

IFQ model when the crack growth law,

da't)QQ[a(t)]b is used and b > 1.0.
dt

c.

3.

1;

Used in

"SCGMC when da(t)

dt

.. ,

The subscript "i"

stress region.
CFA

conjunction with the

bi is used.
is used
refers to the ith

-i%

a\t)]

Conventional Fatigue Analysis

(Palmgren-Miner rule)
D)CGA

Deterministic Crack Growth

Approach
dt

Crack growth rate as a function of time

EIFS

Sf a(%x)

Equivalent initial flaw size

-

EIFS probability density function

dFa a()(x)
dx

. T(.t)

dFT(t)
dt

Fa(O)(x)
SFT(t)

EIFS cumulative distribution function

TTCI cumulative distribution function

FHQ
IFQ

Fasten.er hole quality

M

Initial fatigue quality

xviii

..

S-

Factor uaed in conjunction with the

fractographic results for a single
TTCI data set and the equatioi.:

is the number of
8 - 8"
equalfy-streseed fastener holes per

2.

test specimen in which only the

"largest fatigue crack in any "one"
"hole per specimen is included in the
fractographic data set.

i -

Same as I with the subscript "i"

referring to the ith TTCI data set
and used in the equation:
-i

L()

iI

Total and average number of details,

respectively, in the entire component
having a crack size >xI at any service
time r

Load transfer through the fastener

T(T)

*
LT

8 (2.)*l~fi.

the ith

Total number of details in

Ni

stress region

Total number ef details in the entire

durability critical component

N(it), N(i,T)

Total and average number of details,

respectively, having a crack size
exceeding x 1 at any service time T

NLT

No load transfer through the fastener

Probability or exceedance probability

PFMA

Probabilistic Fracture Mechanics

Approach

p(iT)

Probability that a detail in the ith

stress region will have a crack size
>x at the service time T

Xix

",.r.

*',.,

-. ,*"%""*"

.,:

".',-

""*"

'

' ' , "" : """'-

- """"". "."." ' - " " " "

""

" -"-"

" " """

"""

^Ave

n~
n .

.Normalized

Avien

crack growth parameter, Q , for the

*
ith fractographic data set when the
i
dtt) Q*[a()]
bj.
crack growth equation,
is used and bi - 1.0. Used in conjunction
ith
with the EIFS master curve for the
fractographic data set.
An

'C

Ave.

QO

8i

Qi a

(1 . )

constant for"generic" EIFS cumulative

distribution.
*
Ave QiOX

Q-

Us

Used when

checking IFQ model goodness-of-fit when

"fr.actographyis available only for the
largest fatigue crack in any one ki
fastener hole per test specimen.
=

SCGMC
t,

t1 ,

Service crack growth master curve

Flight hours at t,

t2

t1 ,

t2 ,

T, TTCI

Time-to-crack-initiation

Crack size
Crack size used for p(iT)

respe.ctively.

predictions

Upper bound limit for EIFS

S=

An EIFS in the IFQ distribution

"corresponding to a crack size x

time T in the ith stress regioA.

Value determined using the SCGMC.

V'

xx

at

No, of standard deviations from the

mean

U,

ai

Weibull distribution parameters for

shape, scale, and lower bound TTCI,
Used in conjunction
respectively.
with the IFQ distribution or for a
single TTCI data set.

F,

0i

Weibull distribution parameters for

the ith TTCI data set for shape, scale,
and lower bound TTCI, respectively.
Used in conjunction with fractographic
data pooling procedures (i - 1,...,n

ei

data sets).
Weibull scale parameter for TTCI based

on the TTCI's for a given fractographic

data set in which only the fractography
for the largest fatigue crack in any
"one" of fastener holes per test
Note:
specimen is used to define 0
1
8 = 8, (t) I/a
with the subscript "i" dencting
the 8 value for the ith TTCI data set and
as follows:
is used to determine
B 4- d I/ai.
Same as 8

r( )

Gamma function

*,y

Empirical constants in the equation:

Qi . &ay, where a = stress

-r
2

T2(i)T),

TL(i,T)

Stress or standard deviation

Variance of N(i,T)

and L(T),

A particular service time

xxi

respectively

TERMINOLOGY

1.

Crack

Size

the length of a crack in a structural

-is

detail in the direction of crack propagation.

2. Deterministic Crack Growth

Crack growth parameters are

treated as deterministic values resulting

in a

single

value prediction for crack length.

3. Durability

is a quantitative measure of the airframe's

resistance to fatigue cracking under

conditions.

small

functional

subcritical

impairment,

crack

structural

details

repaired,

economical

exceed

population

of

is susceptible
Therefore,

containing

sizes

structural

such

which

maintenance

crait may not

However, if the

requirements and lie-yc-ohs

pose an immediate safety problem.

cracks

service

Structural durability is normally concerned

with relatively
affect

specified

are

cracks

not

repairs cannot be made when these

limiting

crack

size.

The

entire

structural details in various components

to

fatigue

statistical

quantitatively assess the

cracking
approach

"durability"

in

service.

is essential
of

to

part,a

component, or an airframe.

4,

4,

Durability Analysis

is

concerned with quantifying

the

extent of structural damage due to fatigue cracking

structural details

lug,

etc.)

used

5.

lity

ensure

Life -

structure's

point

fillet,

design

compliance

cutout,

Results

with

Air

are
Force

design requrements.

Economic

damage

damage

rastener hole,

as a function of service time.

to

durf

(e.g..

for

and/or
where

is

that point
state

due

environmental
operational

in

time when an aircraft

to

fatigue,

deterioration

readiness

goals

accidental

reaches
cannot

a
be

"preserved by economically acceptable maintenance action.

6.

Economic Life Criteria - are guidelines and formats for

,defining qua.,titative
aircraft

design
provides

economic

life

structure to satisfy U.S.

requirements.
the

The

for

Air Force durability

economic

life

criterion

basis for analytically and experimentally

ensuring design compliance of

durability design requirements.
for economic

reauirpments

life

criteria

aircraft

structure

with

Two recommended formats

are:

o probability of crack exceedance

o cost ratio: repair cost/replacement cost

xxiv
.-

7.

Economic Repair Limit

is

the maximum damage size that

can be eccnomically repaired (e.g.,

radial

crack

repair 0.03" - 0.05"

in fastener holes by reaming hole to next

size).

8.

Equivalent
crack

Initial Flaw Size (EIFS)

assumed

service.

It

to

exist

characterizes

actual initial flaws in

determined

in

the

time.

The

details

exceedance

or

An EIFS is
actual

EIFS

tool"

structural

prior

equivalent

to

effect of

detail.

It

assumed to have

concept
for

and

extent

is

results.
the

salie

convenient
the

probability

of

initial

flaw

size.

Within

negative,

this context,

depending

on

the

including:

results used (and the test variables

the fractographic crack size range used,

the

than

results and the back extrapolation method

fractographic

cf

crack

of damage as a function of time.

EIFS values depend on several factors,

reflected),

for

IFQ

strictly a mathematical quantity rather

fractographic
used.

is

quantifying

the

EIFS's can e positive or

form

structure

by back-extrapolating fractographic

"mathematical

the

a hypothetical

shape and origin as the observable crack size at a

given

an

is

structural

An equivalent initial flaw is

flaw

the

crack

the

growth equation used for the back-

4XXV

extrapolation,

the goodness of

fractographic

data,

the

structural details,
and

fit,

are not

etc.

if

type,

fit

to

the

quality of the
fastener

type

EIFS's for different fractographic data

are

unless

the

determined

the

applicable

consistently

fractographic crack size range

imposed

curve

manufacturing

fastener hole

comparable

parameters

the

Q,[a(t)jb

used,

same

IFQ

model

(e.g.,

same

"b "

value

crack growth model is used,

same a values imposed for comparable fractographic

sets,

etc.).

Fractographic data pooling is

data

essential to

quantify the IFQ for different fractographic

data

sets

on a common baseline.

9.

EIFS

Master

Curve

tabulation of a(t)
functional form)

vs.

is

t or

curve
curve

(e.g.,
without

equation,
prescribed

used to determine the EIFS value at t=O

corresponding to a given TTCI value at a specified crack

size.

Such

distribution
mastei

curve

curve

from

is

the

depends

used,

the

TTCI

the

distribution.

fractographic

functional

form

equation used in the curve fit,

etc.

xxvi
I

to determine the IFQ

on several factors,

fractographic data base,

range

needed

The
such as
crack

EIFS
the
size

of the crack growth

(Ref.

EIFS).

10.

of

Extent

Damage

is

structural durability at
number

the

example,

fastener holes,

cutouts,

exceeding

details

given

a
of

etc.)

fillets,

F'-:

time.

service

(e.g

details

structural

or percentage

length is the fundamental measure for structural dar

The

predicted

extent

of

is compared wit

damage

specified economic life criterion

compliance

11.

with U.S.

GE 2ric

EIFS

"generic"

if

depends

distributicn

variables,
lcad

only

on

analysis",
(e.g.,

be

en'vironment,

the

given

types/fit,

should

the

EIFS

should

etc

be

of

For
for

percent

"durability
fastener holes

procedures,

justified

the

(Lesign

stress level,

drilling

and

Theoretically,

distribution

material,
etc.)

:sign

material

independent

such as load spectrum,

transfer,

the

An EIFS distriDuticn is

manufacturing/fabrication processes.

EIFS

ensuring

je.

Air Force uurability requiremrnts.

Distribution
it

for

-f

C" ck

crack size limits.

specified

of

measure

quantitative

for

fastener
different

design stress levels and load spectra.

12.

Initial

Fatigue

Quality

initial manufactured state of

details

with

component,

or

respect

to

airframe

characterizes

structural

initial

prior

xxvii

(IFQ)

to

flaws
service.

the

detail
in

a
The

or

part,
IFQ,

represented

by

an

distribution,

must

fractographic

data base.

be

defined

EIFS

(Ref.

EIFS

distribution

specifically

for

and
is

range

used

it

should

to

be

single

used

EIFS

in

(e.g.,

0.0005"

understanding,
for

a fairly

required

other

Whatever

be

defined

durability

distribution

0.10").

will

range

of

not
crack

Based

on

current

the EIFS distribution should

be

defined

small range of crack sizes

0.050" crack size for fastener holes).

is

and

the

necessarily be satisfactory for a wide

sizes

consistent

crack size range of interest for

the structural details

analysis.

EIFS master curve).

used,

the

using

The EIFS distribution depends

on the fractographic crack size

factors

(EIFS)

flaw size

initial

equivalent

(e.g.,

0.020"

Further reE arch

to evaluate the effects and sensitivity of

the crack size range on the EIFS

distribution

and

the

accuracy of the crack exceedance prediction.

13.

Initial Fati4ue Quality Model - is

for quantifying
structural

fractographic
determined

the

IFQ

details.

which

distribution
Using

results,
is

an

EIFS

the

xxviii

for
IFQ

applicable
model

distribution

compatible

distribution.

*1t

a "mathematical tool"

with

the

can

and

be
TTCI

..

14.

..

Probability of Crack Exceedance

(p(ir))

probability of exceeding a specified crack,

given service time,

refers to the

xj, size

at

can be determined from the

It

r.

statistical distribution of crack sizes and can be

used

to quantify the extent of damage due to fatigue cracking

fillets, lugs,

in fastener holes, cutouts,

15.

- This is the specified crack

Reference Crack Size (aN)

is

distribution

reference

to

size in a detail used

on

based

etc.

The

TTCI's.

IFQ

selected reference crack

size.
A'

16.

Service

Crack Growth Master Curve (SCGMC)

is used to determine the EIFS,

yli(T),

This curve

corresponding

to

The probability
exceedance crack size xI at time r.
of crack exceedance, p(i,r), can be determined from the
an

.IFS

cumulative

distribution

(e.g.,

stress

applicable

the

SCGMC is defined for

level,

for a given y 1 i(r).

design

variables

and

it can be

etc.)

spectrum,

The

determined using either test data or an analytical crack

growth program.

All SCGMC's must be consistent with the

corresponding EIFS master curve

data

base.

and

the

fractographic

The SCGMC must be consistent with the basis

for the IFQ distribution.

xxix

.--.

...-

7,

17.

Detail

Structural

to

susceptible

structure

"18. Time-To-Crack-Initiation
to

hours required

(TTCI)

a0 ,

in

a metallic

in

(e.g.,

cracking
etc.).

lug,

initiate

crack size,

fatigue

element

fatigue

cutout,

fillet,

fastener hole,

any

is

is

the time or service

specified

(observable)
(with no

a structural detail

initial flaws intentionally introduced).

19.

TTCI

Lower

Limit

Bound

TTCI's reflected in the IFQ

given
xu,

model.

a0 ,

should

It

for

varies

TTCI's for a given crack

e.

This

basis

for

parameter provides a
distribution

a cutoff value for

depends on the EIFS upper bound limit,

and it

and the EIFS master curve.

size,

e is

Weibull

distribution

quantifying

the

EIFS

for different TTCI crack sizes on a common

baseline.

20.

Upper Bound EIFS (Y1)

initial fatigue
specified

- defines

quality

the largest EIFS in

distribution.

The

xu

by the user should be consistent with

lower bound limit)

and the EIFS master curve.

xxx

the

value
e (TTCI

SECTION I

INTRODUCT IO
1.1

This
Handbook.

is

the

first

GENERAL

edition

the Durability Design

of

The purpose of the handbook is to:p

o summarize the essential Air Force durability

design requirements for metallic airframes (1-31,p

o describe methodology for satisfying the durability]

design requirements,

o provide guidelines and design data for implementing

the methodology and for demonstrating design

o provide a framework, with a loose-leaf format, for

incorporating future durability methodology advancements and design data.

This

document

is loosely

developments and design data

called
are

"Handbook".

required

to

Further

expand

and

refine

the document for efficient design usage.

Therefore,

the handbook reflects the current understanding of

Force's

durability

the

Air

design requirements and provides state-

of-the-art concepts, tools

and

guidelines

for

satisfying

these requirements.

The

material

presented

intended for durability

airframes.

in this handbook is primari .V

design

applications

for

metallic

However, many of the concepts, analytical tools,

da~a and guidelines can also be used to asseiss the extent of

damage due to cracking for in-service aircraft.

1.2

BACKGROUND

Aircraft structures have thousands of structural details

susceptible to fatigue cracking:
cutouts,

etc.

holes,

fillets,

For example, the wing box assembly shown in

Fig. 1.1 has over 3000 fastener

alone.

fastener

Fatigue

cracking

most prevalent forms of

in

holes

in the

wing

skins

fastener holes is one of the

structural. damage

for

in-service

aircra~ft (4-8].

Durability is a measure of the structure's resistance to

fatigue

cracki.ng.

details

in various

The

entire

components

1.2

population

of

structural

is susceptible to fatigue

11101

Fig. 1.1

Wing Box Assemnbly

1.3

cracking in service.

Therefore,

the

extent

structure

or

to assess the durability of

of

damage

(i.e.,

percentage of structural details in a part,

component

number

or

structure,

or airframe exceeding specified crack size limits

that cannot be economically repaired) as a function of time,

the

entire

population

accounted for.

Thus,

of

structural

details

must

be

a statistical approach is essential to

quantify the extent of damage as a function of time.

Structural

durability

relatively

small

functional

impairment,

subcritical

and life-,cycle-costs.
safety

ir generally

problem.

crack

concerned

sizes

which

with
affect

structural maintenance requirements

Such cracks may not pose an immediate

However,

structural

details

containing such cracks are not repaired, economical

repairs

cannot

be

size.

For

fastener

made

when

example,

hole

crack

the

these cracks exceed a limiting crack

a

0.030-0.050"

radial

The

size

in

economical
a

detail

repair
that

without further repair or part replacement.

detaiLs

are

crack

in

can be cleaned up by reaming the hole to the

next fastener size.

maximum

if

limit

is

the

can be cleaned-up
If

structural

not repaired or parts replaced at an opportune

time, expensive repairs or parz replacement may be required.

Also,

unrepaired

cracks may reach sizes which could affect

structural safety during the design life of the aircraft.

1.4

"Aircraft

safety

structural

"

--.

--

by

governed

is

damage

tolerance conditions whichare concerned with the structure's

resistance

to failure due to cracking.

Damage tolerance is

typically concerned with the largest crack size in a

detail.

For

example,

the wing

box

is

details.

in Fig. 1.1 the damage tolerance of

limited

the

However,

by

few

durability

critical

structural

the

box

of

concerned with the entire population of

and

the

size

of

single

the

largest

wing

structural

subcritical

is

details

crack in each

detail.

The

conventional fatigue analysis (CFA)

approach (i.e.,

Palmgren-Miner rule, Ref.

9,10) and the deterministic

growth

[11] do not provide a quantitative

approach

(DCGA)

lescription of the "extent

,ervice

time.

The CFA,

quantify crack sizes

of

damage"

as

crack

f anction

in its commonly used form, does not

for

population

of

details

essential requirement for any durability analysis method.

DCGA can be used to predict the growth of a single crack
a detail as a function of time.

Using the DCGA,

to

analytically

does

not

quantify

the

in

can

be

1.5

size.

However,

probable crack sizes or

ranges of crack sizes for the population

[ Sb"'

assure that the largest crack size in

the group of details will be S a specified

"the DCGA

an

details can

"be grouped and the "worst-case" detail in the group

used

Cf-

of

details,

CFA

and

the

DCGA

have been evaluated

for potential durability

analysis applications [12,133.

'I

'
II

a,

U'

'

,1.6

SECTION

II

DURABILITY DESIGN REQUIREMENTS

AND ANALYSIS CRITERIA/GUIDELINES

2.1

INTRODUCTION
LA

purpose

The

this section is

of

and interpret the important

elements

durability design requirements (1-3],

critical parts
recommended

criteria

and

(3)

to:

(1) briefly review

of

the

Air

Foice's

(2) discuss durability

provide

guidelines

and

formats for defining quantitative ecolomic life

criteria.

2.2

DURABILITY DESIGN REQUIREMENTS

2.2.1

The

objective

of

requirements [1-31 is
costs

and

selection

maximize
of

inspections,

Objective and Scope

the
to

Force

minimize

operational

materials,
and

Air

stress

protection

durability

in-service
readiness
levels,

design

maintenance

through proper
design

systems.

These

details,
design

requirements include both analyses and tests.

2.1
.4\
*

.*-

2.2.2

Essential

durability

described in Fig.

General Requirements

2.1,

requirements,

conceptually

are as follows:

The economic life of the airtrame must exceed one

design service life.

No functional impairment

(e.g.,

loss of control effectivenes,

loss of stiffness,
loss of cabin pressure

or fuel leaks) shall occur in less than one

design service life.

The economic life of the airframe must be demonstrated

Sanalytically and experimentally.

2.2.3

AnerI2yses

are

Analytical Requirements

required to demonstrate tha~t the economic

life of the airframe is greater than the design service life

when

subjected

to

the

desigr

service

loads

chemical/thermal environments.

The economic

must

quality,

account

sequence,

for

initial

material property variations,

must be verified by tests.

2.2

life

and design
analysis

environment,
etc.

The

load

analysis

_
.

cr

"-",,CRACK

CRITICAL CRACK LENGTH

I,DESIGN SERVICE LIFE

SIZE

FUNCTIONAL IMPAIRMENT

of

%d

ECONOMI

LIFE

""--,

~4-Uf.

[]i

SI!

-:,-:-. . '. ,, ' '-:-% ,.,- . .," , -.".. , .:.,'-'-\ ,., -"" "., -:... \ ,., . - ,,..... A,,>.-:.'-

"";to

tf

,.,

TIME

Fig.

2.1

U.

S.

Air Force Durability Design Requirements

2.3

Itt

2.2.4

Design
*

Experimental Requirements

development

early evaluation

tests

are

required

of the durability of

to provide an

critical

components

and assemblies as well as the verification of the durability

analysis.

durability

test of a full-scale airframe may also be

required by the Air Force.

-,

The requirements for

this

test

are:

1.

The airframe must be durability tested to one

lifetime.

Critical structural areas must be

inspected before the full, production go-ahead

dec

2.

S101o.

Two lifetimes of durability testing plus an

inspection of critical structural areas must be
completed prior to delivery of the f~irst production
aircraft.

If

the

economic

life

of

the airframe is not reached

before two lifetimes of durability

options are available:

2,4

testing,

the

following

.'

kb

-%.

'

-,

1.

.'

- "-'

'

Terminate

the

..

".

testing

durability

nondestructive inspection followed

",

..

perform a

and
by

"

destructive

teardown inspection.

Terminate the durability testing and perform damage

2.

tolerance

testing

and

nondestructive

inspection

followed by a destructive teardown inspection.

3.

Continue

the

durability

testing

for an approved
preceding

period of time followed by either of the

options.

DURABILITY ANALYSIS CRITERIA

2.3

2.3.1

are several modes of durability damage,

There
fatigue

"

Durability Damage Modes

cracking,

importance

and

corrosion,

prevalence,

wear,

Criteria

Due

fatigue cracking is

structural degradation considered

2.3.2

etc.

including
to

its

the form of

in this handbook.

Durability Critical Parts Criteria,

must

be developed for determining which parts

of an aircraft are durability critical

(i.e.,

which

parts
2'.

2.5
4'-

I_
.- .. "....

- .'" -." "

" ".

" .." ,"

N % " .. ," ." " . ." ," . ." .'.

'..

,'. '. ',.

-' .'. " . ,' .,

must

to

designed

be

design

durability

tht

meet

The durability critical parts criteria

requirements).

from aircraft to aircraft.

vary

They are especially dependent on

the definition of economic life for the particular aircraft

A typical flow diagram for selecting vhich parts
involved.
In Fig.

are durability critical is presented in Fig. 2.2.

2.2,

refers

durability

resist cracking

whereas

ability of an airframe to

the

to

tolerance

damage

refers

the

to

ability of an airframe to resist failure due to the presence

of such cracks.

2.3.3

Economic Life Criteria/Guidelines

must be developed for determining the economic

Criteria

life of the particular aircraft of interest.

vary from aircraft

fastener

parts criteria, economic life criteria

critical

durability

hole

Similar to the

to

repair

aircraft.
(e.g.,

They

may

based

be

on

reaming the damaged fastener

hole to the next nominal hole size),

functional

(e.g., fuel leakage), residual strength, etc.

impairment

Two promising

analytical formats for quantifying the economical life of an

airframe

are

(1) the probability of crack exceedance, and

(2) cost ratio: repair cost/replacement cost.

require

durability

analysis

methodology

quantifying the extent of aircraft structural

2.6

Both

formats

capable
da-mage

as

of
a

.ESAILISH DESIGN CRITERIA FOR A PART71

00

STATIC STR
YES

PART ALWAYS LOADED IN COMPRESSION?

