Skip to main content

Full text of "Results of the Round Robin on opening-load measurement conducted by ASTM Task Group E24.04.04 on Crack Closure Measurement and Analysis"

See other formats


NASA TECHNICAL MEMORANDUM 101601 



RESULTS OF THE ROUND ROBIN ON 
OPENING-LOAD MEASUREMENT 

Conducted By 

ASTM TASK GROUP E24.04.04 ON CRACK 
CLOSURE MEASUREMENT AND ANALYSIS 

fNASA-TH-101601) EESDLTS OF THE ROUND ROBIN H89-24679 
ON OPENING-LOAD SEASOREHENT CONDOCTED BY 

ASTM TASK GROUP E2U.0'*.Oii ON CRACK CLOSURE n.^i.c: 

MEASUREMENT AND ANALYSIS {NASA., ^angley S?i72U1 

Research Center) 39 p CSCL 20K G3/39 0217241 

Edward P. Phillips 



May 1989 



IVI/\SA 

National Aeronautics and 
Space Administration 



Langley Research Center 

Hampton, Virginia 23665 



SUMMARY 



An ex 
in fa 
Group 
purpo 
consi 
Ident 
parti 
Openi 
inten 
overl 
of sp 
all p 
metho 
openi 
the f 
openi 
resul 
corapl 
data . 
lab-t 
nomin 



per Irn 
t igue 

SB of 

stenc 
ify c 
cipat 
ng-lo 
sity 
oad . 
ecime 
art ic 
ds) f 
ng lo 
atigu 
ng-lo 
ts pr 
iance 
Anot 
o-lab 
ally 



enta 

era 
04.0 

the 
y of 
ause 
ed i 
ad m 
fact 
All 
n CO 
ipan 
or a 
ads 
e lo 
ad r 
oduc 

and 
her 

dif 
the 



1 Ro 
ck g 
M on 

Rou 

ope 
3 fo 
n th 
easu 
or 1 
open 
mpli 
ts , 

giv 
was 
adin 
esul 
ed b 

to 
sign 
fere 
same 



und Ro 
rowth 

Crack 
nd Rob 
ni ng 1 
r obse 
e test 
rement 
evels , 
Ing-lo 
ance d 
al 1 me 
en tes 
very 1 
g cycl 
ts was 
y the 
evalua 
if ican 
nces i 

metho 



bin on 
tests h 

Closur 
in was 
oad mea 
rved in 
ing of 
s were 

three 
ad meas 
ata. Wh 
asur eme 
t Gondi 
arge--t 
e. Part 

ascrib 
var i ous 
te the 
t porti 
n produ 
d of me 



the ra 
as be 
e Mea 
to ev 
surem 
consi 
compa 
made 
crack 
ureme 
en al 
nt me 
t ion 
ypica 

of t 
ed to 

meth 
openi 
on of 
clng 
asure 



easu 
en c 
sure 
alua 
ents 
sten 
ct a 
for 

len 
nts 
1 of 
thod 
were 

iiy 

he 1 
con 

ods 

ng 1 
the 

the 

ment 



r emen 
onduc 
ment 
te th 

amon 
cy . E 
nd mi 
crack 
g t h s , 
were 

the 
s , an 

pool 
spann 
arge 
si ste 
used 
oad f 

scat 
compl 



t of 
ted b 
and A 
e cur 
g lab 
leven 
ddle- 
grow 
and 
based 
resul 
d all 
ed , t 
ing t 
scatt 
nt di 
to me 
rom t 
ter w 
lance 



the 
y th 
naly 
rent 
orat 

lab 
crac 
th a 
foil 

on 
ts r 

dat 
he r 
he 1 
er i 
f f er 
asur 
he c 
as a 

dat 



opening 
e ASTM 
sis . Th 

level 
or i as a 
orator 1 
k spec i 
t two s 
owing a 
the ana 
eported 
a analy 
ange of 
ower ha 
n the r 
ences i 
e spec i 
ompl ian 
scr ibed 
a when 



load 
Task 
e 

of 

nd to 
es 

mens . 
tress- 
n 
lysis 

( from 
sis 

If of 

eported 

n 

men 

ce 

to 
us ing 



INTRODUCTION 



Sine 

c los 

numb 

crac 

Inte 

expe 

spec 

char 

meas 

Howe 

comp 

dete 

diff 

ques 

resu 

clos 

data 

r epo 

and 

Admi 

from 

and 



e El 

ur e 

er o 

k op 

rpre 

rime 

imen 

acte 

urem 

ver , 

1 ian 

rmin 

eren 

tion 

Its. 

ure 

mea 
rtin 
anal 
tted 

a r 
test 



ber ' s 
for t 
f res 
ening 
tat io 
ntal 

comp 
r ize 
ent o 

even 
ce me 
e the 
t app 

ar is 

Judg 
behav 
surem 
g lab 
ysis 
ly, t 
eview 

cond 



paper 
he int 
earche 

or cl 
n of t 
technl 
1 iance 
the cl 
f comp 

for t 
asur em 

openi 
roache 
es as 
ing f r 
ior in 
ent pr 
orator 
proced 
he act 

of th 
it ions 



in 
erpr 
rs h 
osin 
heir 
ques 
, an 
osur 
lian 
he c 
ent 
ng 1 
s be 
to w 
om t 

the 
oces 
les 
ur es 
ual 
e li 

are 



1971 
etat 
ave 
g lo 

era 

inc 
d ph 
e be 
ce h 
ompl 
and 
oad 
ing 
heth 
he r 

lit 
s is 
or t 

giv 
leve 
tera 

sel 



( re 
ion 
publ 
ads 
ck g 
ludi 
otog 
havi 
ave 
ianc 
data 
(ref 
used 
er a 
athe 
erat 

not 
hat 
e sy 
1 of 
ture 
dom 



f . 1 ) 
of fa 
ished 
which 
rowth 
ng ul 
raphi 
or . T 
emerg 
e app 

anal 
. 2-k 

to d 
11 of 
r inc 
ure , 

adeq 
the d 
sterna 

inco 

beca 
repor 



on the s 

tigue era 

exper ime 

they hav 

results . 

trasonics 

c methods 

echni ques 

ed as the 

roach, a 

ysis meth 

) • Cons id 

etermine 

the appr 
ons Istent 
it appear 
uately co 
if f erent 
t ically d 
ns i stenc y 
use data 
ted . 



ighif 

ck gr 

ntal 

e use 

A va 

, pot 

have 

base 

most 

numbe 

ods h 

er ing 

openi 

cache 

body 

s tha 

ntrol 

exper 

if f er 

is d 

for t 



icance 
owth d 
measur 
dint 
r 1 ety 
ent ial 

been 
d on t 

popul 
r of d 
ave be 

the n 
ng loa 
s prod 

of da 
t ei th 
led in 
imenta 
ent re 
if f icu 
he sam 



of 
ata , 
emen 
he 
of 

dro 
used 
he 

ar a 
if fe 
en u 
umbe 
ds , 
uce 
ta o 
er t 

the 
1 te 
suit 
It t 
e ma 



crack 

a large 
ts of 



to 

ppr oach . 

rent 

sed to 

r of 

the 

the same 

n 

he raw 

chni ques 
s . 

o judge 
ter ial 



This report documents the results of a Round Robin test activity 
undertaken by the ASTM Task Group E24.04.0i< on Crack Closure 
Measurement and Analysis in an attempt to gain better information on 
the current level of consistency of opening load measurements among 
various laboratories (and by inference the consistency of data being 
reported in the literature) and to identify causes for observed 



Inconsistency. The results will be used to help guide the 
development of a recommended practice for making opening load 
measurements that will lead to the production of a more consistent 
data base in the literature. 

ROUND ROBIN TEST PLAN 



The i 

condi 

made 

addit 

among 

param 

speci 

openi 

into 

metho 

resul 

were 

Appen 

descr 



ntent o 
tions s 
by all 
ion, th 
measur 
eters s 
men typ 
ng load 
the pot 
ds in i 
ts. The 
sent to 
dix A. 
ibed in 



f th 

uffi 

part 

e te 

emen 

uch 

e. T 

s fo 

ent i 

nter 

det 

all 

The 

the 



e test 
cientl 
icipan 
3t pla 
t and 
as str 
he tes 
llowln 
al ut i 
pret in 
ailed 

part i 
sal ien 

Test 



pla 
y so 
ts u 
n wa 
anal 
ess- 
t pi 

g a 

lity 

g va 

Test 

cipa 

t f e 

Plan 



n wa 

tha 
nder 
3 dr 
ysis 
inte 
an a 
spec 

of 
riab 

Pla 
nts 
atur 

are 



3 to sp 
t openi 

nomina 
awn to 

method 
nsity f 
Iso cal 
ified o 
the var 
le-ampl 
n and i 
are Inc 
es of t 

given 



ecify t 
ng load 
lly ide 
provide 
3 as a 
actor 1 
led for 
ver load 
ious me 
i tude-1 
nstruct 
luded i 
he test 
below . 



he test and measurement 

measurements would be 
ntical conditions. In 
data for comparisons 
function of several test 
evel, crack length, and 
the measurement of 
to provide some insight 
asurement and analysis 
oading crack growth 
ions for the tests that 
n this document as 
s and analysis methods 



and 



The Test Plan defined fatigue crack growth tests on the C(T) 
M(T) specimen configurations shown in Figures la and lb 
respectively. All test specimens were fabricated by the same company 
from a single plate of 9.5 mm (3/8 inch) thick 202it-T351 aluminum 
alloy. Specimens were numbered according to their location within 
the plate and then were allotted to participants by a random draw 
process. Four tests were defined in the Test Plan -- two using C(T) 
specimens and two using M(T) specimens. All tests were to be 
conducted in ambient air at a constant stress ratio (R) of 0.1. One 
specimen of each type was to be tested at 