100 DAMAGE TOLERANCE & DURABCLITY ANALYSISI

1O DURABILITY ANALYSIS
0

DAMAGE TOLERANCE MORE CRITICAL\

__/
T4AN DURABILITY?

URABTY CONTROL STRESS LEVEL?

YES

NO

YES

~~DAMAGE TOLERANCE CONTPOL(E Y

STRESS LEVEL?

fNO
IUSECONTROLLE
O

MATERIAL

YS

-r

N-PE CTAILE?
ONN
PATH OR
/SIGLELOAD

RIYE

AA-0

Fig.

2.2

Flow Diagram For Selecting Durability Critical Parts

2.7
,

SN

L A P T OR

function

of service time.

"life criteria are based on

which

cannot

be

For example,

assume the economic

the

of

number

economically

repaired

fastener holes with crack sizes equal

Then

x1 ).

size

specified

an

is

economic

(i.e.,

to

or

life

are

discussed

further

number of

than

format
2.3.

in Fig.

the exceedance probability.

holes

greater

analytical

"quantifying economic life is presented

2.3,

fastener

for

In Fig.

Various aspects of
in

the

following

subsections and elsewhere [11-22].

2.3.3.1

The

II"

defined

Economic Life Definition

economic life of an aircraft structure is currently

in

widespread

qualitative
damage

not repaired,
operational
"widespread
defined

for

terms:

which

could

is uneconomical to repair and,

cause

readiness"

occurrence

"...the

functional
(1-3].

damage"

and

each

aircraft

problems

Acceptable

"uneconomical

of
if

affecting
limits

repairs"

must

for
be

design ard such limits must be

approved by the Air Force.

A quantitative definition of economic life is not given

in this handbook.
specifyizg
"In any case,

However,

economic

guidelines

life criterion (Ref.

are

presented

for

Section 2.3.3.4).

quantitative criteria for the economic life

of

2.8

4.,..%

>
-

Al
w
N

I SERVICE

P'O.05

LIFE

~
FOR

.ECONOMIC

LIFE

P- 0.5O

FLIGHT

Fig.
,C,

2.3

HOURS

Analytical Format For Economic Life

--

-2.9

BASIS
JUDGING

aircraft

structures

should

based on specific aircraft

bc

requirements and the user's acceptable

for

limits

aircraft

performance and maintenance costs.

2.3.3.2

Economic Repair Limit

The "economic repair limit" is

the maximum crack size in

a structural detail that can be economically repaired.

limits

easily be defined from geometric considerations

can

for fastener holes but such limits

are

difficult

more

define for structural details such as cutouts,

be

governed

by

the

to
etc.
hole

fastener

largest radial crack that can be

cleaned-up by reaming the hole to

the

fastener

next

size

0.03" to 0.05" radial crack).

(e.g.,

The

objective

of

the

presented in this handbook is

number

of

durability
to

must

analysis

analytically

method

predict

impairment.

The

define the uneconomical repairment or functional

impairment crack size for the details to be included in

extent-of-damage
considerations

assessment.
such as

Such

crack

structural detail

-.

the

sizes depend on
type,

location,

2.10

the

structural details with a crack size which would

cause an uneconomical repair or functional

user

fillets,

the economic repair limit for a

For example,
may

Such

, .7

-4-t

accessability,

repairability,

inspectability,

repair costs,

etc.

details

Structural

one

contain

more cracks.

or

structural durability is concerned with the largest

However,
in

crack

may

each

r'

detail

may

which

repair or part

require

2.3.3.3

'

replacement.

Extent of DamaQe

4,

The

extent

of

a quantitative measure of the

damage is

number of structural details containing cracks

specified

crack

that

exceed

size limits as a function of service time.

Structural maintenance requirements and costs depend on

number

of

structural

"durability" of the

requiring

details

depends

structure

on

repair.

the

the
Th

extent

of
a

damage for the population of structural details in a part,

component,

or airframe.

extent

The

analytical tools
damage

of

provided

predictions

can

damage
in

provide

be

this
the

predicted
handbook.

basis

for

using

the

Extent

of

analytiua~ly

ensuring design compliance with the governinq eccnomic

criterion.

4%%

2.11
V..

life

"2.3.3.4 Formats For Economic Life Criteria

Two

analytical

formats

for

defining

"cE.onomic life criteria are recommended:

crack exceedance and (2)
cost

[14-17].

handbook

The

can

be

cost ratio:

analytical

(1)

quantitative

probability

of

repair cost/replacement

tools

described

in

this

used to predict results in these formats.

Various aspects of each format for a

quantitative

life

including examples and

criterion

guidelines

(Ref.

2.3.3.4.1

are discussed below,

Fig.

2.3).

Probability

of

Crack

probability of a crack occurrence which

specified

crack

crack exceedance."

the

durability

economic

size is

analysis

For example,

exceeding

crack

size

is

larger

The
than

in
x,

a fundamental output of

methodology
Fig.
at

2.4

described
the

r is

in

this

probability

initial

crack

at

Crack size rankings in the respective distributions

r.

for two different times are

size

of

represented by the

cross-hatched area under the crack size density function

t

referred to as the "probability of

This quantity is

handbook.

Exceedance.

x,

at

t -

namely,

the

crack

r has the same rank (or percentile) as the

crack size at y 1 i(T)

exceedance

preserved;

can

be

at t = 0.

used

to

expected repairs in a given service

2.22

The

predict
interval

probability

of

the number of
(15,172.

Ic

p0,r)

-%

")

*P

IefT) > x

0.

CRACK
SIZE

S~vii

SERVICE

CRACK GROWTH
MASTER CURVE

(T)

.,I

EF

OISTRIBUT!0ON

TIME

Fig.

2.4

Durability Analysis Approach

2.13

I'_-:
. ,. -,,'-- : ---",".- . . - ,. . . -

- , .

..-

.-.- --

- .-

, ,.

.-

- . -

. ,

. ,. -..

- .,.,--,

- ,

- ,

basis for judging airframe durability and

provides

also

for analytically demonstrating design

compliance

with

the

c'overning criterion for economic life.

Another

explanation

exceedance

concept

will

structural

detail,

in

of

the

now
a

be

probability
given.

of

crack

Each

common

group of details having a common

stress history,

has a single dominant

are

variable and their "initial" size depends on

random

crack.

Such

the manufacturing quality for each structural

population

of

considered.
fatigue

crack

sizes at time r

detail.

The

sizes depends on the group of details

For example,

quality

cracks

in Fig.

2.4

assume

the

initial

dibtribution and the distribution of crack

are

for

fastener

100

holes

(i.e.,

the

population of details).

The

probability of exceeding crack size x, at time z is

represented by the cross-hatched area under the

probability

density

Suppose the

of

crack

sizes

shown

probability of crack exceedance

in
is

Fig.
p(ir)

means that on the average 5% of the details

,N

measure

crack size a xl at time r.

of

distribution,

the
the

extent

of

damage.

0.05.

(e.g.,

fastener holes)in a part or component would be

have

p(i,r) is
Using

This

5% of the

expected

to

a fundamental
the

binomial

extent of damage foc different groups of

2.14
e-i

2.4.

'6i

details can be combined to quantify the overall damage tor a

parw,

a component or airframe.

The

allowable

crack

exceedance

recommended for quantifying economic

handbook

provides

crack exceedance,
demonstrating
durability
tailored

specific

design

acceptable

etc.

values

life.

are

compliance

requirements.

criterion

Although

not

with
Such

specific

aircraft

limit

for

requirements/costs,
readiness,

one

this

guidelines for quantifying the allowable

design

for

is

presented

the

Air

values

structure

The allowable

be

and the user's

maintenance

impairment,
crack

Force's
must

structural

functional

for

operational

exceedance

criterion

for economic life design compliance shall be approved by the

Air Forcp.

The
depends

allowable
on

crack exceedance for a part or, component

several

accessability,

factors,

including:

inspectability,

criticality,

repairability,

cost,

operational readiness, acceptable

risk limits,
etc.
For
example, an expensive fracture critical part may be embedded
into the wing under-structure.
accessible

and

it

is

The

difficult

Suppose the bolt hole for this

part

part
to

is

inspect
governs

not

readily

and repair.
its

economic

life. Then a lower allowable crack exceedance may be desired

2.13

than for an equally critical part that

"and inspectable.

For

example,

is

an

more

average

exceedance at 1.2 service lives might

be

accessible
of

2% crack

suitable

in

the

first case and an average of 5% might be appropriate for

different circumstances
An

example

icriterion

for

the

is as follows.
component

is

reached

details (e.g.,
reached

crack

of

size

(limiting case).

each

when

percent
cutouts,

of

of

part

or

the structural

fillets,

etc.)

The limiting crack size depends

detail,

which

have

on

the

the economic repair limit,

and

would

cause

functional

impairment

Structural safety or damage tolerance must

not be compromised.
for

crack exceedance

size a a specified limiting crack size at

structural

the crack

of

The economic life

fastener holes,

1.2 service lives.

type

probability

Also,

detail

type

the specified limiting crack size

should

account

for

inspecticn

capabilities and requirements and operational readiness.

The

economic

life

criterion described (i.e.,

exceedance)

can be used to demonstrate economic life

compliance

analytically and experimentally.

5% crack
design

The analytical

tools presented in this handbook can be used to quantify the

extent of damage in terms of crack exceedance.
Therefore,
given the criterion foc economic life, design compliance
can
be analytically assured.
Experimental compliance can be
determined based on the
results
of
the
durability
demonstration test

results.

2.16

'=

2.3.3.4.2

Repair

Cost/Replacement

Ratio.

Cost

cost is another recommended

ratio of repair cost/replacement

criterion for quantitative economic life.

reached.

the economic life is

economic life is reached when

the

from

the

aircraft

Allowable

components.

In other words,
ratio

cost

is

user

ratios

acceptable allowable cost

cost

at

to
the
a

1.2 service lives).

specified service life (e.g.,

Input

cost

when

For example,

the cost to repair a part or component exceeds the

replace it,

The

for

needed

to

define

parts

different

or

ratios cculd be specified tor

pairticular design situations and user goals.

closts

Repair

are

proportional

structural details (e.g.,

Spervie

time.
to

the

number

of

fastener holes)
requiring
repair
eime.analytical
trols

described in this handbook

number

to

--an

be

used

to

quantify

the

of details requiring repair as a function of service

Although specific repair cost data may

obtain for different

way vary,

the cost ratio

be

difficult

circumstances and replacement

can

be

estimated

using

costs

assumed

repair and replacement costs.

The

cost

recommended

ratio
for

criterion

demonstrating

2.17

Al.

for

economic

design

life

compliance

is not
unless

acceptable
is

cost data

are available.

tradeotf

recommended for evaluating user design

affecting

the

llfe-cycle-cost

this criterion

However,

the

of

options

airframe.

analytical tools described in this handbook can be

used

The

to

evaluate the life-cycle-cost design tradeoffs.

"."

'V'

24

'"

2.18

Ij
V.'
'...9-

,.

..

i-

.-

'

SECTION

III

SUMMARY OF THE DURAPILIrY ANALYSIS METHOD

3.1

Essential

elements

INTRODUCTION

and

equations

of

the

durability

analysis method are summarized in this section.

Details

of

"the approach and implementation procedures are described in

*0

Sections IV and V and elsewhere

GENERAL DESCRIPTION OF THE METHOD

3.2

The

basic
is

methodoloav

function

objective
to

of

[14-21,23].

quantify

of

the

the

extPnt

holes,

fastener

expected to have a
specified
of

damage

exceedance.

of

crack

of

cutouts,
whose

is

value at a given service time.

is

represented

by

as

details

lugs..
greater
Hence,

probability

etc.)
than

of

4.'

crack

The durability analysis results quantitatively

as

time

basis for aialytically assuring that the

4'

4'

the extent

describe the extent of damage as a function of service

"and serves

The extent

structural

fillets,

size

analysis

damage

service time for a given aircraft

of damage is measured by the number

(e.g.,

durability

3.1

economic life

the

of

structure

design

the

exceed

will

service life.

The

durability

anaiysis

includes two essential steps:

(1) quantify the initial fatigue quality of

structural

the

details considered, and (2) predict the probability of crack

exceedance

using

applicable

design

the

fatigue

initial

the

and

quality

(e.g., load spectrum, stress

conditions

levels, percent load transfer, etc.).

Essential elements of

the durability analysis method are described in Fig. 3.1.

The

durability

analysis

method has been developed and

holes

demonstrated for fatigue cracks in fastener

[15-21).

However, the basic approach theoretically applies to fatigue

cracks
cutouts,

in

other
lugs,

structural
etc.

details,

Further

such

research

as

is

evaluate and demonstrate the method for details

fillets,

required
other

to
than

fastener holes.

3.3

1.

Fatigue

ASSUMPTIONS AND L1IMITATIONS

crack length, measured in the direction of

crack propagation, is the fundamental, measure of

damage.

3.2

durability

II

Ii.i,

aII-

1i.

"2

a
-

'

oo a

(10
..-

I
do

*-

t)

O's)

I7'

'w
1~1

I"i,-- i
.

"

rww

.- '

IFI

..L

'v7-

('

(.

',

4%r

,.

.I,
3. 3

.4-

r-4

ini

INI

4-J

It~i

L6
d6 ar'r

r4

-U1-

r-4

4H

3.40

'4 ,

2.

Each detail

etc.)

lug

(e.g.,

fastener hole

in an aircraft structure has a

fillet

cutout

single

dominant

fatigue crack which governs the durability of the structure;

the size of such a

crack

is

considered

to

be

random

variable.
~4

"4

3.
M

The

largest

fatigue

relatively small (e.g.,

crack

in

each

detail

5 0.05" corner crack in

fastener

hole) and such cracks are statistically independent.

the growth of the largest
significantly

affect

the

crack

in

growth

neighboring details and vice versa.

distribution

can

be

for different details,

4.

An equivalent

back-exLrapolating
crack

growth

law.

one
of

Hence,

does

not

the largest crack in

Therefore,

the binomial

used to quantify the extent of damage

parts,

components or the entire airframe.

initial flaw size can be determined by

fractographic results
An

EIFS

is

describing the IFQ for a given deta.l.

not necessarily an actual initial crack

5.

detail

is

using

rational

mathemati.cal quantity
As such,

the EFS is

size in the detail.

Different EIFS distributions can be developed using

the same fractographic data set and fractographic crack size

range by using different crack growth master, curves.

3.5
I.

..

-.

-.

6.

EIFS

The

wr

-.

distribution

is

'.r

defined for a selected

0.020"

fatigue crack size range (e.g.,

0.050"

crack

in

fastener holes).

"7.

An EIFS distribution can be grown from time zero to

a given service time

using

single

deterministic

crack

growth curve.

8.

A su:.table EIFS distribution for durability analysis

data

can be determined using selected fractographic

for

etc.):
the

load

given

spectra,

% load transfer,

(i.e.,

stress level,

The derived EIFS disr.ribution can be used to predict

extent

of

damage

for

different load spectra,

levels and % load transfers other than

those

stress

reflected

in

the fractographic data base.

9.
can

be

A suitable service crack growth master curve (SCGMC)

developed

for

specific

analysis

durability

conditions.

3.4

The

IFQ

model provides a means for quantifying the IFQ

"of structural
cutouts,

INITIAL FATIGUE QUALITY MODEL

lugs,

details
Etc.)

(e.g.,

fastener

fillets,

holes,

IFQ

susceptible to fatigue cracking.

can be represented by either a TTCI distribution

or

by

an

""',-

--.-

-.

*..

'4<

1,

EIFS

distribution.

The IFQ distribution is

used to predict

the extent of damage or probability of crack exceedance

for

multiple structural details subjected to fatigue loading and

environment.
notations

Essential

are

shown

elements

in Fig.

3.2.

of

the

model

and

The model equations are

summarized in this section and details are

The

IFQ

time-to-crack-initiation

distribution for a reference crack size,

given

(TTCI)
aO,

elsewhere

I'

cumulative
is

represente'

by the three-paramater Weibull distribution as follows:

F(t) P[T'-t]
P

where:

1-(-1

TTCI

Shape Parameter

Scale Parameter TC

S=Lower
Bound of TTCI

3.I7
-"

~3.7

FT(t)

--TTCI

Distribution
dFT(t)

CRACK
SIZE

\--

EIFS MASTER CTJRV

/
.Y-

-t
I

da()

Ohs(t)I h

EIFS

-z

__________
_______________________________________________________________

TIME
faw(x)
dx

Fig

3.2 Thit.al

Fatigue Quality Model

.p.

3.8

41

9A

The

EIFS cumulative distribution,

from a transformation of Eq.

3--l

Fa(0)(x),

and

the

is obtained

following

crack

growth law in the small crack size region.

Q[a(t) I

da(t

where:

Q ,b

(3-2)

Parameters depending on loading spectra,

structural and material properties

Eq.

3-2

is

used

ability to fit

The

model

its

simplicity and general

EIFS cumulative distribution,

compatible

distribution Fr(t)

IFQ

of

crack growth data.

derived

statistically

because

: i.e.,

with
FT(t)=I

equations

variations of Eq. 3-2,

i.e.,

have

the

F a(

TTCI

)(x),

is

cumulative

Fa(0)(x).

been

developed

bol and b=l [15,163.

for

two

In

this

section,

the IFQ model equations are summarized for both b>l

(Cuse I)

arid b=l (Case II).

durability

analysis.

Hence,

Case

II

is

recommended

for

the implementation procedures

presented in this handbook are tailored for Case II.

3.9

for
holes

Several

IFQ

both

straight-bore

(16].

be

less

parameter studies were performed

and

countersunk

fasteners.

For the

fractographic

the computed "b value" in Eq.

than

1.

theoretically

Since

possible

negative
when

b<l,

3-2

EIFS
b

referred

to

data

Section III of Volume VII [16]

sets

was found to
values

values

recommended for the present durability analysis.

is

fastener

These studies were for 7475-T7351 al~uninum and

clearance-fit
considered,

model

are
1

are

The reader
for further

details.

3.4.1

IFQ Model Equations for Case I (b > 1)

"General Crack Size -- Time Relationship

"Interatina Ea.

a(t

3-2 from ti to t--

[a(t

(3-3)

2
where:

one obtains the following,

a(tI),

a(t

) = Crack size at time tj and t 2,

respectively
c = b-l

,Q = Crack growth model parameter

3.10

'o-

--

EIFS MASTER CURVE

Let
'

crack

tl=O,

t 2 =T

and

a(T)-a

3-3 becomes

Then Eq.

initiation.

EIFS

a(O)

(a

reference crack size at

C+ cQT) -1/c(3-4)

EIFS Upper Bound Limit

denoted by xu,

The upper bound of a(O),

Eq.

is obtained from

3-4 by setting the lower bound E for T,

x=

-/c35)

(a 0 C + cQ)

TTCI Lower Bound Limit

The

lower bound

E of T can be expressed in

terms of the

upper bound x,, of a(O),

c-(x
Q=

a0

31

3.11

I4" S.
"

"

aO-Xu

(3-6)

EIFS Cumulative Distribution

The
,..

distribution of a(Q)

given by Eq.

can be obtained from that of T

3-. through the transformation of

Eq.

3-4

as

follows,

Fa(0)(x)

exp

x
-

[0]

-C cQ

S1.0

x
'

(3-7)

!x_
U

or

Fa(O) (x)=

exp

(3-8)

1.0

p~

- x

-u

x~x

where,

(3-9)

Weibull scale parameter for TTCI based

on the TTCI's for a given fractographic data set

in which only the fractography for the largest

fatigue crack in any "one" of t fastener holes

per test specimen is used.

3.12

-*

'

A.5.

of equally-stressed fastener

t -Number

holes per test specimen

in which only

the largest fatigue crack in any "one'

hole per specimen is

included in the

fractographic data set.

%I
*1

Sm Weibull shape parameter for a given

TTCI data set.

3-9 are given in Section 4.5 and Ref.

Details for Eq.

IFQ Model Equation For Case II

3.4.2

16.

(b=l)

General Crack Size-Time Relationship

a(t)

a(t

) exp[ -Q(t2-t

31-0)

EIFS Master Curve

a(0)

EIFS = a 0 exp (-QT)

(3-1l)

EIFS Upper Bound Limit

exp

xI

(3-12)

(-Qc)

3.13
,

?,.-

p.4

TTCI Lower Bound Limit

-.

InI

(ao/x

3-13)

ETFS Cumulative Distribution

exp

FaO (x)

x2x(

(3-14)

=1.0

;xzxu

or

F(

)(X)

exp

--

Q-

; 0<xX
,

)
(3-15)

=1.0

;xzx
U.

"In Eqs.

3-14 and 3-15,

3 "

1/as
(.') (Ref.

Eq.

3-9 and Section

4.5 for further details).

"

O.S

~3.14

3.5

The

DURABILITY ANALYSIS

gcne'al

procedure

PROCD1JR,:S

for implemanting the dLu ,ability

analysis method developed is described and discussed

1.

Decide

predicted for
"parts?,

what
(e.g.,

level
a

the

single

component?,

extent

of damage will be

part?,

complete

below:

several
airframe?,

different
fleet

of

airframes).,

2.

Determine which structural details will be included

in the durability analysis (e.g.,

fillets,

3.

lugs,

fastener

holes,

cutouts,

etc.).

Determine the IFQ or EIFS distribution for each type

of structural detail to be included

assessment.

in the extent of

Use the model shown in Fig.

damage

3.2 and applicable

tractographLc results for a selected crack size -&ajcge

0.020"

0.050" crack in fastener holes)to define the EIFS

distribution expressed in Eqs.

Q

and

3-14 and 3-15.

Determine

QO using the procedures described in Sections IV and

and Ref. 16.

Eq.

3-12.

The x, selected should be consistent

Theoretically,
structural detail.
only

(e.g.,

the

IFQ

However,

with

model developed applies to any

the

model

has

for fatigue cracks in fastener holes.

been

verified

Further work is

3.15
9",

%" ", - ," " " " " "-m ,

".-

," % ,

' "" " % - .

" " " " "%" """ ", '

%<', " %

'

-" -

required to develop
procedures

to

details such as,

fillets,

For each part,

4.

etc.

cutouts,

component,

details by type .nto m

stress

group the structural

etc.,

where

regions

the

maximum

in each region may reasonably be assumed to be equial

stress

for every location or detail (e.g.,

For each stress region,

5.

the corresponding EIFS value,

Lize
If

and

fractographic data for

appropriate

acquire

specimens

test

fatigue

suitable

x1

fastener hole).

ith stress region,

yUi(r),

that

't
determine

grows to a crack

at service time r as illustrated in Fig,

)?

3.3 (16].

applicable fractographic data are available for different

stress

levels and fractographic data pooling procedures are

used, the crack growth rate expression in

.can be integrated from a(O)

i=ia
obtain

where

3-16,

y 1 i(r) to a(r)

x, to

(r) in Eq. 3-17.