,1 /2 



(6 



, . , 1/2. 
ksi-in ) 



and one at K 



max 



= 22 MPa-m 



a constant 
1/2 



max 



:6.6 MPa- 



(20 



ks i-in ) 



for 



the initial portion of the test. (In the remainder of this report, 
these tests will be referred to as the low-K and high-K tests 
respectively.) Crack opening loads were to be measured at three 



specified crack lengths during the constant K^gj^ 



tests. After the constant K 



max 



portion of the 
portion of the tests, the low-K 
1/2 



ksi- 
1 /2 



, 1/2, 
in ) 



and 



specimen was to be overloaded to lU.8 MPa-m " " (13.5 

the hlgh-K specimen was to be overloaded to 38.5 MPa-m (35 ksi 

in^^^). The overloads correspond to overload ratios (OLR) of 2.25 
and 1.75 for the low-K and high-K tests respectively. After the 
single overload, opening loads were to be measured at specified 
numbers of cycles of the constant-amplitude loading that was being 
used just before the overload. The opening-load measurement times 
specified for the tests, either a crack length before the overload 
or a number of cycles after the overload, are listed In Table 1. 

Although the Test Plan did not specify the experimental technique to 
be used to measure opening loads, it was anticipated that the 
compliance approach would be used in most of the tests. For 



part i 

that 

choic 

comrao 

metho 

resul 

The f 

Linea 

nonvi 

for t 

prese 

calle 

slope 

for o 

diffe 

gener 

all o 

analy 

repor 



c i pan 
the c 
e of 
nly u 
ds an 
ts fo 
our m 
r Dis 
sual , 
he ot 
nted 
d for 
-exce 
nly t 
rence 
ally 
f the 
sis m 
ted 3 



ts u 
orapl 
the 
sed 
d va 
r on 
etho 
plac 

non 
her 
in t 

ope 
edan 
he 1 
s be 
the 

lev 
etho 
impl 



sing 
iance 
parti 
metho 
riati 
ly fo 
ds in 
ement 
subj e 
five 
his r 
nlng- 
ce le 
% lev 
tween 
same 
els w 
ds ( s 
y bee 



the 

dat 
cipa 
ds d 
ons 
ur o 
clud 
, In 
ctiv 
anal 
epor 
load 
vels 
el a 

the 
at a 
ould 
ee T 
ause 



compl 
a be 
nt an 
escr i 
of me 
f the 
e thr 
terse 
e met 
ysis 
t for 
dete 
for 
re pr 
Nonv 
11 si 

Glut 

est P 
too 



lanc 
anal 
d al 
bed 
thod 

met 
ee V 
ctio 
hod 
meth 

the 
rrain 
the 
esen 
isua 
ope- 
ter 
Ian 
few 



e ap 
yzed 
so a 
in t 
s id 
hods 
isua 
n , a 
iden 
ods 

fol 
at io 
Nonv 
ted 
1 rae 
exce 
the 
para 
data 



proa 
ace 
ccor 
he P 
ent i 
are 
1. s 
nd R 
tif i 
cite 
lowi 
ns a 
isua 
beca 
thod 
edan 
f igu 
grap 
wer 



ch , 
ordi 
ding 
Ian . 
f ied 

pre 
ubje 
educ 
ed a 
d in 
ng r 
t th 
1 me 
use : 

and 
ce 1 
r es . 
hs H 
e re 



the 
ng t 

to 

Of 

in 
sent 
ctiv 
ed D 
s No 

the 
easo 
e . 
thod 

(1 ) 

the 
evel 

Dat 
. 1 a 
cei V 



Test Pla 
o the me 
several 
the nine 
the Test 
ed in th 
e method 
i splacem 
nvisual ( 
Test PI 
ns. The 
5%. ^%, 
, but th 
the tre 
other m 
s and (2 
a for tw 
nd JJ.2.3 
ed to be 



n re 
thod 
othe 
ana 
Pla 
is r 
S--U 
ent- 
H) . 
an a 
Test 
2%, 
e re 
nds 
etho 
) in 
o of 
) ar 
use 



quested 

of 
r 

lysis 
n , 

eport . 
pper 
-and one 

Results 
re not 

Plan 
and k% 
suits 
in 

ds were 
eluding 

the 
e not 
f ul . 



The fou 
compare 
Figures 

(1) Upp 
determi 
becomes 

(2) Int 
at whic 
upper p 
minimum 

(3) Red 
determi 
equal t 
"reduce 
d i splac 
1 inear 
in) Non 
var iat i 
slope f 
load be 
average 
descr i p 
Appendi 



r CO 
d in 

2a- 
er L 
ne t 

lln 
erse 
h a 
art 

loa 
uced 
ne t 
th 
d" d 
emen 
load 
visu 
on o 
or t 
low 

slo 
t ion 
X A) 



mpl 
th 

2d: 

ine 

he 

ear 

ct i 

lih 

of 

d a 
Di 

he 

e s 

i sp 

t a 

-di 

al 

f s 

he 

whi 

pe 
of 



iance-based opening-load analysis methods that are 
is report are described below and illustrated in 

ar Displacement- From a load-displacement plot, 

load at which the upper portion of the loading curve 



on- From a load-displacement plot, determine the load 
e drawn through the maximum load and tangent to the 
the curve intersects with a line drawn through the 
nd tangent to the lower part of the curve, 
splacement- From a load-"reduced" displacement plot, 
load at which the slope of the loading curve becomes 
lope of the upper portion of the unloading curve. A 
lacement is the difference between the measured 
t a given load and the displacement defined by a 
splacement relation at the same load. 
{^%)- From load-displacement data, evaluate the 
lope with load and compare the slopes to the average 
upper 25? of the load cycle. Determine the maximum 
ch the slope is always at least ^% greater than the 
of the upper portion of the curve. (A more detailed 
the procedure is given in the Test Plan in 



It was recognized that the scatter in results that would be reported 
from the different labs using the compliance approach would have two 
eomponents--scatter due to differences in collecting the compliance 
data and scatter due to differences in analyzing the compliance data 
to evaluate the opening load. It was also recognized that it would 
be extremely difficult to quantify the two components of scatter 
based solely on the Round Robin data. To obtain some indication of 
the scatter in the Round Robin data set due to the analysis methods 
alone, all participants were provided with identical sets of 
load-displacement data and asked to determine the opening load using 
the same analysis methods specified for the experimental Round 



Robin. Analysis of the opening load results from the various labs 
for the same raw data should indicate the differences in mean values 
and scatter to be expected from the different analysis methods. 

ROUND ROBIN DATA SET 

One complete set and ten partial sets of test results were received 
from the participants listed in Table 2. The partial data sets 
either did not include data for all test conditions or for all 
analysis methods. All of the participants used measurement of 
specimen compliance to determine opening loads. Results were 
reported for measurements made using displacement gages at the 
mouth on C(T) specimens and the specimen centerline on M(T) 
specimens; strain gages on the back face of C(T) specimens, and an 
interferometric displacement gage near the crack tip in both C(T) 
and M(T) specimens. 



crack 




et . 



Some of the data that were received were not included in 
data that was analyzed. Data were eliminated if they were not for 
the test and measurement conditions specified in the Test Plan or if 
the reported crack growth data indicated that the accuracy of the 
test loading was suspect. About five percent of the data submitted 
were discarded for the latter reason. 

ANALYSIS OF THE ROUND ROBIN EXPERIMENTAL RESULTS 



In this 
are pres 
making o 
indicate 
measurem 
As expec 
rigorous 
been mad 
var ious 
pooled d 
sets (va 
necessar 
and may 
Signif ic 
evident 



section 
ented : 
pening- 

trends 
ent typ 
ted , th 

Stat is 
e to sh 
ways an 
ata set 
r iables 
ily hel 
indicat 
ance sh 
across 



, the 
(1 ) to 
load m 

in th 
e , dat 
e Roun 
t ical 
ow tre 
d plot 
s . The 

that 
d cons 
e tren 
ould o 
the br 



test resu 
indicate 
easur emen 
e opening 
a analysi 
d Robin d 
analysis 
nds in th 
ting the 

reader i 
may influ 
tant in e 
ds that a 
nly be at 
eadth of 



Its from the Round Robin participants 

the current level of consistency in 
ts among several labs, and (2) to 
-load results as a function of 
s method, and several test parameters, 
ata set is too fragmented to support a 
of the data. Therefore, an attempt has 
e opening-load data by pooling data in 
means and standard deviations of the 
s cautioned that inhomogeneous data 
ence the comparison -are not 
ach set) are generally being compared 
re misleading when interpreted alone, 
tached to those gross trends that are 
the Round Robin data population. 