Y1 i()

Eq.

= xI exp (-Q:-); b. =1

(3-17)

suitable fractographic results are available for the

design conditions (eog.,

load

spectza

load

transfer.,

3. 1"
A.

CRACK
SIZE

_* .

o FT)
([(T)>x1 ]

Xu
OLpT)

SERVICE CRACK GROWTH MASTER CURVE

ii-.i...,

v i !T )

"LIMiti

Fetigs L ility

DOIbl.r S(IIFI
)

Fig.

3.3

"

TIME

Growth of EIFS Distributioni a& Function of Time

3.].7

etc.), Qi may be expressed by a power function

stress level,
as follows.

(3-18)

in

ing spectrum,

andi

the load-

in

the naximum applied stress

c" is

3-18,

Eq.

E and y are constants to be determined

from.

the available fractc-raphic data.

suitable fractographic results are not available,

If

analytical crack growth program can be used to

crack

the

size

range

o:

the

interest.

However,

the analytical crack growth program should first be

"tuned"

or curve-.fitted to the applicable EIFS master curve

is

before it

to

used

predict

the

growth

crack

damage

the crack growth parameter bi and

Then,

a(t).

accumulation

r'

over

growth

crack

predict

an

Q. can be obtained by fitting Eq. 3-16 to predict

the crach

size a(t) as a function of service life t.

i'.,

Compute the probability of crack exceedance for each

6.

stress region, '

IS
using Eqs.

p(i,T)

p(ir)

i.e.,
"1

3-15 and 3-17,

=1l

-'

exp

= P[a(T)

> xl]

1-Fa (0) [IV 1i (l*

wjth the result

(X/Yl(T)

(3-19)
p(i

= 00

3.18

...--

in which 6 is

7.

defined by Eq.

The

average

3-9 (ref.

number

standard deviation o(i,r)

in

of

Section 4.5).

details

N(i,r),

the ith stress

and the

region

with

crack

size greater than x 1 at service-time r are determined

using

the

binomial

distribution

and

are

expressed

as

follows:

N(iT)

oN(iT)

which

p(i,

{Ni

in

N.

(3-20)

p(iTC)

The average number of details with

size

exceeding

L(r),

and its standard deviation,

using Eqs.

(3-21)

Ni denotes the total number of details in the ith

stress region.

x,

crack

at service time r for m stress regions,

aL(r),

can

be

computed

3-22 and 3-23.

"L[(T)

m
-.

JL~T)
L

N(i,T)

2_a(T
N

3.19

4-.%

[l-p(i,T)j(2

(3-22)

i=1i

(3-23)

"Equations3-22 and 3-23 can be used to quantify the extent of

damage for a single detail,
component,

The
-. "

reference

part,

crack size for crack exceedance,

different

"limiting crack

size

detail

that

can

types
be

according

E(r)

ZC'L(r),

where

is

xl,

can

to

the

economically repaired.

and lower bcunds for the prediction can

using

or an airframe.

be defined for

*"Upper

a group of details,

be

estimated

the number of standard

deviations,
23

are

a (r), from the mean, E(r).

Eqs. 3-20 through 3L
valid if cracks in each detail are relatively small

and the growth of the largest crack in each

affected by cracks in neighboring details.
growth

accumulation

independent

each

for

detail

detail
Hence,

is

is

the crack

statistically

[153.

3 .

ii[Z

"N
i

3.2 0

''

not

SECTION IV

INITIAL FATIGUE QUALITY DETAILS

4.1

K'

Initial
the

INTRODUCTION

fatigue quality (IFQ),

"cornerstone"

developed.

for

the

or EIFS distribution,

durability

analysis

is

method

Much has been learned about the characteristics

and traits of an EIFS distribution during the course of this

program

(14-23).

The

purpose

of

this section is to (1)

discuss the current understanding of the EIFS distribution

based on fastener hole experience,

(2) present guidelines for

acquiring the data needed to quantify the EIFS

and

(3)

describe

calibrating

the

and

IFQ

illustrate

model

the

parameters

distribution

procedures
from

for

available

fractographic data.

The

IFQ

model

evaluated further
fatigue

cracks

experience is
the

EIFS

conditions
transfer,

described
using

in

in

Section

existing

fastener

III

fractographic

holes

(e.g.,

should

be

data

for

24-271.

This

needed to further advance the understanding of

distribution
(e.g.,

load

for

different materials and design

spectra,

fastener hole type/fit,

4.1

stress
etc.).

level,

load

-.

-, -

. .. .

--.

'W

1.
-

fatigue

to

flaws

crack

in

in

defines

a part,

equivalent

an

in

exist

to

the

rather

than

is

EIFS
an

flaw is

details

hypothetical

prior to service.

An

crack which would

point

in

time.

a detail.

are

(fractography)

crack nrowth analysis

As

quantity"

"mathematical

flaw in

initial

actual

using

with

flaw size (EIFS)

initial

a detail

strictly

cracks from fatigue tests

backwards

details

component or airframe

an actual crack size at a later

such,

initial

the

or

size of a hypothetical

the initial

result

(IFQ)

An equivalent initial

assumed

EIFS is

W--- v

The IFQ for a group of replicate

by

represented

distribution.

-.-

quality

initial

prior to service.
is

of a structural detail

manufactured state
respect

VWTL-

EIFS DISTRIBUTION

4.2

Initial

'

Observed

extrapolated

to estimate their

EIFS.

time

The
type,

for

required

which

can
a

following

(TTCI).

reliably

be

fatigue

a0

An

test.

in Section III,

EIFS

the

an arbitrary crack

frsctographically
EIFS

population

crack

is

distribution

..

* ..

.-.

-7..

for

determined

4.2

distribution
a

Using the IFQ model described

replicate details.

of

is

observed

quantitatively describes the EIFS

group

of whatever

to become a fatigue crack of size a 0 is defined as the

time-to-crack-initiation

size

an initial defect,

4-

..

--

-%

by

S.

the

coupling

%'-..
.

'

for

''

'.W

distribution with a deterministic crack

TTCI

The IFQ model

growth law.
tool"

. -,

is

the

quantifying

"mathematical

convenient

EIFS

is

which

distribution,

statistically compatible with the TTCI distribution.

.4

EIFS

An

to

fitting the IFQ model parameters

data

for

bolt load

fastener

type/fit,

the EIFS distribution has been established,

of

fatigue

cracks

cumulative distribution of TTCI,

be
(eg..

analytically

workmanship.

load spectrum,

Necessary

basic

this:

Once

cumulative

the

given time and the

a0

can

EIFS

As

is

inherent

property

manufacturing/assembly

of

of

such

techniques,

EIFS should be independent of

such,

stress level,

traits

an

etc.).

and % bolt load transfer.

the

EIFS

distribution

follows:

4.3

i-

is

for a given

FT(t),

stress level,

factors as the material,

and

The

predicted for different service conditions

load spectrum,

Intuitively,

at

stress level,

etc.

the durability analysis approach

distribution

fractogaphic

observed

load spectra,

material,

transfer,
of

premise

given

can be established by

F a(0)(x)

distribution,

aie

as

The

EIFS

observed

the

predict

accurately

must

service,

when; grown forward during

distribution,

cumulative distribution of crack sizes at any time.

it

Alternatively,

must predict observed TTCI values

for any crack size a 0 .

An

if

specimens,

"tested using

set

that

implies

of

on

level.

identical

divided into two or more

different

depend

not

spectrum and load

i.e.,

subsequent service,
This

should

distribution

EIFS

groups

test
and

stress levels or spectra,

This is

should produce the same EIFS distribution.

called a "generic" EIFS distribution.

"The EIFS distribution is

necessarily

the

not:

distribution

defects or cracks in the material

unique.

fairly

wide

EIFS distribution is

ubserved

flaw

distributions

range reasonably well.

obtained

4.4

<"

initially.

In fact many different EIFS distributions

can predict the same

over

actual physical

of

- -

using

Each

different

growth

crack

shown in Ref.

An example is

model.

2Ii

16.

3-8 and 3-15),

Equations for the EIFS distribution (Eqs.

growth

Other equations

3-2).

(Eq.

law

on

based

are

presented in this handbook,

given

for Fa( 0 )(x)

also be developed for different crack growth laws.

the

general and the user can

analytical crack growth program,

growth model,
data base,

his

to

method

the

adapt

could.,

However,

approach proposed herein is

analysis

durability

crack

quite
crack

fractographic

etc.

4.3

Fractographic

TEST/FRACTOGRAPHY GUIDELINES

(i.e.,

data

or

needed to define the IFQ

details (e.g.,

crack

fillets,

guidelines

in

recommended

current

understanding

holes.

As

such,

this

these

section

fatigue

for

lugs,

The test and

used in the durability analysis.

for

distribution

EIFS

fastener holes,

size versus time) are

guidelines

cutouts,

those
etc.)

fractography

are based on the

cracking
should

in

fastener

be considered

preliminary.
I~

Further

work

is

guidelines for details

required to develop test/fractography

other

than

fastener

holes

(e.g.,

4.5
I,

I.

4k~I

J.1

-.

fillet.s,

--

-.

'*-

V"

. -

~~

suitable

Also,

etc.).

lugs,

cutouts,

specimens should be standardized for generating the

cracking

-!
*

.-

test

fatigue

data needed for each detail type to be included in

the extent of damage assessment.

Test Guidelines

4.3.1

The following guidelines are for fastener holes:

1.

fatigue cracking data,

design

technique/tools,
stress level,

2.

-''N

The

fatigue

details,

transfer,

etc.).

design used should provide a maximum

specimen

for

single

test.

For

example,

data can be obtained for multiple details

in a single specimen.
multiple

load

percent

environment,

load spectrum,

cracking

applicable
preparation

hole

material,

fastener type/fit,

amount of information

to generate the

used

should account for the

it

(e.g.,

variables

is

specimen

test

Whatever

To

each

comparable stress level.

justify
detail
And,

using
should

specimen

be

exposed

the largest crack

in any

with
to

a
one

detail smould not significantly affect the crack growth in a

IA

neighboring detail and vice versa.

Structural details

must

be spaced far enough apart so that they will crack independently.

4.6
.,

'.n

Single dog-bone and reverse dog-bone type specimens

3.

bone

specimen

type

fatigue
~

shown in Fig. 4.1 was used with one or

This type of specimlen can be

two holes.

cracking

to

used

generate

data for no load transfer cases.

performed during Phase II suggested that IFQ is

of the percent load transfer (161.

Studies

independent

Further work is required

to justify using no load transfer specimens

dog-

single

The

were successfully used for this program.

to

the

define

IFQ for fastener holes with different percentages of load

If no load transfer specimens can be justified,

transfer.

then the specimen types shmown in Figs. 4.1 and 4.2 could be
promising for economically generating the fatigue cracking
data needed.

34.

The

no

load

transfer

contains two fastener holes.

a

positive

way

to

assure

specimen shown in Fig. 4.2

This specimen design

independence

the

cracking in the two fastener holes.

of

specimen

occurs in a single hole.

holes

when

the

If there

are

fatigue

be

and

TTCI

in

in

test is stopped, fractography canm be

results

described in Section 4.5.

occurs

can

cracks

performed for both holes or for the largest crack in

hole

fatigue

tested to a specified service time or untill failure

fatigue

both

This

provides

can

either

be scaled using the procedure

If an

observable

fatigue

crack

only one hole when the test is stopped, the test

4.7

'@I

'0

o0

Fig. 4.1.

No-Load Transfer Specimen With Multiple

Details

'-'.-2

Fig. 4.2

Two-For-One No-Load Transfer Specimen Design

4.8
*A

--.-

.A,
7''.4.

I ."

can be continued for the uncracked

specimens

as

follows:

by

hole

specimen.

.-

reworking

the

cutoff the test sections containing

the hole with a crack from the thick part at the

the

"

middle

of

The thick part then becomes the lug end that

can be used to continue the testing until

visible

crack

occurs in the remaining fastener hole.

5.

The no load transfer specimens should be tested with

the desized type cf fastener and fit, in the test holes.

6.

The reverse double dog-bcnc specimen design concept

shown in Fig.
program,

4.3

seemed

particularly

to

work

fairly

for

this

the "15% load t.ansfer design".

The

transition between the lug end and test

smooth

and gradual.

Also,

should allow an adequate

well

section

should

be

the specimen length and geompetry

range

of

axial

deformations

to

obtain the desired shear load transfer through the fasteners

and mating holes.

7.

The number of specimensrequired for testing depends

on factors,
or

such as:

multiple hole),

(e.g.,

no.

diameter,

of
bolt

environments,

{..

(1) type of specimen used

(2)

design variables to be accounted for

materials,
torque,
stress

(single hole

fastener

type/fit,

manufacturing variations,
level,

4.9

load

spectra,

fastener
fretting,

etc.).

(3)

'--

1.~000" DIA 1 0.002 7rYP

2.0 R TYP7

4.00"

N.

314"1
3

4,1

~~Fig. 4.3

RvreDouble Dog-Bone Seie

Load Transfer)

4.10

1%Bl

confidence

level

parameters,

(4)

results,

cracking

fatigue

in

scatter

expected

model

IFQ

calibrated

for

desired

etc.

The schedule and budget also influence this

Room

temperature

decision.

8.

quantifying

IFQ.

the

are

tests

Effects

of

recommended

for

can

be

environment

accounted for in the service crack growth master curve.

9.

For a given load spectrum,

test replicate specim-ens

using at least two different stress

levels

are

levels are used,

low

stress

only two stress

tests should be conducted at a high

and

Select stress levels that will cover the

stress level.

range of expected design stress

can

If

affordable.

if

preferred

Three

levels.

levels.

information

This

be used to define the s2rvice crack growth master curve

levels

for different stress

without

analytical

an

using

crack growth program.

4.3.2

1.

Guidelines for Fractographic Data

Select a minimum crack size to be read,

is consistent with the

reading

equipment

and

capabilities
technicians.

of

the
Budget

4.11
V:.
L

such that it
fractographic
and schedulc

j'

should also be considered.

-ize

The minimum fractographic

to be read depends on the smallest crack size,

which

crack

example,

exceedance
crack

predictions

size

range

will

be

crack
x1 , for

made.

For

of 0.020" - 0.050" would be

reasonable for assessing the durability of fastener holes.

2.
in

Take advantage of the g6scaling technique described

Section

"requirements.
fastener

holes

4.5

to

minimize

For example,
and

then

fractoyraphic

reading

use test specimens with multiple

fractographically

evaluate

the

largest crack per specimen in a given fastener hole.

3.

Use automated crack monitoring techniques as much as

possible to minimize fractographic

acquisition costs.

Also,

automatic storing ot the fractographic re uults directly into

the computer can minimize the time and

costs

for

plotting

results and for calibrating the IFQ model parameters.

4.4

PROCEDURES FOR CALIBRATING '."HE IFQ MODEL PARAMETERS

Suggested

procedures

"parameters are described

including guidelines.

for
and

determining
discussed

However,

it

is

the

in

section,

optimization

the IFQ model parameters using pooled fractographic data

4.12

p.

this

model

felt that further work

and experience is needed in the evaluation and

of

IFQ

.W-1

for

materials

before values can be tabulated for different

design purposes.

During the course of this program various techniques and

their

variations

parameters

for

have

determining

been

investigated.

studies were performed in Phase II

"and countersunk

consistent

fastener

Model

for

both

holes [Ref.

16,

IFQ

model

calibration

straight

bore

Appendices A and

The lessons learned from these studies are reflected in

B1.

this section.

4.4.1

The

Generic Nature of IFQ and Data Pooling Concepts

purpose

of

this

section

is

to:

(1)

discuss

fractographic data pooling concepts for determining the EIFS

"distribution,

(2)

explain

the "generic" nature of the IFQ

model

parameters

pooled fractogcaphic

results for

"distribution and (3)

describe how the IFQ

can

using

be

determined

different data sets.

The

IFQ

model parameters in Eq.

for. fastener holes using

for

applicable

one or more data sets.

replicate fatigue tests (e.g.,

4.13
*A

fractc'j-raphic

results

A fractographic data set refe-s

to the fractographic results (i.e.,

for

3-15 can be determined

values)

same: material,

specimen

a(t)

versus

"

."

hole drilling procedure,

geometry,

load spectrum,

results can be used to determine the TTCI and

behavior

for

fastener

holes.

determine an appropriate

crack

growth

This information is used to

distribution

IFQ

stress level,
Fractographic

etc.).

fastener type/fit,

% bolt hole transfer,

*.***
" ". " ".

for

durability

analysis.

Fractographic

results

for

different

data sets can be

"pooled" to quantify the IFQ model parameters

analysis.

Pooling

data sets

(e.g.,

drilling

sample
(xu,

the fractographic results for different

same

procedure

stress levels)
size

is

levels,

An
the

material,
but

can

3-15,

because

this

increases

the

Therefore,

thC

resulting

EIFS

be applied to different design stress

etc.

EIFS distribution is

Theoretically,

and hole

available for fitting the IFQ model parameters

load spectra,

material

fastener type/fit,

different load spectra and

recommended

a and QO) in Ect.

distribution

for durability

and
the

of design variables,

"generic"

if

it

depends only on

manufacturing/fabrication

processes.

EIFS distribution should be independent

such as load

percent bolt load transfer,

spectrum,

environment,

stress

level,

etc.

4.14

-I1

::.

. ..

. .

Fractographic
fastener

results are used to determine

holes.

influence

of

spectrum,

Since

the

specific

stress level,

fractography

fatigue

reflects

the

test variables (e.g.,

percent bolt load

"the derived IFQ distribution may

the IFQ for

transfer,

load

etc.),

not be strictly "generic".

However,

the fractographic data pooling procedures described

herein

can

be

used

to

determine

an

appropriate

IFQ

distribution for practical durability analyses.

TTCI distributions for different fractographic data sets

(i.e.,
fit

u-

same material,

but

drilling procedure,

different load spectra,

"transform
compatible

into

the

same

results (Ref.

EIFS

Fig.

fastener type

stress level,

etc.) should

distribution

4.4).

and

to

For example,

obtain

each TTCI

"value in a given TTCI data set is grown backward from a0

an

EIFS value,

a(0),

at time t=0 using the applicable crack

"growth relationship for each TTCI data set (Ref.

The

resulting

a(O)

value

for

eacl,

TTCI

fractographic data set maps into the same EIFS

(Ref.

Fig.

To
Eq.

value

4.5).
in each

distribution

the

crack growth parameters Q and b in

3-2 for each fractographic data set to be pooled,

1 is used.
"da(t)

In Eq.
_

Eq.

4-

4-1
*b

Qi

[a(t)]

4.15
41

Fig.

4.4).

distinguish

__de

to

(4-1)

-0

-H

c2)n

4L4

-H'

'

-4

f4

0u

'-.1z

-74--.j

44W

E -44

44,

0)

4-4

0 0 *H

)-A

4
-. 4
%9

4.16

V3

144

N.

.4 %

SCRACK

Ti
T

- 'SIZE

TTCI's in

ith TTCI Data Set

"[."

,.-,
a(O)

1:'"",TIME

EIF masBr curve for ith

[f racosrahic data set
a0 exp(-QiT1 ) - EIFS

"EIFS Distribution

'..."

Fig.

4.5

Illustration
Showing TTCI's and EIFS Master
Curve for the ith Fractographic Data Set

. '...

4..1

%O

and

bi

are

crack

fractographic data set.

for

different

distribution,
the

same

3.4.1),

data

To make the

sets

for

the

fractographic

"compatible"

for

value.

For

> 1; for Case II,

The subscript "i"

rceults

single IFQ

is

IFQ

model Case I (Ref.

1 (Ref.

Section

Section 3.4.2).

added to the parameters a, 03, and e

in the TTCI distribution of Eq.

3-1

to

distinguish

values

for the different fractographic data sets to be pooled.

example,
bound

aj,

of

ith

each fractographic data set pooled should have

bi

constants

growth

03j,

and ei denote the shape,

TTCI

for

respectively (Ref.

Fig.

the

ith

scale

fractographic

For

and

lower

data

set,

4.4)

A
Theoretically,

"generic"

EIFS

distribution

can be

obtained for pooled fractographic data sets by imposing

following conditions (Ref.

~2
b

4.4):
an

u1

The

Fig.

conditions

the

u3

of Eq.

n-

4-2 are necessary to obtain the same

EIFS cumulative distribution,

Fa(0)(x),

fractographic data sets to be pooled.

WA

4.18

for

each

of

the

Because

of sampling fluctuations due to limited amounts

of test data,
based

on

Eq.

the

4-2 cannot be satisfied exactly.

experiences of this program,

it

However,

appears that

the conditions of Eq. 4-2 can be reasonably satisfied - at

least to an acceptable degree
ror practical durability
analysis.

Also,

Qi~i

may

it

appears that the parameters ai,

be material constants which are independent

load spectra and stress level (16].

using
needed.

existing

fractographic

In the present investigation,

optimization method (Ref.

4.4.2

The
set can
methods

Further

results

be satisfied using data poolin

4.4.2.1

bi

Eq.

and
of the

investigation

[e.g.,
4-2 is

24,25]

is

forced

to

procedures and the parameter

subsection 4.4.2.4).

Calihration and Data Pooling Procedre

Determination of Qi

crack growth parameter

be determined
are described

in
in

Qi in Eq.

various ways.
this section.

based on the least squares fit

criterion.

4.19

4-1 for each data

Two different
Both methods are

Method I

4-1

Equation

be

can

squares fit for:a as shown in Eq.

In

Using

in

dt

in

transformed

linear least
-,

4-3.

i + in a(t)

(4-3)

the least squares criterion, an expression for Qi can

be determined as follows:

~-r da(t)

E [ln
3E= 0

dt

in Q-i

n a(t)]

in da(t)
dt

D,

in Qi

(4-4)

In a(t)]

-/

(4-5)

.i5i

2-A

Solving Eq.

Eq.

4-5 for Qi,

exp

(4-6)

{.,

Values

of

da(t)/dt

can

be estimated for a given a(t)

using the fractographic data

in

range,

There

i.e.,

aL

to

aU,

the

selected
are

for

example,

the

Direct

crack

size

various

numerical

This

includes,

techniques for estimating da(t)/dt values.

Secant Method [28],

the Modified

4.20

.'-

. . .. .

,*.

obtained.

4-6 is

,_

_-..

.-

.-.--

."..

...

. ."..

. .

,.

,,.-.....

....

.. ,.~..

' ."

" -"

. ..

. .'

Method

and the Incremental Polynomial

Secant Method (29,30]

(28].

Method 2

The

growth

crack

Q. can also be determined

parameter

from the generalized crack size-time relationship of Eq.

L-Qi

a(t.) = a(t.) axp

The notations for Eq.

Equation
fit

4-7

(ti-t.)J

(4-7)

4-7 are described in Fig.

can be transformed

4-7.