Most of the results are plotted on figures which show the opening- 
load ratio (ratio of opening load to maximum fatigue load) against 
test •measurement time. Where mean values or standard deviations of 



opening-load ratios are used, these terms are defined in the usual 
way as 



mean value 



^C I x^]/n 



-.2 



standard deviation = s - [(I(x -x) )/(n-1)] 
! X are Individ 
points in the sample 



0.5 



where x are individual data points and n is the number of data 



Overall Consistency 

All of the Round Robin opening-load data for the f 
tests are shown in Figures 3a-3d. The lines in the 
data points representing the same combination of p 
measurement type, and analysis method. With few ex 
lines in the figures do not fluctuate up and down 
suggests that the large overall scatter shown in t 
mainly due to systematic differences among labs, m 
and analysis methods. It is felt that the large sc 
shown in these figures represents the potential sc 
be reported in the literature because data from al 
measurement types and analysis methods represented 
are also represented in the literature. Scatter of 
would make it very difficult to develop a clear pi 
effects and to verify quantitative models of closu 
data from the literature. 

Effect of Compliance Measurement Type 



our crack 

figures 
ar t i ci pan 
cept ions , 
very much 
he figure 
easur emen 
atter in 
atter tha 
1 of the 
in the f 
this mag 
ctur e of 
re effect 



growth 
connect 
t, 

the 
. This 
3 is 
t types 
results 
t might 

i gures 
ni tude 
closure 
s using 



The 

Robi 

(1 )C 

spec 

on t 

di sp 

behi 

data 

dete 

valu 

C(T) 

aval 

cons 

meas 

cons 

both 

of t 

not 

BFS, 



foil 
n : 

rack 
imen 
he b 
lace 
nd t 

fro 
ct a 
es o 

tes 
labl 
i ste 
ur era 
iste 

typ 
he m 
indl 

and 



owing types of compliance measurement were used in the Round 



ope 
s an 
ack 
ment 
he c 
m al 
ny t 
f th 
ts a 
e. F 
nt d 
ents 
ntly 
es o 
easu 
cate 

IDG 



ning d 

d at t 

face ( 

gages 

rack t 

1 part 

rends 

e pool 

nd the 

rom th 

if fere 

made 

highe 

f spec 

rement 

any c 

metho 



ispl 
he c 
BFS) 

(ID 
i p o 
icip 
in r 
ed d 

one 
ese 
noes 
on C 
r va 
imen 

typ 
onsi 
ds . 



acemen 
enterl 

of C( 
G) loc 
n both 
ants a 
esults 
ata ar 

M(T) 
result 

in th 
(T) sp 
lues o 
. The 
es is 
stent 



t (C 
ine 
T) s 
ated 

typ 
nd a 

due 
e sh 
test 
s, i 
e re 
ecim 
f op 
scat 
show 
dif f 



OD) 
of M 
peel 

0.0 
es 
11 a 

to 
own 

for 
t ap 
suit 
ens , 
enin 
ter 
n in 
er en 



gages 
(T) s 
mens , 
5-0.0 
f spe 
nalys 
raeasu 
in Fi 

whic 
pears 
s fro 

but 
g loa 
in op 

Figu 
ces i 



at th 
pecime 

and ( 
8 mm ( 
c imen . 
is met 
rement 
gures 
h comp 

that 
m the 
the ID 
d than 
ening- 
res 5a 
n scat 



e cr 
ns , 
3)in 
0.00 

The 
hods 

typ 
Ha~H 
arat 
ther 
COD 
G do 

the 
load 
-5c. 
ter 



ack 

(2)3 

terf 

2-0. 
ope 
wer 

e. T 

c f o 

i ve 

e ar 

and 

es a 
COD 
val 
The 

amon 



mout 
trai 
erom 
003 
ning 
e po 
he m 
r th 
data 
e no 
BFS 
ppea 
app 
ues 
se r 
g th 



h of C(T) 
n gages 
etr i c 
in. ) 
-load 
oled to 
ean 
e two 
were 



r to give 
roach on 
from each 
esults do 
e COD , 



Effect of Data Analysis Method 

The opening-load data from all participants and all measurement 
types were pooled to detect any trends in results due to method of 
data analysis. The mean values of the pooled data are shown in 



F igur e 
differ 
analys 
lowest 
Gonsls 
Displa 
values 
analys 
result 
scatte 
No con 
discer 



3 6a- 
ences 
is me 

open 
tentl 
ceraen 

over 
is me 
s , th 
r tha 
siste 
nible 



6d. T 
in t 
thods 
ing-1 
y pro 
t and 
all. 
thods 
e Red 
n the 
nt di 
in t 



hese re 
he open 
. The I 
oad val 
duced t 

Nonvis 
The sea 

is sho 
uced Di 

other 
f f erenc 
he plot 



suit 
ing- 
nter 
ues , 
he s 
ual( 
tter 
wn i 
spla 
thre 
es i 
s fo 



s in 
load 
sect 

the 
econ 
U) 

in 
n Fi 
ceme 
e me 
n sc 
r th 



dicate that there are systematic 

values determined by the four 
ion method consistently produced the 
Upper Linear Displacement method 
d-lowest values, while the Reduced 
methods produced about the same 
opening-load values from each of the 
gures 7a-7d. In the C(T) test 
nt method appears to show more 
thods, which are all about the same. 
atter due to analysis method are 
e M(T) specimen. 



Effect of Overload 



All of the data in the 
determine whether diff 
showed the same capabi 
subsequent opening-loa 
tests, none of the mea 
systemat ic""change in t 
though the crack growt 
in some cases. These r 
detection methods were 
changes in closure beh 
responsible for the dr 
high-K tests, all of t 
available showed a tre 
following the overload 
pronounced in some cas 
a trend towards higher 
participants and this 
results following the 
showed no change in re 



previously presen 
erent measurement 
lity to detect the 
d behavior. In bot 
surement types or 
he opening loads a 
h showed a large d 
esults suggest tha 

not sufficiently 
avior or that clos 
op in growth rates 
he measurement typ 
nd towards higher 
, although the cha 
es. Three of the f 

values, but the t 
is reflected in th 
overload. The Inte 
suits following th 



ted figure 
types and 

effects 
h the C(T) 
analysis m 
fter the o 
rop in rat 
t ei ther t 
sensi t i ve 
ure mechan 
. In both 
es for whi 
opening lo 
nges in va 
our analys 
rend was n 
e higher s 
r sect ion m 
e overload 



s were examined to 
analysis methods 
f an overload on 

and M(T) low-K 
ethods showed a 
verload , even 
e or even arrest 
he opening-load 
to detect the 
isms were not 
the C(T) and M(T) 
ch data were 
ad values 

lues were not very 
is methods showed 
ot found by all 
catter in the 
ethod consistently 



Effects of Other Variables 

The Round Robin results were also examined to detect whether there 
were effects of specimen type, K level, and crack length on the 
measured opening loads. No consistently large effects were evident 
for any of these variables but there were hints in the data that 
small effects might exist. Most notable were the tendencies in the 
C(T) specimen towards lower opening-load values for higher K levels 
and towards lower scatter in the results at the longer crack lengths 
(before overload). 

ANALYSIS OF IDENTICAL COMPLIANCE DATA BY SEVERAL PARTICIPANTS 

Identical load-displacement data were sent to several labs for 
evaluation of opening loads by the four analysis methods described 
earlier in this report. These results should provide a good 
indication of systematic differences in the mean values and the 
scatter of opening-load values produced by the different analysis 
methods. Load-displacement data for seven complete load cycles were 
selected for analysis. Data for six of the load cycles were taken 



i:i 1 



'during the Round Robin and represent a variety of the test 
conditions specified for opening-load measurements on C(T) 
specimens. The load-displacement data for one of the load cycles was 
generated analytically using displacements calculated from the crack 
closure model in ref. 5. The analytical load-displacement data 
contain no experimental artifacts that might influence the analyses. 
Analysts of the data were not told that one of the data sets was - 
generated analytically. Lists of the test conditions for the 
individual load cycles (data sets) and the participating analysts 
are given in Tables 3 and 4 respectively. 



The 

meth 

are 

the 

lowe 

seco 

Disp 

tren 

set 

disp 

open 

give 

anal 

Figu 

subs 

from 

Inte 



mean 

ods 

syst 

diff 

st V 

nd-1 

lace 

ds i 

show 

lace 

ing 

som 
ys is 
re 8 
tant 

the 
r sec 



val 
are 
emat 
eren 
alue 
owes 
ment 
n th 
n in 
ment 
load 
e in 

met 
a , a 
iall 

Red 
t ion 



ues 

show 
ic d 
t an 
3 , t 
t va 

and 
e re 

Fig 

dat 
. Th 
dlca 
hods 
11 f 
y un 
uced 

met 



of th 
n in 
iffer 
alysi 
he Up 
lues , 

Nonv 
suits 
ures 
a (da 
er ef o 
t ion 

to d 
our o 
der es 

Disp 
hod b 



e ope 
Figur 
ences 
s met 
per L 

and 
i sual 

are 
6a-6d 
ta se 
re , a 
of th 
etect 
f the 
t imat 
lacem 
eing 



ning 
e 8a 

in 
hods 
inea 
the 
(1$) 
the 
. Th 
t 7) 
naly 
e re 
the 
Rou 
ed t 
ent 
fart 



loa 
. Th 
the 
. Th 
r Di 
othe 
) pr 
same 
e an 

had 
sis 
lati 

act 
nd R 
he o 
meth 
hest 



ds e 

ese 

open 

e In 

spla 

r tw 

oduc 

as 
alyt 

a k 
resu 
ve c 
ual 
obin 
peni 
od b 

fro 



valua 
resul 
ing-1 
terse 
cemen 
o met 
ed ab 
noted 
icall 
nown , 
Its f 
apabi 
openi 
anal 
ng lo 
eing 
m the 



ted 
ts i 
oad 
ctio 
t me 
hods 
out 
for 

y-ge 

num 
rom 
lity 
ng 1 
ysis 
ad - 
clos 

ana 



by t 
ndic 
valu 
n me 
thod 

(Re 
the 

the 
nera 
eric 
data 

of 
oads 

met 
- wi 
est 
lyti 



he four 
ate tha 
es prod 
thod pr 

produc 
duced 
same va 

Round 
ted loa 
ally-de 

set 7 
the var 
. As in 
hods 
th the 
and the 
cal val 



analys is 
t there 
uced by 
oduced the 
ed the 



lues . 