4.6.

into the least squares

form as f ol lcws:

in

a(t*)

following expression for Qi is

The

invoking the condition

(4-8)

(t-t

aF t)-Qi

obtained from Eq.

4-8 by

E - 0,

Qw

(t4..
i (t i-tj).'
-.

____

_____(4-9)

4.21

.-. ,.-.v
.-.
-S ,-%.-

..-....

S5

v,-

.,.v......

v.-,.....

-......

.....

:.-.......

-.-

--..

-.-'.,-,.Q
..
--

-:.

--

j 7N . -- -!*

-W

-- w ,W
-,

..

-"

'

7 1

Fractography crack size range of interest

N-

No.

of a(t),

-airs

u'

N)

8TJ~

a(t)
tN

CRACK
SIZE
a(tN-1)
tN- 1

.N1.

aL

---

a(tl)
1

Fig.

4.6 Notational Scheme for Determination

of

4.22

"5I

,
L

4-10 in the notational form

Equation 4-9 is rewritten in Eq.

shown in Fig.

4.6.
2

i=NQi-1

N-=

In

Fq.

a(ti)1

i-i

4-10,

( ti
i

t i-1 -N

22

i-i

i=N

j=1

a (ti.

ln

(4-10)

(c.-t.
1

i-j

number of [a(t),t] pairs in the selected

fractographic crack size range,

aL to aU.

Discussion

crack

The

growth

in

Qi

parameter

Eq.

4-1

can

be

determined using the fractographic data for a single fatigue

crack or for all fatigue cracks in a given data set.
be determined using either Eq.
Eq.

4-10

is

4-10.

4-6 or

more appealing than Eq.

determined directly from

the

a(t)

Qj can

Analytically,

4-6 because Qi can be

and

values

without

"having to approximate da(t)/dt values.

two

The
compared

in

methods for determining Q, are illustrated and

Section

4.6.1.

the fractograDhic data neede

set is

V.

given in Fig.

A conceptual description of

to determine Q4

for a given data

47.

4.23

-,i
.4I

J-

.41

'V

.4

CC4 L

ViLi4

I1
cc

4-4

14-i

.44
w

4j

-Li (
rq 4

4'ro

4
4;.w

C..D

4-

p.

U
00

4.4
I'.

*8r

*~41
--

- -

-*.-

-.

.-. -

~aa..aLa2LaI
4. - '~*4 ..

4g4.*
*.

At

14
"-4 60-

41

Determination of ai,

4.4.2.2

Weibull

the

set,

data

fractographic

given

For

6i and Ei

of

distribution parameters for shape, scale and lower bound

are

TTCI

by

denoted

-Land ei, respectively.

ai,

These

parameters can be determined as follows:

Define

"1.
size,

a0 ,

in the range: aL < a 0 S aU

value

given aO.

3-13)

Eq.

(Ref.

Estimate

ranking the (t

xu,

4.7).

in the range:

(Ref.

3-12)

the

a corresponding

distribution

cumulative

ei

value

4.5).

Fig.

- ei) values for the

a 0 , in ascending order using Eq.

in

result

the smallest TTCI for the

For a given xu there is

and vice versa (Eq.

3.

Fig.

The xu valuf., selected should

aL 5 xu s a 0 au.
ei

(Ref.

Select an upper bound EIFS value,

2.

an

v:ilues for a selected reference crack

TTCI

reference

of TTCI by
crack

size,

4-11.

r
(4-ii)
(-1

FT(t) =
T ~n+l

where r = rank of (t

n - No.

of (t-

ei)
ei)

(1, 2,

....

n)

values in the fractographic

set for reference crack size,

4.25

a0 .

data

4.
6q

Determine

distribution.

(ii and Oi.

Eq. 3-1,

least squares fit

in

Y = in

can be transformed into th- following

form.

Z =

where Z

The three parameter Weibull

+ U

{ -in

(4-12)

[l-FT(t)]1

(t-Ei)

Us-."i inci

ci

arid

0i

can

be

determined

using

Eq.

4-13 and 4-14,

respectively.

NYYZ- (EY) (EZ)

NNZy2_ (Ey) 2

=
exp

where

previously

No.
(Eq.

of

=x

a.N

(4-13)

(4-14)

TTCI values for a 0 ; and Y and Z are defined

4-12).

4.26

Determination of a and N__

4.4.2.3

3-15 can be determined ms follows:

a and QO in Eq.

1.

Compute the product Qjij for each fractographic data

Then determine Q0

set to be pooled.

2.

Qij.

Ave.

a using the normalized TTCI results for all

Compbxte

the fractographic data sets pooled.

methods

suggested

Two

are described below.

Method 1

Compute

(t-ei)/Oi

for

each

fatigue

in

crack

each

Pool

a0 .

the

fractogrrnphic data set

for

the

specified

for

all

the

C ta sets and estimate the

values

resulting

cumulative distribution,

FT(t),

using Eq.

4-11.

In

Eq.

4-

11, r = rank of (t-ei)9ii and n = nunber of (t-ei)/3i values

for all the fractographic data sets combined.
is

Weibull

the

scale

parameter

for

each

Note that

0i

TTCI data set

separately.

Equation
transformed

3-1

subscript "i"

the

notation

can be

4-15 as follows,

into Eq.

w:here Z

with

(4-15)

"X

ln { -ln [

-FT (t)

4.27

-.-- --.~-~,. .. , -- S

S -

Using

in

Eq.

-r

(t-e.)/U

4-15

and

iuhe

following equation for a

a ..

least

squares

criterion,

the

is obtained.

4-16)

EX

Therefore, a can be obtained using Eq. 4-16 and the results for
the pool^d fractography.

Method 2

A non-dimensional form for FT(t)

I::

is

given by Eq.

4-17.

it- iIr (a0/u

FTt)

1-

where Qi =(Avr.

and QO

0a/x)

exp -[t

(4-17)

Qiji)/Pi (normalized Qi value for data set i)

Ave.

constant

(for

generic

EIFS

distribution).

4.28

"

Equation
3-1

4-

is

'

obLained by substituting Eq.

and rearranging

terms.

Equation

can

4-17

least squares fit

Z =

where,

into the following

aX+B

X = in

Wi t h

transformed

into Eq.

form,

Z = in

B =

be

3-13

(4-18)

l-FT(t)]}

in

it

aln Q3

the

least

in

(a 0 /xu)]

square

criterion,

Eq.

4-19 for

c is

obtained.

Zz
X-( in

can

thern

be

resu.A'ts and Eq.

obtained

using

4-19.

4.29
-r--

4-19)

the

pooled fractographic

r. -

717

Optimization of Par3meters

44.2.4

)I
.1

The

xu,

parameters

and

Q0

in

Eq.

optimized to meet the user's criterion for

There

ara

different

ways

this

3-15 should be
fit.

acceptable

For

can be accomplished.

example, the sum squared error (SSE2) can be minimized for a

and
Q).
given combination of parameters (i.e., aoX, ;0
x values are assumed and the corresponding o, QWand SEE are
determined.

4.4.2.5

The

"

This procedure is continued utiitil the SSE is minimized.

,
-.

:1

General Ste.ps,

essential

steps

for

determining

the

IFQ

model

parameters are as follows:

1.

suitable fractographic data tor" fitting the

Select

model parameters.
4.:

2.

-elect

fractographic

interest for durabili.ty analysis.

range of 0.020"

;Therefore;

to
to

For example,

0.050" (Ref.

such

a crack size

Fig. 4.7).

The

readings tor each fatigue crack may not cover

"the selected crack size range of interest,

In

size range of most

holes.
0.050" might be used for fastener holes

aL =0.020" and aU

fractograpnh.c

cra~k

cases,

the

fractography

i.e.,

aL

to

7.

aU.

should be extrapolated,

.4.,

.~.30"'

1B

:e-.

to cover the range from aL to aU (Ref.

forward or backwards,
SFiC.

4.7).

Select

3.
Fig.

4.7).

"the

IFQ

a 0 ,for TTCI's (Ref.

a reference crack size,

following "rule-of-thumb" for ao is based on

this

of

experience

modeling

IFQ

experience with pooled fractographic results and

optimization

parameter

needed

is

model

additional

provide

to

More

program

guide2ines,

a0

4., Determine the crack growth parameter,

,in

Use fractographic

se
Se,

for each fra.ctographi:

Tatigue

Determine
crack.

the
Use

rTCI

6.

for

selected

interpolation

procedures as required (Ref.

Fig.

or

a0 for each
extrapolation

4.7).

For a given x. determine the corresponding

for each fractographic data set separately.

the product Qi3i for each data set

4.31

results

he range from aL tcaU,

S5.

6.

in Eq. 4-1

Qi,

(Ref.

Fig.

Then,
4.4).

ci andi
compute

fractographic data

With

sets.

for

the

average

Qj_3i

Qjii

average

the

Compute

7.

the

pooled

value,
0%

determine

crack growth parameter

normalized

the

for

Qi,

each data set as follows:

ave.

n i0I

Qi3i

i(4-20)

1..

of TTCI data sets pooled.

No.

where n

the

Normalize

8.

results

TTCI

pool the results and determine the

described in Section 4.4.2.3;

(x for the pooled data sets.

corresponding

9.

Using the resulting values for xu,

determine

well

how

because the x

chosen is

fractographic

value of xu and repeat

steps

through

"combination of xu,

8.

By

a0 ,

cc,

fit

the

one can choose another

in

described

procedures

iteration,
and ave.

the

fits

improve

To

used,

same

the

Qij

may not be satisfactory

not optimum.
data

and ave.

distribution

The fit

used.

data

the

EIFS

che

fractographic

for

S.v

for each data set as

one can determine the

"best-

the

Qig3 giving

fit"

"4.32

S..

~...

. .

-.

5'%

"

,%

4.5

STATISTICAL SCALING OF 1, FOR MILT II'I, 1KDE'J'AILS

It's

not always practical or cost-effective

a fatigue crack in each fastener hole

that

will

useful

be

large

enough

fractographic

statistical

method

(e.g.,

data
has

Weibull scale parameter,

in

for

to initiate

test

specimen

aU ; 0.050") to provide
quantifying

IFQ.

been developed for determining The

3, for the cumulative

distribution

of TTCI using only the largest fatigue crack in any one of t

fastener holes per specimen [16].
the

TTCI

in

This method accounts

for

each fastener hole per specimen and minimizes

the fractographic data needed to

determine

the

IFQ.

For

example,
fractography
is
required
for only the largest
fatigue crack in any one of I fastener holes per specimen in
the

data

elsewhere

set.

Further

details

discussed below and

(16].

fatigue

test

specimen

replicate fastener holes (e.g.,

diameter,

are

fastener type,

may

contain

some:

fastener fit,

one

drilling
etc.).

or

technique,
In practice,

the test specimens are fatigue tested until a fatigue

aU

is

crack

initiated in at least one of the t fastener holes

per specimen.
"I

more

When the fatigue test is

stopped,

only

one

fastener hole per specimen may have a fatigue crack 2 aU and

the remaining fastener holes

may

observable fatigue crack.

he
.

4.33

or

may

not

contain

an

The

resistance

of

tastener hole per specimen to

each

fatigue cracking should be accounted for when

and

TTCI
higher

EIFS

Fastener

distributions.

resistance to fatigue cracking

Although each fastener hole

in

with

the

longer

have a

TTCI.

a specimen may be drilled and

the IFQ for the fascener

Therefore,

the

holes

typically

the TTCI will

fatigue tested the same way,

defining

vary.

holes should account

for

the TTCI for each fastener hole.

If

each

fastener

replicate

stress

specimen is

hole per test

history,

the

Weibull

scale

subjected

o a common

parameter

for the TTCI distribution can be determined using

Eq.

4-21

[16],

I
(4-21)

where:

Weibull

01

scale

parameter

for

the

TTCI

distribution.

Weibull scale parameter ror the TTCI distribution

based on the largest fatigue

fastener

holes

per

fractographic data set.

Weibull shape parameter

4.34

"."
.

.*

crack

specimen

in

in
the

one

of

complete

S= Number of fastener holes per specimen

be used to "scale" 9, to accounit for the

can

4-21

Eq.

TTCI for each fastener hole in the data set.

description of

, scaling

are

SSfatigue

noted.

test specimen is
stressed

If

fastener

the TTCI data is

been

For

distribution

determined

directly

successfully

used

assumed
(i.e.,

to
ti

goodness-of-fit

""

of

the

four

the

equally

= 4).

available for each fastener hole in

4-21 because
Eq.

using the TTCi data.

for

observations

TTCI

EIFS

the

purposes,

contain

both

countersunk fastener hole applications

Actual

4.8 and the

and

illustrative

there's no need to use Eq.

the data set,

be

holes

conceptual

illustrated in Fig.

is

effects of scaling on the TTC1

distribution

are

theoretical

4 .35

straight

can

4-21 has

bore

and

check

the

[16,19-21].

needed
TTCI

to

distribution.

Hrle with largest fatigue

crack per specimen

C.RACKeullv-stressed

ho~es

holespepier

for largest
fatigue
Scrack Accounts
in 'each
of 9, holes
per
Sspecimen

fractographic data set containing

Xu

nfracograhL:samples

- '

1 .0

0.632

CUMUL-;ATIVE
DI STRItBUTION
OF

...

FT (t)
%I

TTTA

Fig. 4.8

Illustration of A Scaling Concept for the i th

Fractographic Data Set

4.36

", 4

Therefore,

available for only the largest

fractography is

if

"fatigue crack in any one of

specimen,
,

these

observations

fastener

holes

must

used

be

this means that

For example,

goodness-of-fit.
"used instead

test

to check the
;

must

be

of 03 to check the TTCI distribution goodnessavailable

of-fit because fractographic results are not

each

in

fastener

hole

in

the

set.

data

goodness-of-fit of the theoretical

In this case,

cumulative

for
the

distribution

of crack size must also be checked using actual fractography

available. As a result,

used

to

model parameters can

described

in

Section

be

determined

described

in

Table

to illustrate how the IFQ

determined

be

4.4.

for
4-1.

using

the

procedures

To illustrate the procedures,

including data pooling methods,

will

of the

ILLUSTRATION OF PROCEDURES FOR DETERMINING IFQ

The purpose of this section is

the IFQ

further discussed in Section 4.6.3.

"of-fit. This aspect is

define

analysis, but Skis employed to verify the goodness-

durability

4.6

is

the

IFQ

model

parameters

the three fractographic data sets

Essential

features

of

thi

calibration procedures will be illustrated and the goodnessof-fit of the derived EIFS distribution will be demonstrated
for

pooled

data

sets as well as for individual data sets.

"4.37

Description of Fractographic

Table 4-.

Data
Set

Data Sets

Gross
%
Fastener
Stress
Load
I.D.
Transfer & (ksi) Dia

Material

ALR4 7475-T/351

32

15

k"

MS90353

No.
Specimens
Load
Used
Spectrum
F-16 400h

10

Aluminum

AXMR4

38

AFXHR4

Notes:

Pull-Through)

(a)

Countersunk rivet (Blind,

(b)

Fastener holes drilled using Winslow Spacematic drill

(c)

Fastener

(d)

Ref.

(e)

Fractography basis:

Fig.

fit

clearance

hole:

for specimen details

4.3

Largest crack for 1 of 4 holes

per specimen (I - 4)

4.38

* ~~ ~ ~

2-'~

~~

~Y
~$*,.\
~~.w,*-Y-7~.*

#LZI'~r

including

The resulting EIFS distribution will be discussed,

p-scaling considerations.

Determination of TTCI and Qi Values

4.6.1

TTCI

Qi

and

values can be determined from the fatigue

The

crack growth data (fractography).

versus

(a(t)

obtained

t)

from

determining

the

AFXMR4 and AFXHR4 is

for data sets AFXLR4,

Volume

VIII

[24].

data

fractographic

The

for

procedures

TTCI values for a selected reference crack

size, a 0 , and for determining Qi values are

illustrated

in

Tables 4-2 and 4-3, respectively, for data set AFXMR4.

A fractographic

size range of aL

will be used to define the

fractographic
VIII,

EIFS

data for specimen

= 0.020"

interpolation.

bj

0.050".

0.0275",

are

interpolated

selected

desired
0.035",

shown

in Table 4-2,

TTCI's

versus

reference

4.39

a(t)
five
and

is used to

crack

size.

including a comparison of

observed

fractographic results.

for

0.0425"

three-point Lagrangian interpolation

determine the TTICI for the selected

Results

are

for

the

from Volume

AFXMR4 (563HB)

TTCI's

reference crack sizes: 0.020",

Using

distribution.

the TTCI values can be determined

values

to aU = 0.050"

values

from

the

Illustration of Procedure for Determining

TTCI Values for Selected a(t)'s

Table 4-2

.:

.1%

FractIaphicData**
t
a( t)

AFXMQR4
(563HB)

(In.)

(Flt. Hrs.)

0.0196*

4800*

0.0237

*".

(Flt. Hrs.)

(In.)

5200

0.0196

4800

0.020

4847

0.0237

5257

0.0275

5643

0.0262

5600

0.0262

5515

0.0314*

6000*

0.0314

6000

0.035

5295

0.0367

6423

0.0367

Interpolation lesults
t
a(t)

6400

I,0.0425

6803
0.0433

6800

7200*

0.0509*

0.0433

6848

0.0500

7166

0.0509

7200

Notes
*

Fractographicresults used for three-point Lagrangian

interpolation

,*

Ref.

(241

Vol Viii

44

"_

'

*",
" '

,.,-

~4.40

."

"."-.

S"."-

"

-,

*--

..

,.-

'

N4*~*

.- ",""

.2I.,a

[,,

..

L ,

,'.'

-.

-'

,-

-"."

following

The

procedures

are illustrated in Table 4-3

(567HB):

for specimen AFXMR4

o extrapolation cf fractography to aL

[30]

o Modified Secant hiekhod for defining da(t)/dt's

o determination of Qi for a single fatigue crack

Since

the

fractography

crack size > ai

(0.020"),

by

determined

for this fatigue crack starts et a

Various

extrapolation.

techniqres could be used [e.g.,

generalized crack growth relationship of Eq.

determine the TTCI vplue for aL = 0.020".
parameter

Q in Eq.

be

extrapolation

In this

31-34].

must

0.020"

the TTCI at aL=

case,

the

used to

3-10 is

The crack

growth

3-10 (Note: Q = Qi to be consistent with

notation used for fractographic data pooling) was determined

using the a(t) versus t values shown in Table 4-3.
for

Eq.

However,

3-10

were

determined

using

Eqs. 4-6

Qivalues
and

4-10.

the extrapolation for a(t) = 0.020" was made using

Qi= 5.202 x 10-4 as shown in Table 4-3.

The

procedures

illustrated

in

Tables4-2 and 4-3 were

used to determine the TTCI's for five reference crack

for each specimen in three fractographic data sets.

4.4..

sizes
Results

Illustration of I'rocedures for Deterzin'ng

Table 4.3

AF,=4 (Specimen N'o.

Flight

567 RB)

da-

x 110

d_

Hours

a (t)

(In.)

(4243)W)

0.0200

4400

0.021/

i2. 00

4800

0.0265

14.00

5200

0.0329

13.63I

5600

0.0374

13.88

6000

0.0440

16.00

0.0500

(15.4410

6400

0.0502

13.13

6800

004

33

(6387)'V

9i for Data Set

X
,

(In./Hr..)

4-6

(10/hr.)
Eq. 4-10

(1C. 83)(1)
4,947

5.202

Notes

Based on Modified Secant Method r29, 30]

SExtrapolation from a(t 2 ) - 0.0217" to a(tI) - 0.0200"

-4 tUsing Eq. 3-10 and:
- "i 5.202
X 10-4
~2 4400
da(t)
0.0217 - 0.020
dt

4400

- 4243

i0"83x10- 6

SThree - point Lagrangian interpolation

t

-=(0.050 -0.044) (0.050 - 0.0374) (6400)

(0.0502 - 0.044) (0.0502 - 0.0374)
+ (0.050 - 0.0502) (0.050 - 0.0374) (6000)
(0.044 - 0.0502) (0.044 - 0.0374)

+ (0.050 - 0.0502)

(0.050 - 0.044)

(0.0374 - 0.0502)

dt

0.0502 - 0.050
(6400 - 6387)

(5600)
(0.0374 - 0.044)

0.0500
(6387

0.044
6000)_

6387

15 44xI0-6

4. 42

a~~.'

'-

-'I.h-t

are summarized in Tables 4-4,

4-5

AFXLR4,

respectively.

AFXMR4

through 4-6,
five

and

AFXHR4,

and

4-6

for
In

crack

sizes.

Also,

the

parameter Q9 for two different equations (Eqs.

is shown for individtual fractographic
4-6, for data sets AFXLR4,

shown in Table 4-7,

for

Table 4-8.
a0

4-6 and

for determining

fractographic

data

For illustration purposes,

0.035" and for xu = 0.025".

and product Qt p

4-10)

Qj values,

and QP

and Q0, described in

are illustrated in this section.

individual

growth

AFXMR4 and AFXHR4J are

Determination of a

procedures

Section 4.4,

*.he

along with the input data used.

4.6.2

The

of

crack

data sets.

sets

Tables 4-4

average TTCI values are shown for each

reference

based on Eq.

data

were first

sets

Input/output

is summarized in

results are shown for

Th, parameters ei,

determined for

individual

ti' A.
data

sets.

Using the average

value of 2.155 for the three data

sets,
set.

the normalized Qri value was determined fur

For example,

each

Gvalues of 1.739 x 10- 4 , 2.818 x 10-4

and 4.040 x i0-4 were determined for data sets AFXLR4,

and

AFXHR4,

data

respectively.

AFXMR4

Ci values were then determined

4.43

Summary of TTCI Values

Table 4-4

for Data Set AF'LR4

T7351 Alumimum)

(7475

TTCI

Specimen

Data

No.

Set

AFXLR4

(ksi)%=-0.020

(Fit.

%-0. -75a

Hours)
0

-0.035'

__

a 00.0425 a 0 .,

9835

11433

12818

13990

14949

35TA

5100

6267

7474

8734

10034

36HA

11070

12717

14066

15117

15869

120AB

9551

13650

16676

18934

20419

121HB

14608

II

16791

18653

20193

21411

559HB

3649

4089

411

4917

5305

123TA

20844

22334

23536

24451

25078

5707

7964

9905

11532

12844

6435
9469

687 5
10262

7293
10945

7689
11517

11115

12478

13611.

14512

33TA

32

"124TA
57B5973

8566
11__558TA
Ave.

9520

Q- 1.822 X 10

Qj-1.857

4.44

-4

/Hr(Eq. 4-10)

X 10- /Hr(Eq. 4-6)1

:I

S-,-

,'"-i-i
- :i

-,,,

,.

-.. '..

N- :

L.-

N.

,.

..

.-

..