Robin 

d- 

f ined 

should 

ious 

di cated 

values 

ue . 



The 
data 



1 n 



The standard deviations of the opening loads evaluated by 
analysis methods are shown in Figure 8b. These results ind 
the Reduced Displacement and the Upper Linear Displacement 
produced substantially more scatter in results than the In 
and NonvisuaK 1 $) methods. The low scatter in the Intersec 
results was a bit surprising because it had been expected 
three of the subjective analysis methods would show more s 
than the nonsub j ect i ve Nonvisual method. The analysis of t 
in the Round Robin data set had also shown large scatter f 
Reduced Displacement method, but had shown essentially no 
differences among the other three methods. It is not clear 
results from the Round Robin data set did not show higher 
for the Upper Linear Displacement method than the Intersec 
Nonvisual methods, but the inhomogenei ty of the Round Robi 
could have caused the difference to be obscured. 



the four 
icate that 

methods 
ter sect i on 
t ion 

that all 
cat ter 
he scatter 
or the 

why the 
scatter 
tion and 
n data set 



DISCUSSION 



As mentioned earlier, an objective of the Task Group is to develop 
recommended procedures for measuring opening loads and to encourage 
their use. The motivation for developing the procedures is to 
improve the consistency of the opening-load data being published in 
the literature. The current Round Robin serves as the first step 
towards meeting the Task Group objectives by providing results that 
document the current level of consistency, indicate some directions 



for improving the consistency, and identify areas where further work 
is required. 

Considering the differences in results (both mean values and 
scatter) obtained from the different analyses, an obvious way to 
improve the consistency beyond that shown In Figures 3a-3d would be 
to use only one analysis method for all measurements. Also, since 
the data suggest that the values obtained by the IDG measurements of 
compliance near the crack tip are different from those obtained by 
COD and BFS measurements, the consistency could be further improved 
by segregating the data population into groups representing 
measurements remote from the crack tip and near the crack tip. The 
potential Improvement in consistency to be gained by using only one 
analysis method and only remote compliance measurements is indicated 
in Figures 9a-9b using the data from the Round Robin tests on C(T) 
specimens. The analysis method chosen for illustration in Figures 
9a-9b is the Nonvisual method because common sense would indicate 
that a nonsubjective analysis should produce the most reproducible 
results (experimental accuracy being equal) and, indeed the results 
discussed earlier tend to confirm that. As expected, the standard 
deviations of opening loads for the restricted data set are 
consistently less than those for the all-inclusive data set. To give 
an idea of how much further Improvement in consistency might be made 
if experimental accuracy was the same for all labs, the standard 
deviations from analysis of the identical data sets (Figure 8b) 
using the Nonvisual(U) method are also plotted in Figures 9a-9b. 
From these results, it appears that recommended procedures for 
obtaining uniformity in compliance data among labs is necessary to 
achieve high levels of consistency. 



The accuracy of the opening-load measur 
Robin cannot be determined at present b 
establishing the "true" opening load do 
most cases the crack does not open alon 
single load, but rather it opens increm 
Therefore, a question concerning the ac 
opening-load measurements made In the R 
remain. Nevertheless, when considered a 
representing the effects of closure alo 
the single-load opening character i zat io 
correlating crack growth data and inter 
effects in crack growth tests. That bei 
worthwhile to identify and promote the 
nonsubjective measurement procedure to 
consistent opening-load values in diffe 

CONCLUSIONS 



ements reported in the Round 
ecause an accepted method of 
es not exist. Actually, for 
g the entire crack front at a 
entally over a range of load. 
curacy of the kind of 
ound Robin will always 
s an "effective" opening load 
ng the entire crack front, 
n may be useful in 
preting phenomenological 
ng the case, it would seem 
use of a well-defined, 
assure the production of 
rent laboratories. 



The following conclusions are based on the results from the Round 
Robin on crack opening-load measurement conducted by ASTM Task Group 
E2U.04.0M: 



1 . When all of the mea 
all measurement types 



all measurement types, an 
condition (specimen type. 



asured opening loads (from all participants, 
, and all analysis methods) for a given test 
ype, K level, and crack length) were pooled. 



ITT' 



the range of opening loads was very large--typ Ically spanning the 
lower half of the range of the fatigue loading. 

2. The opening loads measured using certain compliance measurement 
methods and data analysis methods were systematically different from 
those measured using the other methods. These systematic differences 
were largely responsible for the large scatter in the pooled results 
described in conclusion 1. 

3. Of the four analysis methods used to determine the opening load 
from the load-displacement data, the Intersection method 
systematically produced the lowest values for opening load, the 
Upper Linear Displacement method systematically produced the 
second-lowest values, and the other two methods (Reduced 
Displacement and Nonvi sual ( 1 $ ) ) produced about the same values. This 
trend was noted in the analysis of the load-displacement data from 
the various participants in the Round Robin and in the analysis of 
identical load-displacement data by several participants. 

4. The opening loads derived from the Inter ferometr ic Displacement 
Gage compliance measurements made near the crack tip appeared to be 
higher than the opening loads derived from crack-mouth-opening and 
back-face-strain compliance measurements made remote from the crack 
tip. 

5. Results of the analysis of identical load-displacement data by 
several participants indicated that use of the Intersection and 
Nonvl sual ( 1 $ ) analysis methods resulted in very little scatter in 
reported opening loads, whereas the Upper Linear Displacement and 
Reduced Displacement methods produced considerably greater scatter 
in results . 

6. When only those Round Robin opening load results based on 
compliance data taken remote from the crack tip and analyzed by the 
Nonvisual ( 1 !t ) method were pooled, the scatter in the results was 
substantially less than the scatter for the overall data set 
described in conclusion 1. This result indicates the potential 
Improvement in consistency of published opening load data that could 
be achieved by widespread use of a single data analysis method by 
the research community. 



7. Substantial lab-to-lab differences were also noted in the Round 
Robin results. To achieve a high level of consistency among labs, 
procedures to assure the generation of acceptably uniform load- 
displacement data in the various labs must be developed and 
implemented . 

8. The capability to detect changes in opening load behavior 
following an overload was not consistent among labs--even among 
those using the same compliance measurement and analysis methods. 
Among the analysis methods, the Intersection method was the only one 
that did not show a general trend towards higher opening loads 
following the overload in the high K test of the C(T) specimen. 



REFERENCES 



El 
To 
fo 
Ba 
Fo 
Al 
Fa 
Me 
Ma 
Do 
Le 
Am 
Ne 
Cr 
fo 
ST 
53 



ber , 

lera 

r Te 

nerj 

rce 

1 iso 

tigu 

Chan 

teri 

nald 

vela 

er ic 

wman 

ack 

r Pr 

P 71* 

-814. 



W. 
nee 
st in 
ee , 
Mate 
n , J 
e Cr 
ics , 
als , 

, J. 
. Me 
an S 
J. 
Grow 
edic 
8, A 



: The Sig 

in Aircra 

g and Mat 

S- : A Re 

rials Lab 

ohn E . : 

ack Growt 

ASTM STP 

1988, pp 

Keith : 

chanlcs o 

ociety fo 

C. , Jr . 
th under 
ting Fati 
inerican S 



nif ican 
ft Stru 
erials , 
view of 
oratory 
The Mea 
h. Elgh 
9^5, A 
. 913-9 
A Proce 
f Fatig 
r Testi 
: A era 
Aircraf 
gue Cra 
ociety 



ce of Fa 
ctures , 
1971 , p 
Crack C 
, Dayton 
surement 
teenth N 
mer ican 

33. 

dure for 
ue Crack 
ng and M 
ck-Closu 
t Spectr 
ck Growt 
for Test 



tigue Crack Closure. Damage 
ASTM STP 486, American Society 
p. 230-242. 

losure. AFWAL-TR-84-4031 , Air 
, Ohio, April 1984. 
of Crack Closure During 
ational Symposium on Fracture 
Society for Testing and 

Standardizing Crack Closure 

Closure, ASTM STP 982, 
aterials, 1988, pp. 222-229. 
re Model for Predicting Fatigue 
um Loading. Methods and Models 
h under Random Loading, ASTM 
ing and Materials, 1981, pp. 



10 



[1 I 



Table 1.- Opening-load measurement times specified for the Round 
Robl n tests 



Figure 
Caption 

CL1 
CL2 
CL3 



Crack Lengths 

C(T) 
a , mm( in . ) 

25.4( 1 .00) 
27.9(1 .10) 
38.1 (1 .50) 



M(T) 
a , mm ( in . ) 

12.7(0.50) 
15.2(0.60) 
25.4(1 .00) 



Figure 


Caption 


NO 


N1 


N2 


N3 


N4 


N5 



Cycles After Overload 
Low-K Test High-K Test 



2.5x1 
5.0x1o' 
1 .0x1 o' 
2.0x1 o' 
4.0x1 o' 




2.5x1 0' 
5.0x1 0' 
1 .0x1 0- 
2.0x1 0- 
4 .0x1 0- 



Table 2.- Participants in Round Robin 
Participants Affiliations 



Noel Ashbaugh 
Anders Blom 
Keith Donald 
Alten Grandt 
Linda Link 
George Miller 
Matt Miller 
Ed Phillips 
Bill Sharpe 
Ralph Stephens 
Dale Wilson 



University of Dayton Research Institute 

FFA (Sweden) 

Fracture Technology Associates 

Purdue University 

David Taylor Research Center 

PSG, Inc. 

Boeing Commercial Airplane Company 

NASA Langley Research Center 

Johns Hopkins University 

University of Iowa 

Tennessee Technological University 



11 



Table 3.- Round Robin test conditions represented in 

load-displacement data analyzed by several participants 



Data Set 



Test Condition 



1 
2 
3 

5 
6 

7 



Low K , 
Low K , 
High K 
High K 
High K 
Low K , 



CL3, 

CL3, 

, N4, 

, CL1 

, NO, 

NH, 



BFS 
COD 
BFS 

, COD 
BFS 

BFS 



(Analytically generated) 



Table U.- Analysts of identical load-displacement data 



Participant 

Noel Ashbaugh 
Linda Link 
Matt Miller 
Ed Phillips 
Bill Sharpe 



Affiliation 

University of Dayton Research Institute 
David Taylor Research Center 
Boeing Commercial Airplane Company 
NASA Langley Research Center 
Johns Hopkins University 



12 



n 1 



-(i>-— IJ^— -<i> 



J 






i 



\< o3 r^ >|< — oJr2 ^ 



S$2 



c 
o> c u 
a 2=5 



■>► 



vi 

0) 

c 
u 



o 
(0 




o 

o 

a. 


1 


vi 


o 


P 


2 




a> 


S 


t5 




E 


n 


ss 



4I> 

E 
o 

(A 

P 

o 



E 

c 

w 

c 

£ 
w 

c 

0) 

E 
'•5 



o 






3 
(0 

c 
o 

3 

c 
o 
o 

c 
E 



(/> 



3 



13 




f 

I 

s 



e 




iS 

(0 

■o 

c 
o 

E 
u 



i 


8- 


S 


S 


§ 


CO 


£. 


2 


« 


E 


s 





■5 




g 


^ 


z 


8 






f 






I' 



c 
c 

o 
o 

3 

n 




u 



I 




« 
1 

3* 



c 

0) 



5 
« 

■8 

o 

3 



n 
o 
OC 

■o 
c 

3 

o 



■o 

SI 

3 
CO 

E 

I 

x: 
o 

CO 



to 

3 



2 

3 

u. 



14 



ri f" 




1 

a 

e 

I 



« 
O 

3 




I 

a 

c 

o 



I 
I 





8| 

£ w 

<u u 

«= c 
c w 

■Q oT 

O Q. 

OC 0) 

11 

CQ C 
« ra 
ki O. 

S5 

S m 
> a 

™ ~ 

■O.S 
■0-2 
Oo 
°S 

^1 

OE 
52 



IS 

o 



3 
C3» 



■ 

O 



I 

s 

s 



CO 

2 

3 
O) 

U. 



15 




If 



■o 

I 

s 








-. 1 




^ 




*^'\ 








i 








\ 






■oS 


.' i^ 




— 


feS~ 


:i 






. Aft 
overl 


i 




— 


' 


i >-' 


••''? 


) 


~" 






.•■/^ 






i 


' 


/ « 




~ 


§1 


■ 1 




- 


"s k 






I 1 


I 1 


L... 


" 



-|S 







i\ 






\ \ 


- 






— 


*i 








• 




S^ 


\ 




, , «i 


y 


~" 


' ' 


<' 


- 


^ 


\ I 




• s 


', 1 




« 5 


\ 




m > 


i 1 




1 1 


1 1 


1 



s - 
2 I 



-2 



- £! = 



if 



-3 

-3 
-5 



Is 



6.S 
c 



</> 

IS, 
3 
(A 

£ 

fl 
II 

o« 
o >, 



^1 

a. 

(0 



-2 



a 
d 



e 



g 



a 
« 

e 

s 



M 



Q> 



re 

c 
re 

■o 

c 
re 



C 

re 

a. 



«E 



2 E^ 

|3 



2-1 
2 |g 



1 



o ' 
O o 

S O 



CD 

_o 

O 

s 






3 
(0 

re 
a> 

E 
So 

.2 «i 

11 

O E 

O o 

•^ Q. 



3 



«? 



e 



1 

e 



■^ 9 



■ 



a 

c 

e 

%. 
O 



5 

5 



16 



[1 ■! 




II- 

Si 

(0-5 



M 

CI 

s 



in 

■OCM 

<gtt 



o 

1 
I 



Type of 

compliance 

measurement 


COD 

BFS 

IDG 



-2 

-S 



II 

Si 



4) 

E 



Ie 



S 

E 

1 
w 

e 

s 



ciS 
T «" 

OS 



-la E 
o Is 



e 



e 



•a 
eg 

1§ - 
?5o £ 

(0-S fi. 
O 



o ■fiS ^ 



a 

o 

o 

3 

a 

K 



31 



e 



P 

I 

I 



O U 


i'r 


,U> flj 


= "■ 


(0 = 


ra <o 


* - 


4> w 




C (0 


ats 


a^ 


11 


o o 


«- o 


o a 




o 


Q.M 




rr 3 


o2 


tr ^ 


^13 




iri 


o 




3 


O) 




u. 



II- i 

o 



i 

3 

3 



17 




If 






\ \ 1 


■" 




\ I 1 






*•. ' I 










V. ' 


_ 




A ^ 






V 

■' 1 


— 


«"5 " 
< Sec 


^ t'. 


- 


g-i 


N \^ 




1 


, H 


A ! f 


- 










■ 




' ••l 1 


- 


•"S 


•/ 




o o 


\\l 


^ 


■ss 




1 1 


1 1 1 





2l 



. 1 




s 






•D 


75. 1 


1 


■S J 


^ 




• 




s. 


= 8|3 


• 






1 : 1 




1 : • 




• : 1 



I J 

- e 
O 



I 

K 

a 
S 



2 
S 

z 

HS 

o • 

-3 
-3 



i| 

9 — 

fl 

Is 



1 



e 



le 

e 



e 






2 
& 

p 

3 
o 

S 

mm 

X 







(0 

T5 

3g 
<A 1 

S>S 

^^ 
o* 

CO. 

" E 
o 
u 

o o 
■n V) 
OS 

^2: 
E? 
«« 

it 

«. (0 



SI 



o 

3 



■ 

Si 

3 

a 



2 - 

S o 

1 g 

S 1 

S O 



s 

3 
01 



18 



Ml 1 





■o 

I 

3 
W 

M 
1 



in 
If" 



s'S 



1 

4 






Before -•— 
overload 


.'Ml 



«1 

e 




d 




. 1 




s 


■g 


i s 


s 


• s 


o 


a S 


E 
• 


Icl^ 


>. 


S 0*0 — 


< 


eriin 
rsecti 
uced 
visua 




Upp 
Inte 
Red 
Non 




1 : 1 




111 




3 
O) 




» 

P 

3 
"o 

I 

1 



= E 
£2 

5 CO 

2 w 

O 0) 

oE 

C Q) 

= C 

Q..S 

OoL 

is 

Z o 
ra S. 

^E 

o a. 
5o 

t (0 
OT CL 

W^ 

gE 
a2 

«m 
■ 

2 

3 



V c 8 

*- O ^ 

■"el's e 

53 e 

(O-O 






■ 

s 

3 

a 






(O-D 



19 



Mean value of 



1.0,- 



0.8 



0.6 



( Opening load \ 
Max fatigue load / 0,4 



0.2 



Analysis method 



Participants 



Upper linear displacement 



intersection 

Reduced displacement 

Nonvisuai 1% 



4 
4 
3 
5 



Analytical 
value 




I 



I 



3 4 

Data set number 



a. Effect on mean values 



Standard 

deviation 

of 



(Opening load \ 
Max fatigue load / 




3 4 

Data set number 



b. Effect on scatter 



Figure 8.- Effect of data analysis method on the mean values and scatter of opening 
loads determined from evaluation of identical load-displacement data sets 
by several participants. 



20 



ii 1 



0.4 r- 



0.3 



Standard 
deviation 

of 0.2 

(Opening load \ 
Max fatigue load / 



0.1 - 



Before ■<- 
overload 



-»■ After 

overload 
(OLR s 2.25) 



Round Robin data - all participant*, measurement 
typet, and analysis methods 

Round Robin data • all participants, only COD and BFS 
measurements, only nonvlsual (1%) analysis method 

Identical toad-displacement data analyzed by 
several participants using nonvlsual (1%) method 




CL1 CL2 CL3 



Specified opening-load measurement limes, 
cracic length or cycle number 



a. Low-K lest of C(T) specimen 



0.4 r- 



0.3 - 



Standard 
deviation 

of 0.2 

Opening load \ 



L Max fatigue load 



0.1 - 



Before .*- 
overload 



J L 



-». After 

overload 
(OLR = 1.75) 



J L 



_L 



CL1 CL2 CL3 NO Nl N2 M3 N4 

Specified opening-load measurement times, 
crack length or cycle number 

b. HIgh-K test of C<T) specimen 



f45 



Figure 9.- Scatter In measured opening loads for all Round Robin data, for a subset of Round Robin 
data, and for identical load-displacement data sets analyzed by several participants using 

tlie Nonvlsual (1%) method 



21 



APPENDIX A 

Test Plan and Instructions for 
E2i|.0i|.0i» Round Robin on Crack Opening Load Measurement 



General Description 



This experimental Round Robin program is being conducted to 
determine the current level of consistency of crack opening load 
measurement among laboratories and to obtain data to guide 
development of a recommended practice for opening load measurements. 
The test program consists of tests on two C(T) and two M(T) 
specimens that are being provided to the participants. Stress 
intensity histories have been specified for each specimen such that 
one specimen will exhibit a nearly-constant, low growth rate and the 
other specimen will exhibit a nearly-constant, high growth rate for 
most of the test. Crack opening loads are to be measured at three 
specified crack lengths during the constant-rate portion of the 
test. After the constant-rate portion of the test, a specified 
overload is to be applied and then opening loads are to be measured 
at specified cyclic intervals following the overload. Participants 
using load-displacement data to determine the opening load are to 
determine the opening load using several specified data analysis 
methods . 