4-r

Table 4-5

Sunmiary of TTCI Values for Data Set AFXM1R4

(7475 - T1351 Aluminum)

TTCI

"Data
Set
AFXMR4

Specimen
(ksi) a
No.
C
34
562TA

-0

0,0' a0
0

TTC,
0.0275 a

Hours)

(FiC.

-0

H
035 a
0

0.0425

2M29

5379

7821

9955

11780

563HB

4847

5643

6295

6803

7166

561 r3

11272

12491

13455

14163

14617

565HA

6476

8412

10077

11471

12595

566HA

2643

5277

7308

8738

9566

567HB

4243

4793

5354

5885

6387

568HB

11078

12669

14014

569HB

3076

4541

5765

6750

7495

570HB

2509

2832

3099

3313

3471

541:

6893

81-32

91,32

9893

Ave.

15962

1 111

4 H-4r

Qi

Qi

S4..

a0

2.027 X 10

(Eq.

2.091 X 10

/Hr(Eq,

4-10)
4-6)

Table 4-6

Sumniry of TTCI Values

"(7475

? : - .D

aa
ata

Set

for Data Set AFXHR4

T7351 Aluminum)

(Fit.

SpcmnTTCI
p e c i me n
S

No.

Hours)

(ksi) a0-0.020f a0-0.0275" a0 -0.035" a00.0425" a0

0.050

0~

AFXHR4

.. .

2651

3539

4192

4611

4794

572HA

5090

5600

6086

6548

6985

573HA

5726

6559

7321

8012

8633

574TB

2973

3312

3607

3859

4067

575TB

8119

8945

9678

10315

10859

576HA

3194

3665

4051

4353

4569

577HB

1784

2085

2321

2492

2597

"579HA

6159

6776

7306

7747

81.01

580-A

1894

2394

2814

3154

3415

4177

4761-

5264

5677

600?

571HB

38

Ave.

Q'

4.46

4.981 X 10

"5.092

/Hr(Eq.

X 10-4 /Hr(Eq.

4-10)
4-6)

Table 4-7

Suwmary of Q0 Results Based On Eq. 4-6.

AFXLR4

Ci

a't

t dt-i
F95t.
rs_

AFXMR4

- xt
t
(

(In.

0,020

9521

4.70

0.0275

1.11.

0.035

/H

1.857

dt

AFXHR4

-1o a[ Qit

dt

l/Hr.)
Flt,
Hr_

(l/Hr. F-t.
*r
14_

(I/Hr.)
i.L /114,

5415

5.07

2.091 4177

12.78

5.10

689.

5.56

4764

13.89

247E

6.06

813

6.78

5264

16.58

0.0425

3611

7 .47

913

28.68

5577

20 .62

0 .050

451

8.32

989

39.86

6002

23.08

'.:i

sz

IkZ
4.47

5.092

io6 ;Q [ %o4
1
i 1

*,

iIj

i :

-4

U)

0Ln

4.

LO~

cn

L4

M_
_
_

~Cco7
W-

0j

_w

M. c 4.

M.. 'T

O'n'

ar-

4.0
0

0D

-sImm

'4

---.

c'4
'o'

'm
U\

T4

m \

Jcu
w

'j

.4,

4.48

o
0

L
r

,o

0,

using the applicabie

and Eq.

values,

x,

--

0.025"

3-13.

The

value for the three fractographic data sets was

determined using
procedures

the

described

determined using Eq.

IFQ

model

data
in

t1 Q

parameters

EIFS

4.6.3

Table

4.4.

4-9

An

and

th'-.

i = 1.801

was

based on

ii

he three da:a sets ar-

Pertinent paraileters and concepts

are

also

sumnarized

and

4.9.

Goodness-of-Fit of IFQ Model

well does the EIFS distribution established for the

pooled data sets fit

question

in

Sect kon

distribution

illustrated in Fig.

How

shown

4-19 and the data given in Table 4-9.

summari-ed in Table 4-10.

for

0.035",

ao

must

be

the observed fractographic data?

This

resolved before the EIFS distribution is

accepted for durability analysis applications.

Kd

The

IFQ

model

parameters

and Q0 in Table 4-10 were

determined for given a 0 and xu values.

This combination of
aO,
xu,
ct and Q0 may not necessarily be the one that
minimizes the sum squared error (SSE).

ft..

In

any

case,

the

4.49
4..
"ft

-1

Table 4-9

Illustration of Procedures and Results for

Determination of a

(a
c
Frac
Datat0ographi
Set
_......

t(TTCI)

(FLT RRS)

AFXLR4

4511

Q1/Hrs._
x104

__xOt

I. D.

0.035"; x u

1.739

.034

.1172

7474

.207

9905
10262

1.722
1.784

12
14

414
.483

12818

2.229

17

.586

14066
16676

2.445
2.899

18
21

3.244

24

.621
.724

4.093

28

.966

.873
1.509
1.624
1.774

2
8
9
13

.069
.276
.310
.448

2.059
2.204

15
16

.517
.552

2.839
3.792

20
25

.689
.862

3.949

27

.931

.937

.103
.138

3099
5354
5765
6295

2.818

7308
7821
10077
13455

S14014
3

2321
2814

Ave.

ri~
r/n+1

_"
Rn
Rank
r

.784

23536

Note:

1.196
1.299

186.5.3

AFXHR4

0.025")

IQr

6875

AFXMR4

for Given

4.040

1.805

2.155

.828

1.137

3
4

3607

1.457

.241

4051

1.636

10

.345

4192

1.693

11

.379

6086

2.458

19

.655

7306
7321
9678

2.951
2.957
3.909

22
23
26

.759
.793
.897

*".
,

"

4.50

hi'

.%

..

A,'.."-

--

,
N

,-..

OC

C-4 0,O

""HLr N
-

cia

'H

-tO

w-

00

-4

0
w

-H

L)

--T

C'4

C1 U''J %0" r-4

CIZ

C'4I

P4

-K H

C4

cqr-4

-r-

"".

"K-H
0 -

.*

cvT

00

.-.

C;
Q

-H

.
NH'J-

"

-C

4-

T
-"CI

_O01

cl

H,-

E-4__

C'

-~

"

w
-

-4

Cf
v..

(IJ

44

eC

-H

"H "H4

>

"
-K

<

-Kq <*H V

*H

zH*

4.51~~

b %.

faU

CRAK

0.050"

SIZE

aL " 0.020"

-- EIFS Dirtributiou

S. ..

Fa(O) (x) - e

];

x ,

-1.0
-

I'<,

Fig. 4.9

4
L'ua(t)u.aU

2.155

0.025"
A -4

xu
a0

O<xUXu

0.035"

Nore.

-(

EIFS Cumulative Distribution Parameters for Pooled

Fractographic Data Sets (AFXLR4, AFXMR4, AFXHR4)

4.52

user

which combination of IFQ parameters will

decide

must

give an acceptable fit.

for a 0 = 0.035"
sets

plot

goodness-of-fit

pooied.

4-9 and Eq.

is

for the distribution of TTCI's

shown in Fig.

4.10

for

the

4-17.

master curve,
cumulative

Eq.

3--11,

(Fa(0)(x)),

can be

distribution

for

used

Eq.

3-15,

to

predict

a given a 0 .

predicted results with the observed TTCI's,

plots

4,12,

for

TTCI (a 0

4.13

and

for

data

These

respectively.

0.035")

show

plotS

data

sets,

fit

and the EIFS

the

that

TTCI

By comparing the
the goodness-of-

are shown in Figs.

sets AFXLRI,

based on the pooled

parameters,
three

data

based on the data given in Table

This plot is

The EIFS distribution

fit

three

AFXHR4,

AFXMR4,
the

fractographic

4.11,

IFQ

model

results

for

the observed fractographic data very

well for the individual data sets.

r.Ci

goodness-of-fit

are shown in Fioss.

AFXMR4

and AFXHR4,

EIFS distribution,

4.14,

plots

for a crack size of 0.050"

4.15 and 4.16 for data

respectively.
based on a0

These plots show that the

0.035",

can

predict the TTCI cumulative distribution foi a 0

4.53

sets AFXLR4,

be
=

used
0.050".

to

r! I

10.0

"Pooled Data Sets:

uI

AFX(LR4,MR4,HR4)

o0 " 0.035"
-u
-

cl-

0.025"
1.805

"QS, - 2.155 (Ave. Q0)

- AFXLR4 (O"-

32 kit)

E_ - AFXMR4 (W"-

34 kni)

- AFXHR4 (0-- 38 kal)

1.0

'-

--

, - ,-.-4

0.1

0.1

--.

\
44.5

,0.02

Fig.

*..,

I I I

1.0

0.1

",Iii

4.1i0

~i0.

I II

TTCI Goodness-of-Fit Plot for Pooled Fractographic

AFXMR4,

Data Sets (AFXIR4,

AFXHR4) ;a

.005

.i

4.54

;
;"

' 5-

.5, ,

'

* . *.'-

. -. . .

AFXLR4 Data Set

a 0 - 0.035"
- 0.025"

3.
0(
Q
..-

1.805

2.155 (Ave.

Qj

,)

-c 1.739xl10 4 /Hr.
'-

- 12388 Hrs.
Test Data
0 For Pooled Data Sets:
*_"

1.0l
0
"

_...,

1.935 Hrs.

"

AFX(LR4,MR4,HR4)

SII

0.1

1000

10000
(t

Fig. 4.11

30000

-1

TTCI Goodness-of-Fit Plot for AFXLR4 Data Set

Based on IFQ Model Parameters for Pooled Data

"Sets; ao - 0.035"

a,q
* a,

4.55
"."

'.

" .,

. .-

. .

-.

-.. .-- - - .'-

.v ---.v ..- . -. ,.--.. .

AFXMR4 Data Set

a0 - 0.035"
xu" -0.025"
3.0
:,

"-

Of - 1.805
Aj - 2.155 (Ave.
',

Q4)$//

2m2818x0"4/Hr.
- 1194 Hrs.

1.0
1.0

Test Data
El0
For Pooled Data Sets:
AFX(LR4,MR4,HR4)
El

0.08
1000

10000
(t

Fig. 4.12

30000

TTCI Goodness-of-Fit Plot for AFXMR4 Data Set

Based on IFQ Model Parameters for Pooled Data
Sets; 0 = 0.035"

4.56

.M '.~

7'i

i 2: '

i .

"

J W.

'. W

F.'

.Fk '

.'

'.

,- . W
r `

y w w,

WW~
.

74

AFXHR4 Data Set

0.035"

0.025"
-.

1.805

S- 2.155

(Ave. Q6,)I

I.- 4.040x10-4 /Hr.

3.0
Ad

833 Hrs.

5334 Hrs.

-F--

Test Data
r Pooled Data Sets:
AFX (LR4,N

*F,

1.0

0.1
0.08

ioo6

4.13

10000
(t

Fig.

30000

-_)

TTCI Goodness-of-Fit Plot for AFXHR4 Data Set

Based on IFQ Model Parameters for Pooled Data
Sets; a 0 - 0.035"

4.57

'

LL
AFXLR4 Data Set
a0 - 0.050"
xu - 0.025"
ot-

1.805*

- 2.155 (Ave.

4:oe.

q - I.739x10"4 /Hr.
. 3986 Hrs.
-12388

Hrs.

Tt?
st Data
*Pid~0 ej
Pooled Data Sets: AFX(LR4,MR4,HR4)

with ao - 0.035"
3.0

Y
1.0

0ii"

r-4,

0
,,.

0-

0.1

10000

1000
(.o

Fi3. 4.14
-

L!-

30000
-t

TTCI Goodness-of-Fit Plot for AFXLR4 Data Set

.
.o.I
.
Based on IFQ Model Parameters for Pooled Data

Sets; a.

0.050"

4.58

AFXMR4 Data Set

a0

0.050"

xu - 0.025"
--1.805

2.155 (Ave. Qi# j.)I

"- 2.818x10- 4 /Hr.

S-

2459 Hrs.
a-7648 Hrs.

S~~El
I- Test Data
Pooled Data sets : AFXtLR4,MR4,HR4)

S with a0o
3.0

0.035#
-7

Z
m0

1.0
%-

El

LrL
VK4

0.1
0.08

1000

-A

10000

30000

10(t
Fig. 4.15

TTCI Goodness-of-Fit Plot for AFXMR4 Data Set

Based on IFQ Model Parameters for Pooled Data
Sets; a 0
0.050"

4.59
",;/

,%

-".,.-,

,7'''V

'\",',''"."'''

"-

" "".

,"

" ".

,'''.'"

'"-..'""'.-

. "

.".

"-".

".

AFXHR4 Data Set

a 0 - 0.050"
u 0.025"

-1.805

.o(

2.155 (Ave.

Qf.- " 4"4x1-/Hr.

i-

1716 Hs

- 5334

Hrs.

A-- Test Data

Poold Dat
g AFX (LR4,MR4,HR4)

"3.0

"

IT',

0A

==

*-'

1.0

r-4
r-4

0.1

0.08

700

1000

II

r I1
10000

(t
Fig. 4.16

30000

-&)

TTCI Gocdness-of-Fit Plot for AFXHR4 Data Set

Based on IFQ Model Parameters for Pooled Data
Sets; a0 - 0.050"

4.60

plot of log QZiversus log a is shown iii Fig.

the three data sets.

shown.

In

Moreover,
EIFS

it

this

A best fit

case,

Eq.

line for Eq.

is

also

3-18 fits the data very well.

would be reasonable to use

distribution

3-18

4.17 for

detined

by

Eq.

3-18

and

the

parameters in Table 4-10 to

predict the cumulative distribution of TTCI's for

different

stress levels.

4.6.4

Parameters

Discussion of ElFS Distribution

for

the

EIFS

cumulative

distribution are

shown in Table 4-10.

These parameters

pooled

results

three fractographic data sets (AFXLR4,

AFXMR4,

AFX:iR4).

for

The goodness-of-fit plots in Figs.

t= 1 rather than t = 4 (actual).
in

Table

specimen.

However,

on

the

4.16 reflected

The IFQ model

parameters

fastener

holes

per

test

Since the fractographic data is not available for

fatigue

specimen,

4.10

based

4-10 were based on the fractogranhic data for the

largest fatigue crack in 1 of

the

were

crack in
the

for

each

goodness-of-fit
durability

fastener

plots

analysis,

hole

in

each

were made for t = 1.

the

EIFS

cumulative

1W.

4.61
I.

Qj- L1.588xl04~O-l)4.699/H 7 7 1
Cr
C

Stress Level (ksi)

5.0

4.0

AFXHR4

S43.o

AFXMR4
r-4

i"

"

r-42.

~x
~Nq

(0:"

AFXLR4

~1.0!

20

30

40

50

"(ks)

Fig. 4.17

Plot of Q

versus 0

SCGMC

62

to Determine Applicable

distribution,

3-15,

-.
q.

should

be

used

with t

4 (Ref.

Section 4.5).

Suggested procedures hiave been described and illustraLed

in this handbook for establishing the EIFS distribution
pooled

fractographic data sets.

Using these procedures and

suitable fatigue crack fractographic

appropriate

EIFS

fcr

data

(e.g.,

24,

253

distributions for durability analysis can

be determined.

4.6.5

The

following

Practical Aspects

analytical

tools

efficiently determine an acceptable

or data are needed to

EIFS

distribution

for

fastener holes:

Suitable

fractographic

versus time

data

information)

for

(i.e.,

crack

fatigue

size

cracks

in

fastener holes.

an analytical crack growth program [e.g.,

computer

program for manipulating large amounts of

fractographic data and

IFQ model parameters,

V.'

4.63
*1

35-38].

for

determining

optimi-..eu

plotting program to evaluat, the goodness-of-fit

to

of the derived EIFS distribution

the

observed

crack sizes and times-to-crack initiation.

can be quantified for fastener holes usinc suitable

IFQ

if

fractogaphy results,
model

parameters

[e.g.,

36].

to

and

or

available,
an

analytical

IFQ

assumed

using

crack growth program

User judgment and experience are also

required

durability

analysis

IFQ

quantify

for

different

applications.

In

fractographic results will not be available

general,

material,

for the desired set of design variables:

type/:it,

stress

the

has

user

fractographic

level,

three

Load spectra,

bc-ic

options:

etc.

In this case,

(1) Use

assumed

IFQ

model parameters,

is

design

used,

or

(3)

based on similar design

variables, and an analytical crack growth program.

option

available

results and interpret for the particular

conditions,(2) acquire suitable fractographic data,

use

fastener

Whatever

the user should satisfy his requirements,

"schedule and budget.

A wealth of fractographic data is available for fatigu3

cracks in fastener
parameters

should

holes

[e.g.,

24-27].

The

IFQ

model

be quantified using such data to provide

4.64

-,

.~

~~A.
.. ' A.

a broad data base and experience for selecting IFQ parameter

values for practical durability analyses.

Recommended

parametric values for the EIFS distribution

of fastener holes are not tabulated

durability

analysis

applications.

and guidelines herein can be

used

in this

for

However, the procedures

to

parameter values for a given condition.

develop

appropriate

Tabulated values of

IFQ model parameters for different materials

will be incorporated into the handbook later.

4.65

handbook

and

variables

SECTION V

DETAILS FOR PERFORI-ING DURABILITY ANALYSIS PREDICTIONS

5.1

The

purpose

of

this

NTRODUCTION

section

illustrate procedures

for

growth

(SCGMC),

master

predictions,
damage

curve

(3)

and (4)

determining
(2)

and

to: (1) describe and

the

service

crack

discuss crack exceedance

present different
illustrate

is

formats

for

extent

of

discuss related durability

analysis considerations.

5.2

A
for

service

each

SERVICE CRACK GROWTH MASTER CURVE

crack

stress

exceedance,

region

p(i,T),

distribution

section,

and

where

prediction

predictions of p(i,r),
E!FS

growth master curve (SCGMC)

a

probability

is

desired.

is

needed

of

crack

For reliable

the SCGMC must be compatible with the

used.

recommended

SCGMC are illustrated.

Guidelines are presented

procedures

Also,

for

in this

determining

refer to references 15,

the

16,

20

and 21 for further discussions and applications.

5.1
&A.

a'.

''-

-.

.~~V..

*.

SCGMC

The

is

used

that

yli(-r),

to determine an EIFS,

a.j

service

time

the SCGMC can be expressed by Eq.

3-17.

~will grow to a selected crack size x. at

:-'

Mathematically,

a given stress region,

the saime SCGMC is

EIFS distribution from t = 0 to t = r

has been determined,

Yli(r)

determined

used

(Ref.

Fig.

For

grow

to

the

3.3).

once

the corresponding p(i,r) can

from tihe EIFS distribution (Ref.

be

Section 3.3 and

5.3).

5.2.1
The

Mathemp(i,c)

If

be compatible with the applicable EIFS

this principle is

predictions

will

not

not

be

followed,

strictly

consistent

and the

of such predictions may be questionable.

Saccuracy

To

'11
obtain

compatible SCGMC,

use the same: (1) crack

range used to define the EIFS distribution (i.e.,

Ssize

au),
iEq,,

must

SCGMC

wdistribution.

Guidelines

(2)
3-2)

crack

aL tonc

growth law as the EIFS master curve (i.e.,

and (3) procedures and -oodness-of -f it

uch
redictions;=:k~...~x.
,.3ccuray of

~5.2

criteria

:' questionable_:.,-.:'
,2:''may be.
5.

to

.,--.

..

u'-

q)

CRACK
SIZE

TIME

S".'

Fig. 5.1

Service Crack Growth Master Curve

5.3

determine

the

SCGMC

master curve(s).
compatible

as

used

determine

t:o

The following guidelines

SCGMC

are

the EIFS

for determining a

based on the understanding developed

under this program.

Two

basic

situations

determining the SCGMC:

are

available

to

should be

considered

in

(1) app'icable fractographic

define

the

SCGMC and (2)

results

an analytical

crack growth program must be used

to

determine

because

is

not available for the

applicable

fractography

selected design conditions.

and

In

below apply; whereas,

the first

case,

the

SCGMC

principles 1

in the second case principle 3

applies.

Use the same crack size ranqe used to establish the

1.

EIFS distribution to determine the SCGMC

conditions
transfer,
crack

(e.g.,
etc.).

load spectra,
For

example,

for

given

stress level,

and bolt load

suppose

fractographic

the

size range used to determine the IFQ model parameters

was: aL = 0.020" to au = 0.050".

Then,
this
should also be used to determine the SCGMC.

2.

design

same

range

The same crack growth law used to determine the EIFS

distribution,
compatible

Fa(O)(x),

SCGMC.

For

should

be

example,

used
Eq.

to

determine

3-16 can be used to

5.4

<,?:~~~~~~~~~~~~... ...........: .,.-.......-............. .................

.....

,...............

.....

.....

define the SCGMC if

(Ref.

Eqs.

3-8

Eq.

3-2 has been used to define

and 3-15).

In Eq.

refers to the i th stress region.

same

"b"

value

3-16,

the subscript "i"

The SCGMC should have the

as the EIFS master curve(s).

Furthermore,

the Qi and bi parameters for the SCGMC should be

for

the

same

fractographic

Fa(O(x)

crack

size

determined

range

used

to

determine the IFQ model parameters.

3.

If

an analytical crack growth program [e.g., -0,5-38] is

used to determine the SCGMC,

should

first

curve(s),

Eq.

be

"tuned"

3-11.

then the crack

to

growth

the "ncrmalized" EIFS master

The EIFS distribution

established

be based on several fractographic data sets.

master curves.

After tuning,

conditions.

The

Qi

the

selected

the crack growth program

can be used to predict a(t) versus t values for the

service

may

Therefore,

analytical crack growth program should be tuned to

EIFS

program

desired

and bi parameters in Eq.

should then be fitted to the a(t) versus t

results

3-16

in

the

section

for

designated crack size range: aL to aU.

5.2.2

Illustrations

N!

Two

examples

are

presented

in

this

determining the SCGMC for the following cases:

.'N - .,.

-.-.

"-"

"-

. " -"-"'

'''

'''

..-

, '

'

-,

"-'."

-?

"- '-'-''

5 .5

' ,

, -V

'

'

,-.

. .

- .

. . -

Case

y 1 i(r)

can

be

defined using applicable

fractographic results.

Case

an

analytical

crack

required to determine the

growth

SCGMC

for

program is
the

desired

load spectrum and stress levels.

5,2.2.1

In

Case 1

this

case,

app]licable

fractographic

data

are

available for the desired material and load spectrum but not
for

the

desired

stress

levels.

It

is

assumed

that

applicable fractographic data are available for two or

more

stress levels.

'Lhe

.-uck

growth

parameter,

Qi'

in

Eq.

de-ermined using the Q., values for the three

""

data",2 sets shown in Table 4-10.