Detailed Description 



1 . Material and Specimens 

1.1 Material - Single plate of 3/8 inch thick 202i|-T351 aluminum 
alloy . 

1.2 Specimens - C(T) and M(T) (See Figures A1 and A2). 

Note 1 : The initial crack starter configuration may be 

modified to accommodate a participant's particular 
measurement apparatus as long as the length of the 
notch is not Increased and the configuration remains 
consistent with the guidelines of ASTM Standard E647. 

Note 2: The M(T) specimen can be tested by using either 
clamping-type grips or a pin/clevis arrangement. 

2. Displacement Measurement Methods 

All methods with which the participant has experience (COD gages, 
CMOD gages, back-face strain gages, laser interf erometry , strain 
gages across crack, scribe marks, etc.) 



22 



n r 



3. Displacement Measurement Locations 

3.1 C(T) specimen - Crack mouth, back face and as many other 
locations as possible. 

3.2 M(T) specimen - Specimen centerline and as many other 
locations as possible. 

M. Procedures for Evaluation of Opening Load from Displacement Data 

4.1 Customary method 

Evaluate opening loads using the method you would use if you 
published a report today. 

4.2 Visual (subjective) methods 

4.2.1 From a load-displacement plot, determine the load at 
which the upper portion of the loading curve becomes 
linear. (See Figure A3) 

4.2.2 From a load-displacement plot, determine the load at 
which a line drawn through maximum load tangent to the 
upper part of the curve Intersects with a line drawn 
through the minimum load tangent to the lower part of 
the curve. (See Figure A4) 

4.2.3 From a load-displacement plot, determine the load at 
which the slope of the loading curve becomes equal to 
the slope of the upper portion of the unloading curve. 
(See Figure A5) 

4.2.4 From a load-"r educed" displacement plot, make the same 
determination as in 4.2.3. A "reduced" displacement is 
the difference between the measured displacement at a 
load and the displacement defined by a linear load- 
displacement relation at the same load. (See Figure A6) 

4.3 Nonvisual ( nonsubj ect ive ) method 

1. Collect load-displacement digitized data for a complete 
load cycle. At least 100, preferably more, data pairs 
should be taken to describe the load-displacement curve. 

2. Starting with the first data sample below maximum load on 
the unloading curve, fit a least-squares straight line to 
a segment of the curve spanning approximately the 
uppermost 25 percent of the cyclic load range. The slope 
of this line is assumed to correspond to the fully-open 
crack configuration. 

3. Starting with the first data sample below maximum load on 
the loading curve, fit least-squares straight lines to 



23 



segments of the curve that span approximately 10 percent 
of the cyclic load range and that overlap each other by 
approximately 5 percent of range (See Figure A7). Store 
the slope and corresponding mean load for each segment in 
an array. 

H. Starting with the first (highest load) slope in the array, 
compare the slopes to the open-crack slope and identify 
the location in the array beyond which the slopes always 
exceed the open-crack slope by at least a specified 
percentage . 

5. Starting at the array location identified in Step M, find 
the nearest, higher-load array location beyond which the 
slope is always less than the criterion level in Step ^. 

6. Determine the opening load corresponding to the specified 
exceedance criterion by interpolating between the two 
slope-load points identified in Steps U and 5. (See 
Figure A8) 



For the Tas 
evaluated 
levels of 



sk Group test program, ope 
for several exceedance lev 
0.5, 1, 2, and 4 percent. 



^ning loads should be 
vel criteria. I suggest 



5. Methods of Opening Load Determination Not Based on Displacement 
Data 

Participants are encouraged to make measurements by all methods 
that they feel are viable approaches. 

6. Test and Measurement Conditions 

6.1 Ambient laboratory environment 

6.2 Test loads 



The K 



spec 

shed 

cont 

tip 

inch 

fact 

meth 

for 

assu 
spec 
the 

at t 
this 
spec 



max 
if ie 
ding 
rol 
betw 
es a 
ors 
ods 
the 

med 
imen 
init 
he e 
poi 
imen 



and R values are to be maintained constant at the 



d va 
loa 
is u 
een 
nd s 
shou 
give 
test 

crac 
, an 

ial 
nd ( s 
nt a 
and 



lues 
d ei 
sed , 
adju 
houl 
Id b 
n in 
by 

k le 
d 2a 
load 
) of 
nd c 
2a = 



dur 
ther 

the 
stme 
d no 
e ca 

Sta 
usin 

ngth 
-0.8 
s un 
the 
ont i 
2.00 



ing the 
by man 
increm 
nts to 
t excee 
Iculate 
ndard E 
g the s 

s of : a 
inche 

til era 
EDM si 

nue unt 
inches 



init 
ual o 
ent o 
the 1 
d 0.0 
d ace 
647. 
pecif 

= 0.90 
s for 
cks a 
ot . L 
il: a 
for 



ial pa 
r comp 
f crac 
oad sh 
3 inch 
ording 
Determ 
ied CO 

inche 
the M 
re det 
oad sh 
=T. 5 
the M( 



rt of th 
uter con 
k growth 
ould be 
es. Stre 

to the 
ine the 
nstant-K 

s for th 
(T) spec 
ected on 
edding s 
inches f 
T) speci 



e test by 

trol . If manual 

at each crack 

nominally 0.02 

ss intensity 

equations and 

initial loads 

value with 
max 

e C(T) 

Imen. Maintain 

both surfaces 

hould begin at 

or the C(T) 

men . 



6.2.1 R=0.1 for all fatigue loading 



24 



rrT" 



6.2.2 Low growth rate specimen 



o Constant-K 



max 



6.0 ksi-ln 



1/2 



o Overload-K 



max 



13.5 ksi-in 



1 /2 



Apply a single load cycle such that the stress 
intensity factor cycles from 0.6 to 13.5 to 0.6 

ksi-in 
o After overload 

After the overload, the fatigue loading should be 
R»0.1, constant-amplitude loading such that K 



max 



1 /2 



6.0 ksi-in for the crack length immediately after 
the overload. 

6.2.3 High growth rate specimen 



o Constant-K ■« 20.0 ksi-in 
max 



o Overload-K 



max 



35.0 ksi-in 



1/2 



1/2 



Apply a single load cycle such that the stress 
intensity factor cycles from 2.0 to 35.0 to 2.0 
ksi-in^/2 



After overload 

After the overload, the fatigue loading should be 

R-0.1, constant-amplitude loading such that K 



max 



20.0 ksi-in 
the overload 



1 /2 



for the crack length immediately after 



6.3 Opening load measurements 

6.3.1 Low growth rate specimen 



o At crack length no . 1 



Make 
measu 
measu 
dur i n 
and r 
displ 
load 
perce 
fatig 
ments 
not g 
gr owt 



mult 
reme 
ring 
g th 
emou 
acem 
cycl 
nt o 
ue c 
. Th 
o be 
h in 



iple, 
nts ( 
appa 
e fat 
nted 
ent d 
es th 
f the 
rack 
e min 
low t 
creme 



indep 
prefer 
ratus 
igue 1 
for ea 
ata ar 
at hav 

max . 
growth 
. load 
he min 
nt . 



- C(T): a=1.00 +or- 0.01 Inches 

M(T): 2a=1.00 +or- 0.01 inches 
endent crack opening load 
ably as many as 10). If the 
does not normally remain in place 
oading, then it should be removed 
ch measurement. If load- 
e taken, make recordings during 
e max. load - kO , 70, and 100 
load for the last increment of 
before the opening load measure- 
in the recording cycles should 
. fatigue load for the last 



25 



At crack length no 
(Same instructions 
At crack length no 



2 - C(T): a=1.10 +or- 0.01 inches 

M(T): 2a-1.20 +or- 0.01 inches 
as at length no . 1 ) 



3 _ c(T): a=1.50 +or- 0.01 inches 
M(T): 2a-2.00 +or- 0.01 inches 
(Same instructions as at length no . 1 ) 

o At the overload length-C(T): a-1.50 +or- 0.01 inches 

M(T): 2a-2.00 +or- 0.01 inches 
After making the opening load measurements at length 
no. 3, apply the specified single overload cycle. Then 

at 0, 2.5x1o\ 5x10\ 1x10^. 2x10^, and HxlO^ total 
fatigue cycles after the overload, make opening load 
measurements during load cycles that go to 100 
percent of the post-overload max. fatigue load. 