""

a is

is

shown in Fig.

also shown.

can be

.o

3-18 can be

fractographic

A plot of log ^Q

versus log

4.17 and the empirical equation

In this case,
-armine

the

the SCGMC,
y 1 j(r)

for

given by Eq. 3-17,

value

for

selected

stress levels.

The

empiricF

relationship,

very well in this case.

given by Eq.

3-18, worked

Equation 3-18 also worked very well

5.6
-N
.

-.

* .

*-

for

selected

straight-bore

and

countersunk fractographic
of this program

data sets investigated under Phase II

However,

there's no guarantee that Eq.

for all fractugraphic data sets and

existing fractographic data is

5.2.2.2

further

evaluation

of

needed.

an analytical crack growth program [e.g.,

"35] is used to determine

the

SCGMC

because

fractographic

is not available for the desired load spectrum,

level,

and % bolt load transfer.

EIFS

distributica

has

crack growth program

IFQ

has

been

fractographic

3-11

sets

that

established
the

Also,

and

stress
the
using

analytical

to the EIFS master

the

Qi

parameters

in Table 4-10 (for example).

in

parameters
data

been

assumed

"tuned"

i in this case)

model

is

results.

curves - represented by Eq.

(note:
k

It

already

fractographic

applicable

The

3-18 will be adequate

Case 2

"In this case,

data

[16].

Table

AFXLR4,

4-10

are

based

on

AFMMR4 and AFXHR4 and the

procedures described in subsection 4.4.2.

SCGMC

is

needed

for

load

maximum stress level of 42 ksi.

growth
"

program

can

be

used

spectrum

"A" arid for a

A suitable analytical crack

to predict the a(t) versus t

values for the specified conditions.

5.7
x_%.
, -"",,' .","
"

," -"

-".", -"- - . -'- " -"

-... "

' " "

-"-

-.. - "*
. -.

; " " " , ,.

L . V .

-.

.[ .*.., ....

Analytical

crack

growth

shown in Table 5-1

for

results

load

apply

to

results for a(t) versus t are

0.020"

a(t)

0.050".

These

spectrum "A" and a = 42 ksi.

Note

that the predicted a(t) versus t values cover the same crack
size

range

used

to

determine the IFQ model parameters

in

Table 4-10.

The SCGMC,
Table

5-1

4.4.2.1.

and
For

usinq Eq.

Eq.

3-17,

the

can be determined using the data in

procedures

example,

Qi

in

described
Eq.

Qi

1.697

In this case,

was obtained.

crack size xj, in the range from aL to au,

the

corresponding yji(T)

subsection

3-17 can be deterrined

4-6 and the data from Table 5-1.

"best-fit"

t,

in

For a given

and service time,

can be determined using Eq.

3-

Two examples for

(r), based on Eq.
3-17 and Qi
-4
1.697 x 0,
are given below for different x, and T values.
17.

S'.67x

o W=

0.025",

o x0 = 0.035",

4000 flight hours : y 1 i(7) = 0.0127"

8000 flight

5.8

hours

i(r)

0.0090"

Table 5-1

Analytical Crack Growth Results

for Spectrum "A" (o =42 ksi)
a(t)
(In.)

(Flt. Hrs.)

0.0200
0.0209
0.0220
0.0232
0.0246
0.0267
0~.0297
0.0336
0.0383
0.0427
0.0486
0.050

0
00500
1000
1500
2000
2500
3000
3500
4000
4500
5000
*5114

5 9

-'.
CRACK EXCEEDANCE PREDICTIONS

5.3

analytically predicted for

given

x1,

and

distribution

and

size,

crack
EIFS

3-9,

Eqs.

can be determined using

p(i,r)

applicable SCGMC,

For

r.

time,

"service

given

can be

p(i,r),

exceedance,

crack

of

probability

The

3-17 and 3-19.

Suppose

EIFS distribution has been established and

the

the following parameters from Table 4-10 apply: a0

1,8G5,
1=

xu = 0.025",

QOZ.m 2.155 and

load spectrum "A" and a = 42 ksi

where

these parameters

predicted,

for

at x,

this case,

hours.

In

0.080.

Therefore,

Yli(r)

by

defined

= 0.00772"

Eqs.

damage

of

extent

variance for different

using

stress

3-20 through 3-23.

described and illustrated

and

the

S~5.1.0

p(i,r)

be

expected

8000 flight hours.

and

regions

the corresponding
can

be

determined

Extent of damage formats are

in Section 5.4.

be

approximately 8% of the fastener holes in

to have a crack size 0.030" at r

overall

Using

0.030" and r =-8000 flight

the specified stress region (' = 42 ksi) would

The

3-17

Eq.

can

p(i,r)

equations,

and applicable

example,

The SCGMC for

= 4.

Subsection 5.2.2.2).

(Ref.

1.697 x 10'

Qi

is

-t

-- 0.035",

EXTENT OF DAMAGE FORMATS/ILLUSTRATIONS

5.4

to quantify

objective of the durability analysis is

The

the "extent of damage" as a function of

formats are described and illaj.3rated in

Various

details.

selected

for

time

this section for presentirl the durability analysis rcsults.

Extent Of Damage Formats

5.4.1

example

Several

in

illustrated

formats

Fig.

5.2.

for,

damage" are

of

objective of the
1. to analytically assure design
TI'e

durability analysis results

"extent
basic

ccmpliance with the secified economic life criterion and to

evaluate

design

requirements

and user life-cycle-costs.

5.4.2

overall

The

tradeoffs affecting structural maintenance

Extent of Damage Illustrations

extent

damage provides a quantitative

of

measure of structural durability.

crack

in

relatively

each
small

As long

the

as

largest

detail included in the damage assessme-t, is

and

cracks

such

are

statistically

"independent, the extent of damage can be estimated using the

binomial distribution.
binomial

distributioni,

Eqs.

3-20 and 3-21,

can be used to predict the extent of

damage and its standard deviation,

5.11
-*A

derived from the

respectively.

The extent

1.

'
-Q-

" Service Life

- Stress level

Z"

Stress Level (Fixed Value)

Service Life (Fixed Value)

Average

Average

Percentage

Pircontage

Of Crack
2

Exceedance

or.

1Of

Crack

T2N

or

"

Ave.

Ave. No. Of

'r-

Exceedance

a 3o

No. Of

Details With
x1

Details With
aa

Crack Length, x,

P -

Crack Length,

Exceedance Probability

xI

For Given x,

T &4aare fixed values

* I~

'Per-entage

Average
Percentage

Tof

Crack
Exceedance

Of Crack

"Exce-sdance

or

or

No. f
Deta.
..
With

ve. No. Of
Details With

'C9

xI

L
Crack Length, x,

i.a

Flight Hours,

III

.. i-'

Damage

Rult

"

i:xI

P - Exceedance Probability

&ZTare fixed values

r-

fixed value

"."1.0

Percentage
Crack
Exceedance

eaOf Coat
Replacement $

j##

or
No. Of
Details
With

Flight Hours, 2"

Flight Hours,

Fig. 5.2 Formats for Presenting Extent of Damage Results

(Continued)

5.13

L,

Tel.0

of

damage

components,

for

the

desired combinaticn of details,

can be determined using Eqs.

etc.,

Three examples are presented to illustrate

3-22

damage

for

different

levels.

and 3-23.

how Eqs.

through 3-23 can be used to quantitatively define

of

parts,

3-20

the extent

The following situations

are considered:

Example 1

One detail

two stress regions;

type:

extent of damage for one control area.

Example 2 -One

detail per airplane; extent of

damage for fleet.

Example 3

Three different detail types; different

stress regions; different crack exceedance
crack sizes(x

); extent of damage for a

component.

Example 1

01

An extent of damage estimate is desired for a wing lower

skin at a critical
holes for x, - 0.0r."

"-5,14

control

area

containing

at = r= 8000 flight hours.

90

fastener

The skin is

divided into two stress regions.

p(i,r)

values

are

assumed

For illustration purposes,

for

each stress region: other

durability analysis details are shown in Table 5-2.

Results

from

Table

5-2 may be interpreted as follows:

"The average number of fastener holes with a crack size 2

at

8000

hours

is

Nip(i,r)

corresponding standard deviation is

average
"at r =

for

7.6

aN(i,r)

x,

holes.

The

2.60.

The

percentage of fastener holes with a crack size > x,

8000

flight

hours

is

Nip(i,r) and oN(i,r),

can be

made

exceedance

for

8.4%.

Us'Ing these average values

upper and lower bound estimates

the

extent

of

damage

at

desired

probabilities.

Example 2

Suppose

an

extenit

of

damage

assessment

average plus upper and lower bound estimates)

one

detail

per

airplane

in

Assume the following probability

been

computed

for

fleet
of

the applicable

(including

is desired for

of 1000 airplanes.

crack

exceedance

initial fatigue quality,

stress level and load spectra for x, = 0.03" and r =

flight

hours:

p(i,r) = 0.05.

Using Eq.

3-20,

10,000

the average

number of details in the fleet with a crack size > 0.03"

(i,r)

1000

0.05

50

has

details

is

fleet.

per

5.15

v~~~~~~~~.................

..--......,,.....-+..o+.......

......-.......

.,.

Table 5-2

Extent of Damage Assessment for Wing Skin

Containing Fastener Holes

NO.

STRESS

FASTENER

HOLES (Ni)

REGION
1

80

0.07

5.6

2.28

10

0.20

2.0

1.26

7.6

2.60

90

"Ave. percentage of details with a crack size

7.6xI00%

x,

90
8.4%

5..

.5.16

,',!.

-*."r

.-.

'-.2"

.-

.-

.2

..-

..

......

.j:

c"

.. r .

' . - -.

-. - .,.

-.

..

-" -. g -

-. ,

" ..

Upper and lower bound predictions

for the extent ot damage

can be estimated for selected probabilities using N(i,r) +

ZaN(i,r),

where Z is

from the mean,

the number of standard deviations,a'N(i,T),

N(ij).

For example,

assume N(i,T)

is

normally

distributed and Z = +3.

From statistical tables for areas under

the normal

Z = 3 and Z = -3

distribution,

bability of 0.0013 and 0.9987,

viation for N(iT),

based on Eq.

prediction for N(i,T)

is

The standard de-

respectively.
3-21,

Using the information above,

6.89.

correspond to a pro-

is:

aji,r)

=[50(0.95)]=

the upper and lower bound

70.67 and 29.33,

respectively.

These results may be interpreted as follows:

The pro-

bability of exceeding 70.67,

50 and 29.33 details with xI -0.03"

at T = 10000 flight

P = 0.0013,

hours is

respectively (with 50% confidence).

0.50 and 0.9987,

This information provides

average as well as upper and lower bound estimates for the

extent of damage for the fleet.
Example 3
An
holes,

aircraft
fillets

component
and

contains

cutouts

in

countersunk

selected

governing

the structures durability.

example is

to show how the extent of

fastener

control

areas

The objective of this

damage

for

different

details with different crack exceedance crack sizes (x,)

can

be combined to quantitatively define the overall damage

for

the

component.

Details

shown in Table 5-3.

component

(two

of

the

The total

control

areas)

analysis and results are

number
with

of

details

the

a crack size 2 x, is

estimated to be 15 with a standard deviation of 3.72.

5.17

in

-,

'3

',

0-

O4O0 w-4

CL.

--

in

V-4

Cr

r4

014
U~-4

'

0C

0
.1-

T-40 C14u

C-I

94

Oev-4

.0

v-4 go4

41L4
0~~
'

E-4
'4--4

1C

$4

.,

4
0

444

01

$44

L2

E4J
14C.
444

C1,-

bOn'

1..4L

04
rn
S141

~41*5.18

60
41

ccI

-94

>1

5.5

5.5.1

ADDITIONAL CONSIDERATIONS

Details Other Than Fastener Holes

durability analysis method described in Section III

The

applies

theoretically

susceptible

details

to

holes, lugs, cutouts,

i

different

to

fatigue

types

cracking

fillets, etc.).

of

structural

(-.g.,

Although

fastener

the

method

fastener holes,
has been demonstrated for clearance-fit
ftrther research is required to verify the method for other
detail types and for" different combinations of details.

Theoretically,

the

IFQ

for a given detail type can be

quantified using applicable fractographic

desired

detail.

Crack

diffe;.ent detail
levels,

data

initiation
types

for

should

different:

data should be used to verify and refine,

fillets,
be

the

cutouts,

generated

materials,

load spectra, manufacturing techniques,

durability

for

Suitable specimens need to be designed for

acqu:,ring fatigue cracking data for lugs,

etc.

results

if

for

stress

etc.

required,

These
the

analysis method for those details which may have

significant

effect

on

the

requirements and economic life.

N.!
5.1

5.19

structural

maintenance

,.r...

_ ,, - .-,.
Yr. -

.i-

--

i...

'*1

5.5.2
The

Large Crack Sizes

durability

analysis

method

was

developed

for

predicting crack exceedance for relatively small crack sizes

(e.g.,

0.10")

in structural details.

The largest crack in

each detail was assumed to be statistically

independent

justify using the binomAl) Oistribution for details.

If

to
the

largest crack in

a give:

ot-.il doesn't significantly affect

the

crscks

ir
t neighboring details, perhaps the

growth

of

proposed durability analysis method can be extended to crack

sizes > 0.10".

The crack growth power law (Eq.

the IFQ model may not be

suitable

cumulative

for

distribution

for

crack

defining

sizes

functional forms for crack growth rate may

justify

used in

the

EIFS

> 0.10".
be

Other

required

to

the same EIFS distribution for both small and large

crack sizes.
predictions

Eq.
for

3-2 may be acceptable

larger

curve" is curve fit

result

3-2)

in

crack

sizes

for
if

making

p(ir)

the "EIFS master

to the larger crack sizes.

This

would

a different EIFS cumulative distribution for the

small and the large crack size range.

5.5.3

Effects of Scale-Up and Hole Interactions

In general,
specimens,
structures

experimental crack growth results for coupon

full-scale structural
exhibit

scile-up

components

and

prototype

and interaction effects.

5.20

The

'

possible

effects

distribution,

of

scale-up

on

the

VII.

demonstrations

EIFS

of

A preliminary investigation of the effects of

scale-up on the durability analysis has been made

further

or

based on coupon fatigue test results, are not

accounted for in the durability analysis

Section

IFQ

research

is

required.

(16]

but

Major sources of scale-up

effects are:

. Increase in the number of fastener holes

* Change in stress field

"-'

Increased variability in workmanship

. Increased variability in material properties.

The

interaztion

of

the

dominant

crack

in

neighboring fastener holes on crack exceedance predictions

have not been evaluated under the present program. Such

K-'.

effects are
largest
...

effects

(e.g.,

not

crack

considered
in

to

fastener

be
hole

significant
is

0.10".

5.21

'4l

the

relatively small

:5 0.10") and the spacing between fastener

considerably larger than

when

holes

is

.75

5.5.4

Functional

Impairment

"The durability analysis method can also be used to

investigate

functional

impairments

For example,

ligament breakage.

through-the-thickness

such as fuel

a leak may occur when a

crack develops in

a fuel tank.

frequently originate at fastener holes.

-

leakage and

Therefore,

resistance of the structure to functional

Cracks
the

impairment due to

fuel leaks can be estimated from the predicted number of

"fastener holes with a through-the-J.1 ,ickness crack.

The durability analysis method has been verified only
for rnlatively small crack sizes
hole).

Since through-the-thickness

"further work is
crack sizes.
*

(e.g.,

< 0.10"

in

a fastener

cracks may exceed 0.10",

required to verify the method for larger

Also,

through-the-thickness

be further investigated for fillets

assess structural

durability.

".4..-.

.- 2

type cracks need to

and other details to

A9

10

SECTION VI

COMPARISON OF DETERMINISTIC AND PROBABILISTIC

APPROACHES FOR DURABILITY ANALYSIS

INTRODUCTION

6.1

',S
fracture mechanics approach (PFMA) for
A probabilistic
durability analysis has been developed.
The deterministic
crack growth approach (DCGA)

was used to analytically assure

the Air Force's durability design requirements for the

For several years now,

[39,40].

airframe
well-known

DCGA

have

used

been

F-16

variations of the

extensively

for

damage

tolerance analyses.

The

objectives of this section are to: (1) conceptually

describe the DCGA used

analysis,

(2)

compare

for

the

F-16

essential features of the DCGA

the

with the PFMA developed under this program and

A,

(3)

discuss

compare the type of output that can be obtained using

~the two approaches and their significance for quantifying,,
and

'A

durability

airframe

"structural durability."

6.1

A hypothetical durability problem is used to explain the

essential

features and differences

in

the DCGA and the

PFMA

for fastener hole applications.

durability

been documented
F-16

analysis

[12,13].

durability

state-of-the-art

assessment has

Details of the DCGA used

analysis

are

given

in

Refs.

for

the

39 and 40.

Details of the PFMA are documented herein and elsewhere

[14-

21].

6.2

Two

U.

different

analysis of

the

growth approach
[ 9,10].

F-16 DURABILITY ANALYSIS APPROACH

approaches

F-16
(DCGA)

were used for the durability

airframe:
and (2)

(1)

deterministic

crack

conventional fatigue analysis

Details of the F-16 airframe

durability

methods are given in References 39 and 40.

only the DCGA will be considered.

Also,

In
the

analysis

this section,
approach

will

"be discussed for the durability analysis of fastener holes.

the

The essential
features of the DCGA approach, used for
durability analysis of
the
F-16
airframe,
are

conceptually described in

Fig.

6.1.

For fastener holes,

basic objective of the F-16 durability analysis was to

that

no

fastener

hole in

the
show

a part or component would have a

6.2

'DL-

CRACK

Xi

T=1Service Life

Deterministic Crack Growth

TINE

Fig. 6.1

*1

F-16 Durability Analysis Approach

6.3

crack size greater than, the repair,

life.
,

aRL is

aRL,

after one servicc

the maximum crack size in a fastener hale that

can be cleaned up by reaming the ,hole .o the next fastener

size.

Typical aRL values range from Cl.03" to 0.040".

Each fastener hole was assumed to have an initial flaw.

F-16 durability analyses were performed using an initial flaw

size of either ai =j 0.005" or 0.010".

These initial flaw sizes

are based on the results of the F-4 tear-down inspection

(7].
The fol]owing general procedures were used to evaluate
the durability of fastener holes in

the F-16 airframe.

1.

Select part

2.

Divide a part into control areas or stress regions:.

3.

Group the structural details

for durai].Iity analys i s

(e.g.,

fastener holezl-

according to applicable stress region,

4.

Select the most critical detail in each stress

region for durability analyses.

6.4

5.

Assume

an

initial

flaw

size

critical detail in each region.

is

size

(ai)

for the most

The

flaw

initial

"re3presentative" of the initial quality of

the detail.

6.

a suitable deterministic crack growth computer

Use

program,
at

[e.g.,

36] to grow ai to a crack

service

specified

life

size

x1

for the applicable

maximum stress for each stress region.

7.

Show

for

the

mcst critical detail in each stress

region that x, is S aDL (durability limit flaw size

for

functional

impairment and/or economic repair)

at r = 1 service life.
size

in

aDL is

a fastener hole,

the

maximum

for example,

that can be

economically repaired by reaming the fastener

hole

aDL = 0.03" was commonly used.

to the next size.

-.

crack

6.5

rrrrrwn

MEHNC-APOC

'S

PROBABILISTIC FRACTURE MECHANICS APPROACH

6.3

is

PFMA

The

5-."

described

conceptually

details are given in Section III

approach

will

be

further

of

this

in

6.2 and

Fig.

handbook.

This

described later for the example

problem.

6.4

durability

containing
assume

.-

200

chat:

region

analysis
fastener

(2)

holes.

are equally-stressed,

each

required
For

all fastener holes

"holes in each region is

for

is

for

analysis

structure
purposes,

(1) the fastener boles are grouped into three

ztrers regions,

EXAMPLE DURABILITY PROBLEM

fastener

and (3)

known.

hole

is

If

in

given

stress

the number of fastener

the economic repair limit

0.03",

how

"durable" is the

structure at the end of one service life?

Conceptually

(scribe
'.

the

durability analysis of this

structure using the DCGA and the PFMA.

discuss

the

type

6.6
4-4

T*

5.

KK72.

5 *--

-- y "'.~5

compare

arid

of information that can be obtained from

"each approach.

T*

Then,

X1'1

CRACK
SIZE

SERVICE CRACK GROWTH MASTER CURVE

-,--

K/

LIPS OISTRIBUTION

TM

L Initial Fatigue Quality

"

TIME

Fig.

6.2

Probabilistic Fracture Mechanics Approach

VN

--

'-

--.

'SA

6 .7 .

6.4.1

The

Durability Analysis Based on the DCGA

fastener hole from each stress region

"worst-case"

is used for the durability analysis.

Using

the

DCGA,

analysis proceeds as follows for each stress region.

an initial flaw

ai,

size,

is

assumed

to

exist

flaws

[e.g.,
stress

using

typical

the

initial

is predicted for each stress

a deterministic crack growth computer program

36,37],applicable material properties,

level,

in

The size of the initial flaw, ai,

in fastener holes.

at the end of one service life,

region

First,

The size of the

fastener hole in the most adverse position.

flaw is considered to be representative of

the

and

load

spectra.

analysis, for illustration purposes,

the applicable

Assumed results for the

are summarized in Table

6-1.

The

assumed

results

shown

in

Table

6-i

All fastener holes in

interpreted as follows.

each

can

be

stress

"region will have a crack size less than the economic repair
5-.

"limit [ e.g.,
"The following

aRL = 0.03"] at the end of one

can

also

service

life.

be stated for stress region I: at

least 1 out of 100 fastener holes will have a crack

size

0.020" at the end of one service life.

Similarly, at least 1 out of 50 fastener holes in stress

regionsIl

and III will

0.015", respectively,

-l.%

have

crack

size

after one service life.

6.8
----------------------------------------------------------------------------------------------

0.027"

and

7,--

Table 6-1

Illustration
of the Deterministic Crack
Growth Approach and the Type of Information
Obtained from the Analysis

Stress Region

No. Holes/Region

100

II
III

I.

SJ
,

6"A

6.9

.010"

Max Crack Size 1

x, @TSL

0.020"

50

0.027"

50

0.015"

200
SL = Service Life

ai

Durability Analysis Based on the PFnA

6.4.2

Using

the PFMA,

follows.

First,

cumulative

El

initial

stress

using

24,25] and

the

is

region.

available

determined

described

stress

predicted

for

crack

distribution,

For

each
Fa(0)(x)

shown

for

region.
oid

and

EIFS

results [ e.g.,
illustrated

master

using

Then,

region

the

curve,

the

in

details

standard

is

procedure

probability

using

the

of

cumulative

purposes,
are

probability
assumed

Table 6-2.
procedures
on

for

Refer to Eq.
described

computing

p(i,r)

in

of
each

crack
stress

3-19 and the

Section

3.5

for each stress

Next, the average number of fastener holes

the

in

and the applicable SCGMC.

p(i,T),

"durability analysis
herein

stress

illustration

as

are

size, xI = 0.03" at one service life is

"exceedance values,

region

or

parameters

and

region

described in Section 5.2 herein.

exceeding

(IFQ)

A service crack growth master

compatible with the

each

model

fractographic

procedures

for

quality

defined for the fastener

IFQ

IV of this handbook.

curve (SCGMC),

fatigue

distribution

determined

Section

the

EIFS

holes in each

the durability analysis is performed as

N(i,r)

deviation

o(i,r) for each stress region

N
with a crack size greater than xI = 0.03" at T = 1 service
life

is determined using Eqs.

3-20 and 3-21.

6.10

The extent of

zo

00

ON
I

S4

%0
0

r-4

CDU'

-r04C40

r.4'

r-4

L4

AN

ci

$4 0-4

44$

44

H,

rLn

IIW

-r0

0
$4 00
41)

cuW

411
c

4-444Lr00
F-4

-r 0

0000

4-4
~

'41
0

0
1-4

C6
**J

4
P 1. .-

__u_

...7.-..-.

damage

L(r) and its standard deviation aL(r) for the

stress

regions

combined are determined using Eqs.

"3-23, respectively.
""

Results

for

three

3-22 and

the

analysis

the

predicted extent of

described

above are summarized in Table 6-2.

and

,Upper

lower

bounds

for

damage

can be estimated using L(T)

number

of

standard deviations,

ZOL(T), where Z is

aL(T),

from the mean,

the
"(f).

"Extent of damage" predictions are illustrated in Table

V..

for

three

probabilities.

exceeding L(r) - 15.7,

For example,

6-2

the probability of

lu.5 and 5.3 is P

0.05,

0.50

and

"3.95, respectively.

6.4.3

Conclusions

-V.

The DCGA does' not account

for the
initial fatigue
quality variation for the population of fastener holes.
A

single

initial

flaw

size is used to characterize "initial

quality" and the results of the analysis do

not

provide

quantitative description of the extent of damage for all the

fastener holes.
m

The "damage" is determined for a single fastener hole in

a given stress region and it
other

fastener

holes

in

is

assumed

that

all

of

the

the region are no worse than the

6.12

For the simple problem considered,

hole analyzed.

only

the

"worst-case" hole out of 200 holes would have to be analyzed

to show that the size of the crack

in

any one hole would

be

-a

RL.

The

PFMA provides a lot more information than the DCGA.

the PFMA provides the following

For example,

number

Average

deviation

in

of

fastener

information:

holes and its

standard

each stress region with a crack size

aRL.

The

extent

of damage and its

standard deviation for

the population of fastener holes.

Upper

and

for the extent of damage for

lower bounds

selected probabilities.

The

information

the

"extent

of

above
damage"

gives a quantitative description of

as

function

of

time.

This

judge
the durability of the
can be used to
information
structure, to assess structural maintenance requir'ements and
costs,

and to evaluate durability design tradeoffs.

6.13

%..".

m .

',.

-. '.9,

--A'.:

SECTION VII

DURABILITY ANALYSIS DEMONSTRATION

7.1

INTRODUCTION

demonstration of the durability analysis methodology,

described in Sections III-V, is presented

for:
N

in

this

section

(1) the F-1.6 lower wing skins and (2) a complex splice

subjected to a B-i oomber load spectrum.

correlated

vrith

test

data.

Both analyses

are

Also, the durability analysis

results and practical aspects are discussed.

7.2

durability

F-16 LOWER WING SKIN

analysis

of

(durability test article) is

methodology described.
of damage

in

each

the

F-16

presented

to

lower wing skins

illustrate

the

Analytical predictions of the extent

wing

skin

are

presented

in

various

formats, and results are compared with observations fromn the

tear-down inspection of the F-16 durability test article.

The

F-16

durability

test article was tested to 16,000

flight hours (equivalent to 2 service lives)

hour

block

spectrum.

using

500--

Each wing received the same loading.

7.1

Following the test,

skins

were

inspected

"Fastener holes with

fractographic

using

crack

eddy

The

respectively,

confirmed

hand

and left hand

twenty

wing

preliminary
skins

reflected:

and

seven

with a crack size 0.03" at r

was

durability
presented

fastener

(1)

results for protruding

rates

for

analysis

in

Ref.

hole

IFQ

head

for the F-16 lower

19.

This

fasteners,

the IFQ model based on Eq.

(2)

parameters

level,

F-16 400-hour block spectrum,

regions

on

3-2 (bil),
IFQ

growth

(3)

three-

model,

Ref.

25),

(5)

considered for the lower wing skin,

an analytical crack growth program

block spectrum

crack

(4)

single data set (one stress

model

stress

based

analysis

based on fractographic

parameter Weibull distribution used in the

(35]

and

three
and (6)
500-hour

used to define the "service crack growth

master curve" for each stress region.

Essential

features

fractographic results
quantify

IFQ

of

the

for

present

countersunk

(countersunk

durability test article),

analysis are:
fasteners

used

(1)
to

fasteners were used on the F-16

(2)

7.2

,i-i--

six

by

16,000 flight hours.

09

techniques.

were

right

lower wing skins were found to have

wing

lower

the

current

indications

evaluation.

fastener holes,

in

all fastener holes

crack growth rates for the IFQ

model

based

on

Eq.

3-2 (b=l),

distribution used in the IFQ

based

on

pooled

model,

three-parameter Weibull
(4)

model

parameters

fractographic results for three different

data sets (three stress levels,

lower

(3)

400-hour block spectrum), (5)

wing skin divided into 10 stress regions,

and (6)

service crack growth master curve for each stress

based

on

Eqs.

results.

3-17

region

and 3-18 and applicable fractoqraphic

There were no significant differences

in the

400-

hour and 500-hour spectra.

The

F-16

lower

wing

regions as shown in Fig.

the

corresponding

skin was divided into ten stress

7.1.

number

were

based

analysis results.
were

on

and

of fastener holes in each stress

region are shown in Table 7-1.

I--IV

Applicable stress levels

strain

The stress levels for

gage

The stress levels

Zones

data and finite element

for

Zones

V,

VII-IX

determined using a coarse grid finite element analysis

az 9 a theoretical stress distribution for a circular hole in

an infinite plate under uniaxial tension.

The stress levels

for Zones VI and X were based on a fine grid finite

element

analysis.

Fractographic results for three data sets (i.e., AFXLR4,

AFXMR4,
32

ksi,

and AFXHR4)
34

ksi,

[24] for maximum gross stress levels

of

and 38 ksi were used to calibrate the IFQ

7.3

VIi

Vill
IVi

Fig. 7.1

Stress Zones for F-16 Lower Wing "kin

704

%.1.

Table 7-1

STPESS
ZONE

Stress Levels and Number of Fastener

-Holes for F-16 Lower Wing Skin

LIMIT STRESS

LEVEL

(ksi)

NUJMBER 01'
FASTENER HOLES

28.3

59

II

27.0

320

24.3

680

IV

16.7

469

28.4

VI

29 .2

30

VII

32.4

VIII

26.2

IX

26.2

12

25.7

20

11614

7.5

The F-16 400-hour block load spectrum was

model parameters.
used.

AFX series specimens were designed for 15% load

The

contained

and

two MS90353-08 (1/4 dia.)

rivets as shown in Fig.

aluminum

countersunk

blind,

typical

All specimens ret2.ect

4.3.

tolerances

quality,

production

aircraft

7475-T7351

of

The specimens were made

transfer.

and fastener fits.

Nine specimens were tested per data set.

4-1 (bi=i) was

The crack growth rate parameter Qi in Eq.

determined for each of the three

AFX

determined

from

the

fractographic

square fit

of Eq.

4-1

[16'.

range

data

size

Usirg Eq.

3-13
e.,

the corresponding lower bound of TTCI value,

ao, was determined

for each reference crack size,

data set.

crack

fractographic

An upper bound EIFS of x, =

0.005" - 02.0" was used.

Q = Qj,

was

results usinq a least-

0.03" was assumed for the IFQ distribution.

and

Qi

sets.

for

each

The results of Q* and e are shown it Table 7-2.

Fractographic results for the three AFX s-ries data sets

were combined to deL.ermine
value

using

the

results

crack

(a 0

0.03",

each of the three data sets.

corresponding
procedureF.

followinq

initiation (TTCI)
sizes

the

for

three

0.05",

"pooled"

rt

Time-to-crack-

different

reterince

an-d 0.10") were used for

The TTCI-c

results

Aor

each

reference crack size were normalized for each data set using

LV."

4A*

7.6

"-

"Table 7-2
""

Sunmary of IFQ Model Parameters fo F-16

400-Hour Spectrum

PMAY
"DATA,

SET

(KS1l

a0

Q*10

(INCH)

(ORSS)I

0.03
AFXLR4

32

0.05

1.201

0.10

0.03

AFXMR4

34

AFXIIR4

38

(ICH
(KSS)
(I)(j

[0

15,033

1.805

4,253

12,916

1 551

10,025

13,421

1.612

8,721

1.777

0.05

2 .037

1.823

2,508

7 ,759

1.581

0.10

5,910

9,093

-. 852

0.03

5,469

2.587

1,079

5,098

2.412

2,545

4,598

2.175

0.05

4.731.

0.10

POOLED

1. 823,

AVERAGE

Q:A

1.928

Notes
1.
2.
3.

xu
0.03"
0.005"'a(t)S0.10'

(fractographic crack size ra:,.ge u~sed)

-4

797

A:
-., " - .: : ' : - : , :: ' , " -' "- : ._. ' ' ' , , , ._" 'i ..: . ? "?,j . "- "

"i ._'i - ; . ' "_1. '

: 2.: : : .1: 1 '' -, -': ' : ': '1 - -

K'

the average TTCI values (T).

Results

sets

and

were

pooled

together

for

the

the

three

data

(TTCI-e)/X data were

ranked in ascending order.

Equation

into

form to determine the pooled

least-squares

value (16].

fit

3-1

was

The pooled value was found to be

transformed

1.823

(Table

7-2).

After

determining

for

the

pooled

d.ata sets, the

adjusted TTCI's for each reference crack size for each

set

were

considered

separately

corresponding t3ivalues [16).

Table

7-2.

Also

to

data

determine

the

These values are presented in

summarized

in

Table

values for the nine cases considered.

7-2 are the Qihi

For generic EIFS,

the

Sand

Q OZ. values should be constants.

1.928 and a. = 1.823 are used

analysis.

for

A plot of Qi versus 0.,

the

An average Qiki
present

durability

is shown in Fig.

7.2 for

the three data sets considered (9 cases).

SCGMC

shown in Fig.
this

cese,

is

needed

for each of the ten stress regions

7.1 to compute the corresponding

suitable fractographic

determine the SCGMC for each stress

the

results

from

Table

generalized SCGMC based on Eas.

7.8

7-2

p(i,r)

In

results are available to

reqion.
can

For

example,

be used to define a

3-17 and 3-18.

~ ~ ~

W,-~~
7T

7KM'c7
7,-.

w4
AFXHR4

2e
0U

cc

AFL

0.

00

=0.01

00

1.00
SYM0

10

310

Fig. 7.2

VersusR

for the F-16 400-Hour Spectrum

7.9
7-

30S

In

Fig.

7.3,

in

Qi

3-18 can

parameters t and 7 in Eq.

least-squares

fit

plotted

is

and

the

be

determined
of

results

U:

in
Eq.

using

The

is

given

7-1.

Qi

1.427 x 10-16

7.928

(7-1)

can

A generalized SCGMC, as a function of stress level,

be obtained by substituting Eq. 7-1 into Eq. 3-17.

Crack

exceedance

as

the

for

predictions

skin were determined using Eqs.

well

no superscript

expressed in ksi units,

used for a SCGMC),

is

The

7.3.

Fig.

resulting crack growth equation for Qj (note:

"-""

in o-.

against

3-9,

the F-16 lower wing

3-17,

3-19 and

following parameters: xu = 0.03",

Qiki= 1.928 (average),

I = 4 and

various

7-1

as

a = 1.823,

values.

The

results are presented in various formats as described below.

The extent of damage predictions for the F-16 lower wing

skin are summarized in Table 7-3 at
for

each

16000

of the ten stress regions shown in Fig.

number of fastener holes with a crack size

and the standard deviation,
"and 4.077,
right

hand

flight

respectively.
and

left

aTr(),

7.1.

0.03",

hand lower wing skins,

7.10

The
L(r),

were estimated to be 17.6

Based on the test results for

w'

hours

the

an average of

GROSS
150

MPP

STRESS,
200

250

AF

C.)

4.

300

HR4

.I

AFX MR4

LU

AF I LR4

I.0
<

O" MP 0

7.92

0.

I-I

20
GROSS

.4.

Fig. 7.3

40

30
STRESS

(KSI)

Q. Versus Gross Stress for the F-16

46 0-Hour Spectrum

7.11

50

lia

Table 7-3

Durability Analysis Results for

F-16 Lower Wing Skin

NO.

STRESS

REGION

(HRS

HOLES WITH

16,000 HRS

P)RFD IICTED

"

N(i,T)

O-N(1,)

TEST
R.H.

WING L.H.

WING

AVERAGE

0.4620

0.0426

2.5

1.547

3.5

II

0.3182

0.02182

6.9

2.598

4.5

111

0.1380

0.00480

3.3

1.812

2.5

IV

0.0071

0.00002

0.0

0.

0.4751

0.0448

0.4

0.618

0.5

Vi

0.5921

0.0662

1.9

1.332

3.0

VII

1.3504

0.2649

2.1

1.242

1.0

VIII

0.2507

0.0142

0.1

0.314

1.0

Ix

0.2507

0.0142

0.2

0.444

0.5

0.2152

0.0108

0.2

0.445

L(T)

17.6,

TLL(T)

4.077,

.2

r7.1

0.03"11

Q0 x 10

TOTAL

TEST AVERAGr

16.5

p.

16.5 fastener holes had a crack size,

hours.

In Table 7-3,

0.03" at

16000

the predicted extent of damage results

track the average test results

for

the

individual

stress

regions very well.

In

Fig.

7.4,

the

predicted

percentages

exceedance versus fastener hole crack size are

x 100%/N*,

1, 2 and 3 are based on L(r)

100%/N*

and

and aL(T)
is

[L(r)

x 100%/N*,

-uL(T)]

are defined by Eqs.

3-22 and

crack

plotted

F-16 lower wing skin at r = 16000 flight hours.

the

N*

of

[L(r)

Curves

aL(T)]

respectively.

3-23,

for
R'

L(O)

respectively.

the total number of fastener holes in the F-16 lower

Since the number of

wing skin (i.e.,

1614 holes).

holes

stress region is

in

each

approximate

the

distribution.

binomial
The

for curves 1, 2,

large,

it

distribution

corresponding

3 are shown in Fig.

is

fastener

reasonable to

by

the

normal

exceedance probabilities
7.4 in parentheses.

Test results for the right and left hand lower ving skin
(at x.
1
circle
.P

0.03" and r
and

square,

16000

hours)

are

respectively,

plotted
in

Approximatley 1.1% of the fastener holes in the

Fig.
F-16

as

wing

%1

16000 hours.

This compares with an average of

7.13

CIA'

7.4.
lower

skin are predicted to have a crack size 0.03" at r

1.02% based

%
V

-W

r.-

"CRACK SIZE
"

"

10 . 00

(Mm)

' .............
...

0. C

50% CONFIDENCE
1.928
*u

jQW

003"1

x=

",''

F~2

-t4,-

-'

w
4

M4

\..\
\

it1

N-~

0.1

CL

SYMBOLS

0 R.H. LWR WING SKIN

a L.H. LWR WING SKIN

0.01I

(84.13 %)

II

0.02

0.03

CRACK

Fig.

TEST
RESULTS

7.4

0.05
SIZE

0.07

O.O0

(INCH)

Percentage of Crack Exceedance Versus Crack Size

"at 16000 Hours for 3 Probability Levels (F-16
Fighter)

7.14

Pb

'V

test results for the right hand and left hand lower wing

on

skins.

Fig.

In

the predicted average percentage of crack

7.4,

exceedance decreases rapidly for larger

crack

sizes.

For

the average percentage of crack exceedance for the

example,

fighter lower wing skin decreases from approximately 1.1% at

x

0.03"

to

approximately

0.14% at x, = 0.05".

Crack

exceedance predictions are based on the service crack growth

master

curve

defined

by

Eqs.

3-17

and

3-18.

A single

service crack growth master curve may not adequately fit

full

range

of desired crack sizes for all crack exceedance

predictions.
master

For example,

curves

different

are required to fit

ranges as illustrated in Fig.

shown

in

Fig.

7.5

apply

7.5.

service

crack

Curve

and

Curve

Fig.

of

to crack size ranges Al and A2 ,

Crack exceedance predictions based on

growth

two different crack size

respectively.
and

the

curves

7.5 will be different for the same crack

exceedance size, xI.

For example, p(i,r) predictions
based
on Curve 2 for xl, Ti
and x2 , T2 will be larger than those
based on Curve 1.

The

extrapolation

of

crack

exceedance predictions to

larger crack sizes should be consistent with the

crack

growth

process

for

applicable

given design conditions and the

7.15
I.
- .J..Im..

..... ,...

..

. .

.,

.. ;.

. . .

. . . .....

. .

, .

.,

.,

. .

.o.

2
0,

(r
SGGMCQ

C.)
(I,

TIME

Fig.

7.5

%.

Service Crack Growth Master Curves for

Different Crack Size Ranges

7.16
J,.4

.1

crack exceedance crack siz(,

to
-

in crack

better understanding and confidence

develop

Further research is needed

xI.

materials,

exceedance predictions for different crack sizes,

and design conditions.

as

the predicted number of fastener holes in the

this case,

of

function

flight

hours

L(T)

for different exceedance

The plots

holes.

24.3

1.65

L(r)

upper

The

there is

a probability of 0.05

probability of 0.50 and 0.95,

at

There

that more

respectively,

16000

hours.

flight

The

upper/lower bound predictions for the F-16 lower

compare

very

that

fastener holes in the F-16 lower wing skin

than 17.6 and 10.9 fastener holes will have a crack

0.03"

bound

will have a crack size a 0.03" at 16000 flight hours.

is

and

approximately 24.3 fastener

is

ZL(r),

In other words,
than

rnmcre

'L(r)

4.077.

oL(r)

holes,

at r = 16,000 hours,

example,

fastener holes and

prediction,

"

= 1614 fastener

For

ZOL(T).

S17.,

Ni

t = 4,

plotted

are

probability values (i.e., P - 0.05", 0.50, 0.95).

-19," x,
0.3" cc=88 3
= 1.823,28
are based on Eq. 3-19, x = 0.03",
(average),

format.

probability

F-16 lower wing skin with a crack size ; 0.03"

.-.

damage are

of

extent

the

7.6 in an exceedance

"presented in Fig.
In

of

predictions

Analytical

size

average

and

wing

skin

well with test results for the right hand and

7.17

,/ ,.,.'" . <.,'
'.

.- ' ..'. ,' " ". . .''- .,

' .'

', . . .

"

. +'

-. - " +' .

. .

'-..' - '

.' . - .'

'- .

. .

. -"+ ,,,' ' . ., _ ' -- . -

.-.. ' '.-"

",

," '

SYMBOLS

0RH-W

40*

L.H. LWR

ISSKN
WING SKIN

TS

EUT

PREDICTED

50% CONFIDENCE

r30-

Q~e-

0034
1.928

20t

S.'.
N1

oURS5
H

LFLGH
-ND

Fig.

t7.1

7.6

Number Of Holes with Crack Size

0.03" Versus
Flight Hours--Exceedance Probabili~ty Format
(F-16 Fighter-)

left hand lower wing skins at r = 16000 flight

hours

(Fig.

7.6).

extent

The

of

damage

predictions

stress level format in Fig.

"baseline stress
Eqs.
to

3-9,
the

zones.

3-17,

(w),

7.7.

maximum

Curves are shown

1.lr, and 1.2,r.

3-19 and 7-1.

stress

for

level

for

the

Results are based on

"he "baseline stress"

For prediction purposes,

each

are presented in a

refers

each of the ten stress

the baseline

stresses

for

stress zone were all increased by the same percentage.

The results shown in Fig.

7.7 can

extent

of

hours.

This format is particularly

damage

durability
damage.

design
For

approximately

be

used

to

assess

the

as a function of stress level and flight

tradeoffs

example,

at

in

useful
terms

of

16000

r-

for

evaluating

the

extent of

flight

hour5,

1.1% of the fastener holes in the F-16 lower

wing skin would be predicted to exceed a crack size of 0.03"

for

the

baseline

were increased to
percentage

of

stress levels.
and

1.1i-

holes

with

If

1.24,
a

crack

the baseline stresses

the

size

approximately 4% and 12%,

respectively.

quantitative

the

measure

of

structural

function of stress level and flight hours.

7.19

NO

predicted

average

: 0.03" would be
This

provides

durability

as a

SYMBOL S
20.0

R.H. LWR WING SKIN

SL.H. LWRWIGSI)TS
-- PREDICTED
50% CONFIDENCE

TS

dUT
SLT
-

1.2 r

10.0

0Al

%w
00,

-BASELINE

1.0

DESIGN

STRES,

0
1

A,

CL~Q

1.928

=1.823

8000

12000

16000

FLIGHT

Fig. 7.7

_'1:
a_

HOURS

Average Percentage of Holes with Crack Size

S0.03" Versus Flight Hours--Stress Level
Format (F-16 Fighter)

7.20
:1%

Z0000

L -2

7.3

COMPLEX SPLICE SPECIMENS SUBJECTED TO A B-I BOMBER SPECTRUM

durability

analysis

of

complex-spl--e

subjected to
a B-I
bomber
load spectrum
Analytical
predictions
of
the
extent of

specimens

is
presented.
damage
in the

specimens are presented in various formats and compared with

fractographic

results.

The analytical/experimental

results

are summarized here and described in more detail in Ref.

The

complex-splice

Fig.

7.8.

and

countersunk

geometry

is

presented

in

Specimens were made of 7475-T7351 aluminum plate;

spectrum [16,24]
analysis

specimen

20.

and

steel rivets were used.

was applied.

A B-1 bomber load

Based on a simplified

strain gage results,

stress

the maximum gross stress

in th? outer row of fastener holes at the faying surface was

estimated
to be 35.8 ksi.
The eleven sperimens were tested
to two service lifetimes

(27,000

flight

hours)

or

failure,

whichever came first.

After testing,
inspected.

all fastener holes in the outer rows were

Fractography was performed for the largest crack

in each fastener hole in the outer rows.

110 faster.er holes in the outer rows
0.05"
in

at

13,500 hours.

Hence,

Twenty-five out of

had

crack

size

22.7% of the fastener holes

the outer rows had a crack size 0.05" at 13,500

hours.

7.21

........................................

6..........

gyM
- 5.50
N

1.88

MS 90353 -06
22 PLACES
(SEE DETAIL A)

0.84 REF.

TYP.

0.75
2.2

TYP.

1.50

o. 500

0.0o6
0.12

"

AA

0.155T

~~~T

,-.28---

990
MATLLIAL:

0.080 MAX

)01

Note:

7475 - T7351 ALUMINUM

All Dimensions (Inches)

DEAIL

0. 2010

Fig.

7.8

Complex Splice Specimen

72

N-.,

4 50

The

IFQ

fractographic
reference

of

the

fastener

sizes.

The

T7351 aluminum and contained

levels

overall

However,
amount

of

15%,

length

the lug end

of

countersunk

basic

load

transfer

transfer

fractographic

IFQ

with

depending

level.

and

the

on

Specimens

holes.

the

were

Therefore,

the

for specimens varied depending on

axial stiffness

of

test

etc.

crack

size range of 0.005" - 0.1" was

An upper bound EIFS of xu = 0.05" was used

distribution.

for

Since the largest fatigue crack in

one of four fastener holes per specimen

was

used,

subsection 4.5).The same data pooling procedures were

used to

determine

previously
average
2.823,

4.3)

All

Three maximum stress levels were

fasteners

stress level,

considered.

(Ref.

Load

for a given % load transfer assuming a perfect fit

specimen.

the

rivets.

section dimensions.

varied

factors such as fastener-hole fit,

(Fig.

test

dimensions

load transfer.

between the mating

actual

the

30% and 40% were considered.

and

considered for each load

designed

on

specimens were made of 7475-

"specimens had the same configuration

same

based

W3s

results for nine data sets and three different

crack

transfer

holes

the

described

a and Qi

IFQ
for

model
the

values
ri

parameters

which

were

fighter demonstration.

The

were

found

to

be

2.702

and

..

respectively.

7.23
2' ?'

'q '~

2.

' .2

,,.-

, '-g

.. -'',

..

..

- '-

---.-.

.--

'

,.

'..,

. . . ,.-

. - .

gener-lized

obtained

using

fractographic
in Fig.

7.9.

best fit

SCGMC,

Qi
data

based

and

gross

sets.

The solid

on Eqs.
stress

3-17 and 3-18,

(o-)

values

The Ln Qi versus

line

represents

through the plot points.

na- is

the

was

for

plotted

least-square

The dashed lines have the

same slope as the solid line and they encompass all the plot
points.
18)

The corresponding best-fit equation for Qi (Eq.

as a function of

gross

expressed in ksi units is

Qi

stress

level

when

stress

3is

as follows:

6.151 x 10

(7-2)

l3.

Crack

exceedance

predictions

specimens were determined using Eqs.

2..

At

13-500

hours,

for

the

3-9,

complex-splice

3-17,

3-19,

and 7-

an average of 9 fastener holps

(8.3k%) were predicted to exceed a crack size of 0.05".

The

"test results showed an average of 25 fastener holes (22.7%)

exceeding a crack size of

0.05".

The

predicted and test crack exceedances

difference

in

the

is attributed mainly to

the stress level used in the predictions.

The actual stress

level and distribution in the outer row of fastener holes is

far more complex,

due to lateral bending effects,

considered

for the damage assessment.

pr_dictions

are very sensitive to the gross

level

used.

rhis

is

LW
'..I.

The crack exceedance

illustrated in Fig.

7 24

than those

applied
7.10.

stress

The solid

GROSS STRESf, MP10.0

300

250

200

1O
IS

9.0
d01 1.895x1O IT 5.3B1

7.0

//

6.0 /("Me

/-

0'J MPS

//
*

I-

,o.--

S~/
ih

"=

4.0 -

/HR

I-/
Uii

A HARXiR4
/Z
ABZvR
eXMR

a:=
,

Uii

r.)

AXLR4@ /

ABYHR4

5,/

ABZLR4,~

II/"

1.0
cc

20

30

40

GROSS STRESS, KS

.2.

0-

ABXR44

'.,

~ 7.
Fig
*

..

ABYH.

~
.%
Q ~Versus Gross
Stress for rB-I Bomber ~
I

.~

,.

*.A5

Load Spectrum

7.25

N *

.*

**

**I '

-*--

..,. ,,-.

,,,

~~100I

-4

/T
low

00

-,

t ..

..-. .

o~0(1*

"V:''

i-i

--

oz

"cn.

o.1

,.,.,

-.

rdc
30 ~

DT,~ut~T

--

40.

,'.,

...-

-,

0-

.5

2.0

-.

,,"

O2

rE

13.20,2
Ill-'..

w-;1

;.

c,:

loe.

0-ro.-oS-t

Fig.

7.10
l

~ QC: '~
AveraePrenaef
o~u-

2-,j

''

7.26
E.t0,

Q t XA'C. '-.'

I"

Hoi3s

Size
005" VersusFiignt
Level Format (B-I Bomhe')

'"

."--".

24.2

Fl-ot~ts

... St:rz'ss

'

q-

line represents average crack exceedance predictions for the

gross stress level of 35.8 ksi obtained using the simplified
stress

analysis

average

crack

approach.
exceedance

The
for

other

Also plotted as a single point is

exceedance

at r = 13,500 hours.

gross applied stress level

used

38.6ksi rather than 35.8 ksi,

dashed

represents

gross stress levels.

t.ie
It

line

average

test

can be seen that if

in

the

accurate

stress

the

predictions

were

the predicted crack exceedance

at r = 13,500 hours would match the test results.

more

crack

analysis

could

Hence,

result

in improved

formats,

previously

pr-di-i:. -Jns.

,t',e
In...rssd
i...

oseful
fo

crack

exceedance

the fighter demonstration,

an

in

CONCLUSIONS AND RECOMMENDATIONS

Probabilistic

fracture mechanics methods for durability

analysis have been described and

fighter

aircraft

demonstrated

structure

splice subjected to a bomber spectrum.

used

presented

7,12 for the complex-splice specimens.

7.4

Full-scale

are

and

for

both

for a complex

These methods can be

to analytically assure zompliance with the Air Force's

durability

design

requirements.

"7.27

The

analytical

tools

"-4

CRACK SIZE, mm
0.22
0?

100

N'

mi

SMD

LU
LA.I
LUJ

I-J
LU

EXCEEDANCE PROBABILITY

0.95

LU

Predicted

13500 HWs

0 Test Results @QT

a 0 " 0.05"
"O,35.8

Q..W.
S....

0.005

ksi
2.823 (50% Confidence)

'

2.702
0.1

0.01

"CRACK SIZE, INCH

Fig.

7.11

Percentage of Crack Exccedance Versus Crack Size and

Exceedance Probability at 13500 Hours (B-i Bonber)

7.28

.7

1 .f-

.Al

NOTES
50% Confidence
-

Predicted

Test Results @T- 13500 Hrs

"a0 W0.05"
7 - 35.8 ksi

r
,"Q/

60 -_J

Ag
04,

DE

5/
50

^,-.

40

2.823

(AVE.)

-2.702

./

& 30

20

Ol2

u-~

10

P Fxr* danre PrnhakuItv

FLIGHT HOURS, 103

Fig.

7.12

Number of Holes (Outer Row) with Crack Sizeo0.05"

Versus Flight 11ours--Exceedance Probability Format
(B-I Bomber)

7.29

described can be used to quant.ify

tie

extent

of damagc

"ai

function of the durability design variables for structural

details in
life

Once the economic

a component or airframe.

a part,

and durability

critical

are established,

parts criteria

the extent of damage predictions can be used to assure

design

compliance with Air Force durability requirements.

fatigue quality model can be used to define

An initial

the EIFS cumulative distribution

using suitable

fractographic

Procedures and guidelines have been developed

results.

determining

the IFQ model parameters

for

for pooled fractographic

data sets and for scaling TTCI results.

The parameters a and

Qz~i provide the basis for putting fractographic results on a

fatigue

common baseline for quantifying the initial

For generic EIFS,

a and Qii should be constants

data sets

fractographic

levels and percent load transfer.

larther research is

the IFQ distributions

etc.

levels,

for different

required to confirm

for different materials,

fastener

types/diameters/fit,

load spectra,

% load transfer,

A considerable amount of fractographic results exist

7.30

',

stress

Encouraging results have

for crack exceedance predictions

design conditions.

stress

loading spectra,

the use of the same EIFS cumulative

been obtained to justify

distribution

for different

fastener type/fit,

(same material,

technique per data set),

and drilling

quality.

-.

>2:, %

>.t

which need to be evaluated using the IFQ model [e.g.,

The

effects

of

fretting,

clamp-up,

effect (scale-up from coupon to component),

sealant,

'3

interference-fit fasteners,

investigated.
transfer

Also

specimens

the

Thi.,

corrosion,
faying

24].

size

surface

etc.,on IFQ need to be

feasibility

of

using

no-load

with multiple holes to quantify the IFQ

should be evaluated using spectrum

loading.

16,

and

constant

amplitude

could provide an economical way to generate

the fractographic results needed to quantify the IFQ.

Theoretically,

the IFQ model can be used to quantify the

EIFS cumulative distribution for various structural

as

long

as

fractographic

results

are

available for the

details to be included in the durability analysis.

model

has

been

fastener holes.
be

developed

evaluated

generating

crack

initiation

growth results for other details such as,

lugs,

etc.

Fractographic

described

can

IFQ

need

to

and crack

cutouts,

fillets,

results should be developed and

evaluated for such details so that the

methods

The

using fractographic results for

Suitable specimens and guidelines

for

details

durability

analysis

be efficiently applied to different

types of structural details in typical aircraft

7.31

structures.

The

accuracy

of crack exceedance predictions,

based on

the same EIFS cumulative distribution, needs to be evaluated

hi

sensitivity
understand

studies
the

need

average

to

IFQ model parameter

Also,

different design conditions.

for

be

performe&

to

better

parameter values and variances and

the impact of these parameters

on

the

IFQ

for

different

fractographic data sets.

The

durabili~ty

analysis

crack exceedance

predictions

sizes

(e.g.,

0.10")

crack

in

each

",,independent

detail

to

methodology was developed for

for

crack

details.

If

in structural details.

was

assumed

exceedance

the

largest

perhaps

be

crack

The largest

statistically

the

law of Eq.

for

structural

crack in a given detail doesn't

of

proposed

methodology can be extended to crack

growth

to

predictions

significantly affect the growth

crack

small

justify using the binomial distribution for

combining

details,

relatively

cracks

in

neighboring

durability
sizes

>

analysis

0.10".

The

3-2 may not be suitable to use for

crack exceedance predictions for crack sizes > 0.10".

However,
a general service crack growth master curve can be
s ,4.

generated under given design conditions which is valid

"crack sizes

>

0.10" [14-16].

Nevertheless,

has not been demonstrated in the present study

for

this approaci
and

further

7.32

V.f.

*'

-.

-41

research

is

required

to extend the probabilistic fracture

mechanics approach developed to larger crack sizes.

15)

Two

different F

were

presented

equation
(1) It
0,

(x)
for

works but Eq.

equatiGns (i.e.,
representating

3-15 is

Eqs.

the

3-8 and 3-

IFQ.

recommended for two reasons:

assures all EIFS's in the IFQ distribution will be

and (2)

>

the crack growth rate parameter Qi can be easily

determined from the fractographic results and the

Qi

Either

values

for

comparable.

If

different

Eq.

3-8 is

data

sets

will

be

resulting
directly

used, the same b value (Eq.

3-2)

must be imposed for different fractographic data sets to put

*

the Qiji values on a comparable baseline.

all

EIFS's

in

Eq.

3-8

will be > 0.

needed to evaluate the accuracy of Eqs.

The

EIFS

independent of
illustrated

cumulative
the

in

distribution

for

3-8

different

Further studies are

3-8 and 3-15.

distribution,

reference

Eqs.

As long as b > 1,

crack

and

3-15.

reference

size,

Fa(O)(x);

is

This

is

a0.

Therefore,
crack

the TTCI

sizes

will

transform into a common Fa(O)(x).

The

resulting

IFQ

model

EIFS's

is

must

simply

"mathematical

tool" for

be considered in the context of the

7.33

f.%'

IFQ model and the fractographic results

the

model

parameters.

EIFS's

hypothetical cracks used for

used

should

crack

be

to

calibrate

considered

exceedance

as

predictions

rather than actual initial flaws per se.

Back

extrapolations

of fractoqraphic data must be done

"consistently to put the EIFS's

different

data

obtained if

sets.

the EIFS

extrapolating

the

if

results

Inconsistent

common
EIFS

distribution

is

fractography

results

specimens then fitting a

EIFS

on

statistical

for

resulting
compatible

EIFS
with

the

to

the

Two problems result

(1) the EIFS's are not on a common

fatigue

distribution
the

back

individual

the resulting EIFS

distribution is not statistically compatible with

and

by

distribution

baseline for different data sets, and (2)

distribution

for

results will be

determined

for different data sets.

this approach is used:

baseline

TTCI

wear

out

should

be

distribution.

the

process.

TTCI
The

statistically
The

IFQ

model

presented in this section satisfies this requirement.

Several

useful

applications of the durability analysis

methodology developed are: (1) the evaluation of

design

tradeoffs

in

(2) the evaluation of

before

or

after

durability

terms of structural design variables,

structural

maintenance

requirements

aircraft are committed to service,

7.34

and (3)

..

the evaluation of
cycle-costs,

aircraft

structural

user

options

maintenance

affecting
requirements,

lifeand

operational readiness.

.,.-,
w
:rz,.<
-,
-
,..:;..-,.:
.. ?.<
..>-.;.
,-..: .=
.;.,,./-,-.-:,
....;->>
-.-,: .,-.
:-

S7.35

'4

REFERENCE S
MIL-STD-1530A,
December 1975.

2.

MIL-A-8867B, "Airplane Strength and Rigidity Ground Tests,"

22 August 1975.

"3.

MIL-A-8866B, "Airplane Strength, Rigidity and Reliability

Requirements; Repeated Loads and Fatigue," August 1975.

4.

B. J.
bility

5.

"Fatigue Strength Study Aimed at Improving Test Procedures,"

Aviation Week and Space Technology, April 6, 1979, p. 53.

6.

C. F.

"Aircraft Structural Integrity Program"

1.

-bility
.

Pendley, S. P. Henslee, and S. D. Manning, "DuraStructural DuraMethods Development, Volume III,

Survey: State-of-the-Art Assessment,',' Air Force
Flight Dynamics Lab., AFFDL-TR-79-3118, September 1979.

Tiffany,

"Durability and Damage Tolerance Assessments

of United States Air Force Aircraft,"

Paper presented at

AIAA Structural Durabilfrty and Damage Tolerance Workshop,

Washi.,i.'tcn,

D.C.,

April 6-7,

1978.

7.

"F4 Fatigue and Damage Tolerance Assessment Program," McDonnell Aircraft Company, Report NDC A 2883, Vol. I and II,
28 June 1974.

8.

R. J. Gran et al, "Investigation and Analysis Development

of Early Life Aircraft Structural Failures," Air Force
Flight Dynamics Lab., AFFDL-TR-70-149, March 1971.

9.

A. Palmgren, "Die Lebensdauer Von Kigellogran,"

des Vercines Deutschoe Ingenienro, Vol. 68, No.
1924.

at-.

Zeitschrift
14, April 5,

Fatigue," ASME Journal

3, Sept. 1945, pp. A-159

10.

M. A. Miner, "Cumulativo Damage in

of Applied Mechanics, Vol. 12, No.
- A-164.

11.

H. A. Wood, R. M. Engle,Jr., J.P. Gallagher, and J.M. Potter,

"Current Practice on Estimating Crack Growth Damage Accumulation with Specific Application to Structural Safety, Duraand Reliability," Air Force Flight Dynamics Laboratory,
bility
AFFDL-TR-75-32, January 1976.

R.I1

12.

S. D. Manning, M. A. Flanders, W. R. Carver, and Y. H. Kim,

"Durability Methods Development, Volume II,
Durability
Analysis: State of the Art Assessment," Air Force Flight
Dynamics Lab., AFFDL-TR-79-3118, September 1979.

13.

S. D. Manning, W. R. Garver, Y. H. Kim and J. L. Rudd,

"Durability Analysis - Format Requirements and Critique
of Potential Approaches," Failure Prevention and Reliability_
- 1981,
ASME, 1981, pp. 223-229.

14.

S. D. Manning, J. N. Yang, M. Shinozuka, and W. R. Garver,

"Durability Methods Development, Volume I - Phase I
Summary," Air Force Flight Dynamics Lab., AFFDL-TR-79-3118,
September 1979.

15.

J. N. Yang, S. D. Manning, and W. R. Garver, "Durability

Methods Development, Volume V, Durability Analysis Methodology
Development," Air Force Flight Dynamics Lab., AFFDL-TR-793118, September 1979.

16.

S. D. Manning, J. N. Yang, et al, "Durability Methods Development, Volume VII - Phase II Documentation," Air Force
Flight Dynamics Laboratory, AFFDL-TR-79-3118, Vol. VII,
January
1984.

17.

J. N, Yang, "Statistical
Estimation of Economic Life for
Aircraft Structures," Proc. AIAA/ASME/ASCE/AHS 20th Structures, Structural Dynamics, and Materials Conference,
April 4-6, 1979, St. Louis, Mo., pp. 240-248; Journal of Aircraft,
AIAA, Vol. 17, No. 7, 1980, pp. 528-535.

18.

J. N. Yang, "Statistical
Crack Growth in Durability and
Damage Tolerant Analyses," Proceedings of the AIAA/ASME/
ASCE/AHS 22nd Structures, Structural Dynamics and Materials
"Conference, Part 1, Atlanta, GA, April 6-8, 1981, pp. 38-49.

19.

J. L. Rudd, J. N. Yang, S. D. Manning and W. R. Garver,

"Durability Design
andFracture
AnalysisResistant
for Metallic
Airframes,"
Design Requirements
of Fatigue and

Ile,

Structures,
,o,

20.

ASTM STP 761,

1982,

pp.

133-151.

J. L. Rudd, J. N. Yang, S. D. Manning, and B. G. W. Yee,

"Damage Assessment of Mechanically-Fastened Joints in the
Small Crack Size Range," Proceedings of the Ninth U.S.
National Congress of Applied Mechanics, Symposium on Structural Reliability and Damage Assessment, Cornell U., Ithaca,
NY, June 21-25, 1982.
R.2

14
21.

J. L. Rudd, S. D. Manning, J. N. Yang, and B. G. W. Yee,

"Probabilistic Fracture Mechanics Analysis Methods for
Structural Durability," Proceedings of AGARD Meeting on:
Behavior of Short Cracks in Airframe Components, Toronto,
Canada, Sept. 20-21, 1982.

22.

S. D. Manning and V. D. Smith, "Economic Life Criteria for

Metallic Airframes," Proceedings of 21st AIAA Structures,
Structural Dynamics and Materials Conference, Part 1, 1980,
pp. 504-511.

23.

J.

N. Yang and S.

Equivalent Initial

D. Manning,
Flaw Size,"

"Di stribution

1980 Proceedings Annual

Reliability and Maintainability Symposium,

January 1980, pp. 112-120.

."

of
San Francisco,

24.

S. M. Speaker, D.E. Gorden et al,"Durability Methods Development, Volume VIII - Test and Fractography Data," Air Force
Flight Dynamics Lab., AFFDL-TR-79-3118, November 1982.

25.

P. J. Noronha et al, "Fastener Hole Quality," Vol. I & II,

Air Force Flight Dynamics Lab., AFFDL-TR-78-209, WPAFB, 1978.

26.

W. R. Garver and R. D. Bruner, "Spectrum Fatigue Testing of

Mechanical Fasteners for Structural and Fuel Integrity,"
General Dynamics, Fort Worth Division, ERR-FW-2069, April
1981.

27.

"Fractographic Examination of A-7D ASIP Initial Quality

Specimens," Vought Aerospace Corp., Report ETR-53452078, 19 June 1975.

28.

"Standard Test Method for Constant-Load-Amplitude Fatigue

Growth Rates Above 10-8 m/cycle," ASTM Standard
Crack
E647-81, 1982 Annual Book of ASTM. Standards, Part 10,
Recommended Data Reduction Techniques, pp. 784-788.

29.

D. A. Virkler, B. M. Hillberry and P. K. Goel, "The

Statistical Nature of Fatigue Crack Propagation," AFFDL-

30.

TR-78-43, April 1978.

D. A. Virkler, B. M. Hillberry and P. K. Goel,

"The

Statistical Nature of Fatigue Crack Propagation," Trans.

ASME, J. of Engr. Materials and Technology, April 1979,
Vol, 101, pp. 148-153,

R.3

31.

M. G. Salvadori and M. L. Baron, Numerical Methods in

Engineering, Second Ed., Prentice-Hall, Inc., New Jersey,
1961.

32.

R. K. Penny and D.
Hill Book Co. (UK)

33.

J. B. Conway, Stress-Rupture Parameters: Origin, Calculation and Use, Gordon and Breach, Science Publishers, Inc.,
New York, i969, pp. 258-271.

34.

A Mendelson and S. S. Manson, "The Extrapolation of

Families of Curves by Recurrence Relations with Application
to Creep-Rupture Data," Trans. ASME, J. of Basic Engr.,
Dec. 1960, pp. 839-847.

35.

W. S. Johnson and T. Spamer, "A User's Guide to CGR-GD, A

Computerized Crack Growth Prediction Program," General
Dynamics, Fort Worth Division, Report FZS-241, November 1976.

36.

S. D. Forness, "Fracture Mechanics Methodology Update,"

General Dynamics, Fort Worth Division Report, ERR-FW-2219
(Properietary), December 1981.

37.

R. M. Engle, Jr. and J. A. Wead, "CRACKS - PD, A Computer

Program for Crack Growth Analysis using the Tektronix 4051
Graphics System", Air Force Flight Dynamics Laboratory,
Technical Memorandum AFFDL-TM-79-63-FBE, June 1979.
J. B. Chang, et al, "Improved Method for Predicting Spectrum

28.

C. Marriott, Design for Creep, McGrawLtd., London, 1971, pp. 192-238.

Loading Effects - Phase I - Identification of Controlling

Damage Parameters, Vol. I: Results and Discussion,"
AFFDL-TR-79-3036, Vol. I, Air Force Flight Dynamics
Laboratory, WPAFB, Ohio, 1979.
39.

"F-16 Airframe Preliminary Durability Analysis," General

Dynamics, Fort Worth Division Report, 16PR308, 3 April 1978.

40.

"F-16 Airframe Final Durability Analysis," General Dynamics

Fort Worth Division Reprot, 16PR768, 3 September 1979.

R.4

1.