6.3.2 High growth rate specimen 

o At crack length no . 1 - C(T): a=1.00 +or- 0.01 inches 

M(T): 2a=1.00 +or- 0.01 inches 
(Same instructions as for crack length no . 1 for low 
growth rate specimen except only record load- 
displacement data during load cycles to the 100 
percent fatigue load level.) 

o At crack length no. 2 - C(T): a=1.10 +or- 0.01 inches 

M(T): 2a=1.20 +or- 0.01 inches 
(Same instructions as at crack length no.1) 

At crack length no . 3 - C(T): a-1.50 +or- 0.01 inches 

M(T): 2a-2.00 +or- 0.01 inches 
(Same instructions as at crack length no.1) 

o At the overload length-C ( T) : a=1.50 +or- 0.01 inches 

M(T): 2a=2.00 +or- 0.01 inches 
After making the opening load measurements at length 
no. 3, apply the specified single overload cycle. Then 

at 0, 2.5x10^, 5x10^, 1x10^, 2x10^, and 4x10^ total 
fatigue cycles after the overload, make opening load 
measurements during load cycles that go to 100 
percent of the post-overload max. fatigue load. 

7. Documentation of Tests and Results 

Participants should thoroughly document their experimental 
apparatus and procedures. All details will not be required in the 
initial reporting of the results, but the need to investigate the 
effects of differences in experimental procedure may become 
evident after an initial analysis of the results. Raw data should 
be stored so that they can be easily retrieved for further 
analysi s . 



26 



n I 



The following should be Included in the Initial report of results 
for each specimen: 

o Specimen number that is scribed on the specimen 

o Plot of growth rate vs. crack length 

o Experimental approach - load-displacement, potential drop, 
ultrasonics, etc. 

If load-displacement approach is used, indicate type of 
displacement measurement apparatus (COD gages, strain 
gages, interf erometry , etc.) and measurement location 
on the specimen. 

o Crack length at measurement and, after overload, the cycle 
number 

o Maximum load in recording cycle 

o Opening loads reported in chronological order of 
determination 

If load-displacement approach is used, then opening 
loads should be reported for all the analysis methods 
listed in paragraph ^. 



27 



Plate 

rolling 

direction 



Detail A 



3.60 



3.75 



^.750 




\r 



See detail A for 
enlarged view 



eniargea view — ». ^— ^ 



A 



___ +.002 ., 
.750 ..000 «"«•• 

2 places 

T 

1.80 + .01 



825 



©- 



.825 



1.80 + .01 



^.750 
Note: all dimensions In Inches 




Figure A1.- C(T} specimen to be used In the Round Robin 



Plate 

rolling 

direction 




I.OOOdla. 
2 places 



.105 



.01 OR 
2 places 



-\ 



.295 


.295 







ID ±. 



105 



Enlarged 
view 



Cross 

section 

A-A 




60* 



-.020 




L.40 J. .40 J 
r±.01*p±.01*| 

Note: all dimensions In Inches 



i 

3/8 



Norn. 3/8 1 



Figure A2.- M(T) specimen to be used in the Round Robin 



28 



TT"T: 



Load 

Opening 
ioad 




straight line ^^^^ 


X 


^^ Opening load Is load at which upper 
< portion of load-displacement curve 
}ecomes linear 

- Data (loading & unloading) 



Displacement 



Figure A3.- Determination of opening load from load-displacement 
data according to method of Test Plan paragraph 4.2.1 



Load 



Opening 
load 



Upper straight line 




Opening load Is load at Intersection 
of straight lines drawn tangent to 
upper & lower portions of load- 
displacement curve 



Data (loading & unloading) 
Lower straight line 



Displacement 



Figure A4.- Determination of opening load from load-displacement 
data according to method of Test Plan paragraph 4.2.2 



29 



Opening load Is load at which the slope of the 

loading portion of the load-dispiacement 

curve becomes equal to slope of upper portion 

of unloading curve 



Load 



Openini 

iOd' 



Parallel 
lines 



Loading curve 




Unloading curve 



In low-hysteresis situations, this method 
Is equivalent to that shown in figure A3 



Displacement 



Figure A5.- Determination of opening load from ioad-dlspiacement data 
according to method of Test Plan paragraph 4.2.3 



Load 



Opening 
load 



Opening load is load at which the slope of the 

loading portion of the reduced displacement 

curve becomes equal to slope of upper portion 

of unloading curve 



Straight line 




Difference between 
data & straight line 
is "reduced displacement 



Data (loading & unloading) 



Parallel 
lines 




Displacement 



"Reduced" displacement 



Figure A6.- Determination of opening load from load-displacement data 
according to method of Test Plan paragraph 4.2.4 



30 



ni Ti 



Load 



Line segment 




Each segment spans 
approx.iO%of the 
cyclic load range 



Displacement 

Figure A7.- Evaluation of slopes of load-displacement curve for use In evaluating 
the opening load according to method of Test Plan paragraph 4.3 



Opening load is the maximum load at which the 

slope of the load-displacement curve Is always 

at least "X"% greater than the average slope 

of the upper portion of the curve 



Load 



Opening 
load 
(1%) 




2 3 4 

Slope Increase, % 



Figure A8.- Determination of opening load from slopes of load-displacement 
curve according to method of Test Plan paragraph 4.3 



31 



APPENDIX B 
Round Robin Opening-Load Data Set 

Table B1.- Opening-load ratios for low-K test of C(T) specimen 



Participant 
number 



CL1 



CL2 



CL3 



Measurement time 

NO N1 N2 N3 



N4 



(A) COD compliance measurement 



7 
8 
1 1 

8 

1 1 

8 
9 

1 
6 
8 
10 



0, 










20 
30 
21 

21 
21 



19 
32 
15 

22 

15 



Upper Linear Displacement 
0.211 0.25 — 0.26 0.2i4 
0.33 0.27 0.27 0.27 0.32 
0.15 0.10 0.18 0.18 0.15 
Intersect ion 



o.zn 0.31* 



H5 
MO 
33 
33 



36 
142 
33 
33 



0.18 0.19 

0.14 0.10 
Reduced 
0.311 

0.19 
Nonvisual {^%) 

0.54 0.27 0. 

0.36 0.35 0. 

0.32 0.28 0. 

0.36 — 0. 



19 
18 



19 
18 



0.1 9 



Di splacement 
28 0.30 0.29 



(B) BFS compliance measurement 



N5 



0.26 0.28 

0.36 0.35 

0.16 0.18 

0.20 0.20 

0.16 0.18 



0.31 0.36 0.37 



H7 0.37 0.55 0.32 0.119 

35 0.36 0.36 0.36 0.36 

29 0.29 0.30 0.32 0.32 

32 0.32 O.3M 0.33 









Upper 


Linear Displacement 








5 


-- 


0.35 


0.25 


0.30 


0.31 


0.31 


0. 


32 


0.31 


8 


0.33 


0.39 


0.35 


0.27 0.33 


0.32 


0.32 


0. 


,40 


0.37 


9 


0.26 


0.21 


0.21 


0.23 


-- 


— 




. — 


— — 


1 1 


0.19 


. 1 11 


0.18 


0.15 0.20 


0.20 


0.20 


0. 


, 19 


0.20 








Intersection 












5 


— 


0.29 


0.25 


0.30 


0.31 


0.31 


0, 


.32 


0.31 


8 


0.23 


0.23 


0.18 


0.19 0.19 


0.1 9 


0.20 


0, 


,20 


0.20 


9 


0.21 


0.18 


0.18 


0.17 


— 


— 




— 


— 


1 1 


0.17 


0.1 4 


0.18 


0.15 0.18 


0.18 


0.20 


0, 


.18 


0.18 








Reduced Displacement 










8 


0.1»3 


0.116 


0.41 


0.32 0.117 


0.62 


0.41 


0, 


.58 


0.58 


9 


0.21 


0.19 


0.19 
Nonvi 


0.20 
sual (1$) 












1 


0.21 


0.23 


0.37 


0.27 0.119 


0.37 


0.41 


0, 


.34 


0.26 


8 


O.HO 


O.MI 


0.36 


0.30 0.31 


0.32 


0.32 


0, 


.35 


0.37 



32 



Table B1.- Opening load ratios for low K test of C(T) specimen 



Part icipant 
number 



CL1 CL2 
(C) IDG compliance measurement 



Measurement time 
CL3 NO N1 N2 



N3 



^4 



5 
1 1 

5 
1 1 

5 

5 



N5 



Upper Linear Displacement 

0.59 O.J49 0.31 0.29 0.37 0.37 0.36 0.37 0.^40 

0.1J9 0.42 0.47 0.20 0.28 0.30 0.33 0.30 0.30 

Intersection 

0.53 0.38 0.30 0.23 0.37 0.37 0.36 0.37 0.40 

0.40 0.37 0.40 0.18 0.20 0.23 0.25 0.23 0.23 

Reduced Displacement 

0.56 0.50 0.40 0.37 0.37 0.35 0.37 0.37 0.55 

Nonvisual {^%) 

0.64 0.38 0.31 0.43 0.47 -- 0.41 



33 



Table B2.- Opening load ratios for high K test of C(T) specimen 



Participant 
number 



CLl 



CL2 



CL3 



Measurement time 
NO N1 N2 



N3 



nM 



N5 



(A) COD compliance measurement 

Upper Linear Displacement 

2 -- — 0.1M 

7 0.15 -- 0.16 0.14 0.13 O.IM 0.15 0.15 0.16 

8 0.17 0.20 0.2M 0.16 0.18 0.17 0.32 O.HH 0.53 
11 0.10 0.10 

11 0.10 0.10 0.12 0.10 0.10 0.10 0.10 0.10 0.10 

Intersect 1 on 

8 0.1^4 0.17 0.18 O.IM 0.15 0.15 0.16 0.16 0.18 
11 0.10 0.10 

11 0.10 0.10 0.12 0.10 0.10 0.10 0.10 0.10 0.10 

Reduced Displacement 

8 0.15 0.21 0.29 0.18 0.19 0.19 0.20 0.65 0.69 

Nonvisual (1?) 

1 0.25 0.117 0.38 0.49 0.47 0.50 0.49 0.57 0.52 

3 0.17 0.22 0.26 0.10 0.12 0.11 0.10 -- 0.57 
6 0.21 0.28 0.29 0.17 0.18 0.17 0.17 0.44 0.61 
8 0.16 0.21 0.27 0.20 0.21 0.22 0.23 0.43 0.62 
10 0.33 0.31 0.30 -- -- -- 0.22 0.30 0.56 



(B) BFS compliance measurement 



2 
5 
8 
9 
1 1 
1 1 

5 
8 
9 

1 1 
1 1 

8 
9 

1 
3 
8 





0, 



0, 





10 
24 
15 
10 



0.10 



10 
18 
13 
10 
10 



0.54 
0.15 

0.16 
0.30 
0.30 











10 
25 
18 
1 
10 



0.10 
0.20 
0.15 
0.10 
0.10 

0.56 
0.16 



46 
29 
31 



Upper Linear Displacement 



15 
10 
28 
16 
1 1 
12 
Inter 
0.10 
19 
13 
1 1 
12 
Reduc 
0.54 
0.16 
Nonvi 
0.47 
0.35 
0.31 











0. 
0. 
0. 
0. 
0. 



10 

20 

19 

1 2 

10 

section 
0.10 0. 

15 

1 2 

1 2 

10 

Di spla 

59 0. 

1 3 0. 

aual (U) 

0.38 0. 

0.20 0. 

0.24 0. 



0, 
0, 
0, 
0, 
ed 

0, 



0. 
0. 
0. 
0. 



10 

31 
14 
14 
10 

10 
15 
12 
14 
10 

















10 
37 
16 
1 
10 

10 
16 
1 2 
1 
10 



cement 
60 0.73 
13 0.13 



52 

15 
24 



0.50 



0.47 
























10 
44 
18 
1 2 
10 

10 
17 
12 
1 2 
10 

77 
13 

.54 
,29 
.55 












10 
53 
19 
1 4 

10 



0.10 
0.18 
0.13 
0.1 4 
0.10 

0.76 
0.13 

0.57 

0.64 

















10 
59 
19 
1 2 



0.10 



1 
19 
13 
12 



0.10 



74 
13 

28 
59 
67 



34 



ilT' 



Table B2.- Opening load ratios for high K teat of C(T) specimen 



Part icipant 
number 



CL1 



CL2 



CL3 



Measurement time 

NO N1 N2 N3 



N4 



N5 



(C) IDG compliance measurement 



5 
1 1 



Upper Linear Displacement 
0.49 0.51 0.49 0.55 

0.37 0.43 0.36 0.17 0.54 0.58 
0.42 0.39 0.32 -- 0.38 0.52 

Intersect ion 
0.44 0.45 0.44 0.52 

0.34 0.39 0.32 0.15 0.51 0.54 
0.37 0.35 0.28 — 0.31 0.38 

Reduced Displacement 
0.43 0.46 0.46 
0.43 0.52 0.43 
0.59 0.62 0.59 

Nonvlsual i^%) 
0.67 0.55 0.55 0.51 
0.59 



55 
63 



65 
65 



0.62 
0.69 



53 
43 



0.61 
0.44 








57 
53 



35 



Table B3.- Opening load ratios for low K test of M(T) specimen 

Participant Measurement time 

number CL1 CL2 CL3 NO N1 N2 N3 NM N5 

(A) COD compliance measurement 

Upper Linear Displacement 

8 0.38 0.32 0.29 0.24 0.27 0.24 0.26 0.27 0.29 

9 0.37 0.35 0.30 0.37 

Intersection 

8 0.2U 0.21 0.19 0.17 0.18 0.18 0.18 0.18 0.17 

9 0.28 0.26 0.2U 0.2M 

Reduced Displacement 

8 0.50 0.39 0.39 0.29 0.30 0.29 0.33 0.36 0.37 

9 0.il8 0.41 0.37 0.41 

Nonvisual {^%) 

3 — ~ 0.28 0.33 0.24 0.23 0.30 0.23 0.37 

6 0.45 0.44 0.43 0.42 0.41 0.41 0.41 0.41 0.42 

8 0.44 0.36 0.34 0.28 0.31 0.28 0.30 0.32 0.32 

(B) IDG compliance measurement 

Upper Linear Displacement 

5 0.49 0.53 0.48 0.29 0.32 0.39 0.36 0.42 0.40 

11 0.38 0.31 0.52 0.30 0.32 0.29 0.32 0.31 0.33 

Intersect i on 

5 0.45 0.39 0.43 0.22 0.25 0.26 0.26 0.29 O.31 

11 0.31 0.27 0.49 0.27 0.28 0.26 0.28 0.28 0.29 

Reduced Displacement 

5 0.45 0.43 0.43 0.26 0.37 0.41 0.46 0.38 0.36 

11 0.41 0.38 0.58 0.34 0.41 0.40 0.39 0.34 0.35 

Nonvisual (1J) 

5 0.55 0.62 0.62 0.62 0.65 0.53 O.6I 0.62 0.67 



36 



ii I 



Table B4.- Opening load ratios for high K test of M(T) specimen 

Participant Measurement time 

number CLl CL2 CL3 NO N1 N2 N3 NU N5 

(A) COD compliance measurement 

Upper Linear Displacement 

8 0.23 0.26 0.27 0.30 0.29 0.30 0.i«6 0.5^ 0.60 

9 0.20 0.21 0.33 0.35 0.38 0.38 0.37 0.36 0.34 

Intersection 

8 0.15 0.19 0.18 0.17 0.17 0.20 0.21 0.22 0.26 

9 0.16 0.18 0.25 0.27 0.22 0.22 0.26 0.25 0.25 

Reduced Displacement 

8 0.27 0.31 0.33 0.38 0.40 0.52 0.63 0.65 0.68 

9 0.23 0.26 0.32 0.M3 0.36 0.38 0.37 0.39 0.36 

Nonvisual (1?) 
3 0.49 0.43 0.41 0.25 0.31 

6 0.34 0.35 0.37 0.36 0.36 0.41 0.47 0.51 0.61 
8 0.25 0,30 0.32 0.26 0.32 0.35 0.47 0.6l 0.65 

(B) IDG compliance measurement 

Upper Linear Displacement 
5 -- — 0.46 

Intersect ion 

5 -- — 0.43 

Reduced Displacement 
5 __ -- 0.41 

Nonvisual (1%) 
5 -- — 0.69 



37 



NASA 

Nanooai f^or^auKS ana 



Report Documentation Page 



1. Repon No. 

NASA TM-101601 



2. Government Accession No. 



4. Title and Subtitle 



Results of the Round Robin on Opening-Load Measurement 
Conducted by ASTM Task Group E24.04.04 on Crack 
Qosure MeasurenKnt and Analysis 



3. Recipient's Catalog No. 



5. Report Date 

May 1989 



6. Performing Organization Code 



7. Author(s) 



8. Performing Organization Report No. 



Edward P. Phillips 



10. Worl< Unit No. 

505-63-01-05 



9. Performing Organization Name and Address 

NASA Langley Research Center 
Hampton, VA 23665-5225 



11. Contract or Grant No. 



12. Sponsoring Agency Name and Address 

National Aeronautics and Space Administration 
Washington, DC 20546 



13. Type of Report and Period Covered 

Technical Memorandum 



14. Sponsoring Agency Code 



15. Supplementary Notes 



16. Abstract 

An experimental Round Robin on the measurement of the opening load in fatigue crack growth tests has 
been conducted by the ASTM Task Group E24.04.04 on Crack Closure Measurement and Analysis. The 
purpose of the Round Robin was to evaluate the current level of consistency of opening load measurements 
among laboratories and to identify causes for observed inconsistency. Eleven laboratories participated in 
the testing of compact and middle-crack specimens. Opening-load measurements were made for crack 
growth at two stress-intensity factor levels, three crack lengths, and following an overload. All opening- 
load measurements were based on the analysis of specimen compliance data. When all of the results 
reported (fh)m all participants, all measurement methods, and all data analysis methods) for a given test 
condition were pooled, the range of opening loads was very large-typically spanning the lower half of the 
fatigue loading cycle. Part of the large scatter in the reported opening-load results was ascribed to consis- 
tent differences in resuhs produced by the various methods used to measure specimen compliance and to 
evaluate the opening load from the compliance data. Another significant portion of the scatter was ascribe*^ 
to lab-to-lab (Hfferences in producing the compliance data when using nominally the same method of 
measurement 



17. Key Words (Suggested by Author(s)) 

Cracks 

Crack Closure 

Crack Propagation 

Fatigue (Materials) 

Metol Fatigue 



19. Security Classif. 

Unclassified 



(of this report) 



18. Distribution Statement 



Unclassified - Unlimited 
Subject Category - 39 



20. Security Classif. (of this page) 

Unclassified 



21. No. of pages 

38 



22. Price 

A03 



NASA FORM 1626 OCT 86 



: