Académique Documents
Professionnel Documents
Culture Documents
F1
~~ ~~~ T~ - 77-0717
n p ,3 r I ~ nu ~~ip ~~n
13N ~~~fl flU ~~ 3 ~ N (
: c
>-
1
~ ~ Tel-Aviv, Israel
C,
Prepared for
AIR FORCE OFFICE OF SCIENTIFIC RESEARCH
Grant AFOSR 7630 14
1 L U N I T ~ ( . LA S ,Ii 0 A T L N )
~
T~~~S PA ,F ~ ~~~~~~ t) , , , t , F , ~ f . , 1)
~ 0 1 -
_______________________-
4 TITLE S T Y P E OF R E P O R T
~~ P E R I O D CO ~ E R E D
,.r,d So b t , I I . I
INTERIM
/ ,.~ A CUMU LATIVE DAMAGE THEORY OF FATiGUE FA LURE
1 ______________________________
6 P ERFQRJ ,jIN
-~~ IT A U S O E / 3 9 5 7 7 . ~~~~~~~~~~~~~~~~~~~
7 A j T H O R ~
. ~~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Z / II A SH IN
4
A ROTO M
AF ~ SR ~~~~3O l 4 ~~~~ .
- -
9 PERFORMING O RGA N IZATION N A M E AND A D D R E S S 0. P R O G R A M E L E M E NT . PR O J E C T . T A S K
A R E A & WORK UNIT NUM B --
~~~~~~
TELAV I \ UNIVERSITY .
~ /1 /
S C H O O L OF E N G I N E E R I N G : . ~~~~~ -
~~ 6 1102F
-
R A M A T A V I V , TEL AVIV 6q978 , ISRAEL
II. CC r 4 T R O L L IN G O F F I C E N A M E A N D A D D R E S S 2. REPORT DA T E
U N C I ASS I F l E D
l5 DECLASSIFICATION DO W NG R A D I N U
~
SCHEDULE
DD ,
~~~~~~~~
1473 F C TI tl OF I NOV hS IS O 8 S O L E T F .
II N( 1 ,\ 5S I rI I
II
( -- S F C U R I I Y C L A S S I F I C A T i O N OF T N I S PA O F (i$?,Cfl t).ta Rote,,. ,)
_ _ _ _ _ _ _
~~~~~~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~~~~~~~
- - - - -
~~
-
~~~~T
T~~~
by
De p a r t m e n t el Sol Id N i i h an . i c s ,
M a t erj ) , 1 s and S t r uc t u r es
O f f i c e of A er o s p a c e R e s e a r c h (EO AR ) , United St at e s A i r Fo r c e .
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
_
_
_
_ _ _
-A
~I CE OF S CI T . 1 I I C
j~~::~~~,J.
. SMITT AL TC .
~
pu~ 1i~ r m o _
~J ..v AFR 1 3 L -
.
-I I~ u o t L - .i ~~~~~~
~~~~~
orantlon O ffloe r
~~~~~~~
_ _ _ _ _
_ _
A CUMULATIVE DAMAGE THEORY OF FATIGUE FAILURE
by
*
Z. Hashin
Dept. of Solid Mechanics , Materials and Structures
School of Engineering
Tel Aviv University
Tel Aviv , Israel
and
A. Rotem **
Dept. of Mechanical Eag ineering
Technion Israel Institute of Technology
Ha i fa , Israel
ABSTRACT
concept of damage curve families and on an equivalent loading postulate which de-
fines specimens that have suffered identical damage under different loading pro-
grams. Lifetime analysis has been per tormed far various cases of p i e c e w i s e con-
stant and continuous variation of cycle amplitude. For continuous variation , the
:.
-
~
.
~ htt ~~ ,. - . ,, i
-
*
Professor tnd Chiirnu n
.
** S e n i o r F ,u t u r i . r
- .
t~ Glt i~~~; ~~ . ;~-
, ..
~
.
- ~~~~~ 2- -( (
1
1. Introduction
,
~ structural part when the amplitude of cyclic load varies in prescribed fashion
program.
~slrio n . It should , h owever , be borne in mind that fati gue failure test d a t a a re
l-
~u b J e c t to c o n s i d e r a b l e s c a t t e r . In o r d e r n ot t o f u r t h e r comp l i c a t e .i v e i -.
d i ff i~ u t
,
~en e. . The problem considered in t h i s sense has become known as tumu l tit IVE d , min ~~~e
t h e o ry . It has been t h e s u b j e c t of n u m e r o u s i n v e s t i g a t i o n s, p a r t i c u l a r l y in t h e
l ist thirty years. Since there are many excellent reviews of previous work , E . g.
- --- .-- I
~ ------ .--- - - , - _ _ _
- TI: T:--
-
~ ~
to the case w h e r e t h e l o a d i n g is p i e c e w i s e c on s t a n t , h e r e r e f e r r e d to as m u l t i
~~ 2
F( , ~ , . . . ) (i.l)
1 2 k
it is stipulated that
0~~~ F a ~ l (1.2)
A simplistic and well known damage function has been postulated by Palmgren
[5] and by Miner [6] and has become known as Miner s rule. Ac cording to them ,
~
- = 1 (1.3)
whe re
n. number of c y c l e s at level o.
1 1
N l i f e t i m e at c o n s t a n t a m p l i t u d e c .
1
This simplistic approach does not in general comp l y with reality. It is knowi
It should be noted that the only material characteristic entering into Mimer s
ule is the SN curve for constant amp litude loading. It seems unreus.nab le to
assume that such simple information would be sufficient for lifet ime predicti on
Ma rco and Starkey [7] assumed that damage produced by n cycles at level ~
cilure condition which for two -stage loading assumes the form
n
1 102 2
( )~ + 1 (1.~+)
N N
1 2
,:nd can be generalized to multista ge loading. Appar ently , the dependenc e of the
cppear to be an easy task. It will be seen that (1.4) is a specifi special case
f the theory to be developed in this work which does specify the dependence of
u. on ~ in unequivocal fashion .
Shanlt - [ 1 h~ s c o n s t r u c t e d a t h e o r y in w h i c h i t is p o s t u l a t e d t h a t d a m a g e
~~
tn be described in terms of C~ 0n. Corten and Dolan [9] described damage in the
a
t ,, rm Cn . A detailed account of t h e s e and o t h e r a p p r o a c h e s may he fo u n d in [ 1 ] .
I Ii damage I ur v e I ar e d e f i n e d in t e r m s of t h i s q u a n t i ty .
- - .
,
~ -- - -
- - -
~~
The C o n c e p t of Damage C u r v e s
st~ he ~. onsidered as a damage region. Each of the points within the damage region
p I . t n c , Fig. 1 , which is called the damage curve. Evidently, the damage curve u n d e i
s~, -~c n t tally defined by its residual lifetime which is a measurable quant ity.
The basic assumption is made that a damage curve is uniquely defined by its
n it i al point. This implies that if any other point on the dam ogt c u rv e n , 0 2,
-
2
at , is interpreted as the first stage of a twostage cycling experiment , the
- - - --
~ ~~ - -- - - ----- - -- - - -- - - - -- - - -
:: - -
.. . ~~~~~~~~~~~~~~~~~~~ ~~ ~~~~~~~~~ ~~~~ ~~~ ~~~~~~~ ~~
i
i
- 5 -
Uni queness of the damage curve passing through a point and subsequent analyti cal
flow be explained. For this purpose we shall introduce time concept of equivalent cycl i c
lo idiugs . Cyclic loading programs are termed equivalent for stress amplitude a if
t a r subsequent ccn stant amp litude loading , at same stress level a , residual lifetime s
ire the same . The equivalent loading postulate is now stated as follows : Cyclic load-
ings which are equivalent for one stress level are equivalent for all stress levels.
the damage curve , Fig. 1, as having been obtained by a series of twostage loadings
loading programs the specimens are cycled to failure at some other stress level , c~,
r
the residual lifetime will be the same n , Fig. 1. Therefore the loadings
nsa the average of maximum and minimum amp litudes in a cycle and R
-- - . - -~~~~~ . -.-
- - - - - - - --
- - ~~~~~~~~ -~~~~~~ --~~~~~~~~ - - - . -~~~~-
- 6 -
Recnuse of the inherent scatter of fatigue testing, all damage curves must
be interpreted iro some best fit sense, as is the case for S-N curves.
The equations of the damage curves are ~-itten in the general form
o = o(n,y) 0
8
f(n,y) (2 .1)
Evidently, since static failure occurs at one quarter cycle which is consid~red
t(O,y) 1 (2.2)
Powever, curve fitting difficulties may require use of a 0 value which is not
s
~~unl to the static test result.
To prove this assertion we consider the cnse of two low-high load cycling pro-
in the first progrnm nnd nl ' > in the second program. Obviously the remainin~
Ill u !tnt' N - n
2 2
nt a2 , ln the flrRt progrnm must be lnrger than the retr.,linin~
-,
0 (2.3)
2
to as in F i g . 2b .
ii a twostage l o a d i n g p r o g r a m w i t h a m p li t u d e s O ,a the l i f e t i m e is s m a l l e r t h a n
L). But the lifetime must be N(a) since the loading is teally one s ta g e .
v i e w of (a) .
.c--~-, t h r o u g h the p o i n t O;a and approach the n ,c axes cmsvmp tot i m ! iv. Ihe :r _ N
-
Lot t Im e t r m at h i n ii t it a l d e s c r i p t i o n is s i m il a r to t h a t of the SN curve. let
-
me eq un t ion if the SN cu r v e t,c r ep ct t ;e n t ed in t h e f o r m
8
= a f ( N , F) (2.4)
I-mere N denotes value of n for a point on the SN curve . Then the equati n
0 = o t ( n , y) (2.5)
ln ~- par ameter y for a damage curve passing through the point n ;0 is determin u d
1 1
by i n s e r t i n g t h o s e values i n t o (2.5) and solving for y.
I
0 = a(1 + F log N) (2.6)
0 0 N ~~~
N-O
It follows t h a t t h e c o r r e s p o n d i n g f o r m s of damage c u r v e s p a s s i n g t h r ou g h a
I t l e~ 1 pa t n t n 0~ a r e :
1
a = c ( 1 -t log n ) (a)
S
( 2 . 8 )
01 /0 1
= (h)
log n
1
A
~
a = on ~ ; lo g(a/o )= y log n (a)
(2.9)
lo g(0 /ct)
1
(b)
log n
1
The form (2.8) is generally more suitable for fiber composites while the form
(2.9) is more s u i t a b l e f o r m e t a l s .
N
ax is contrary to theorem (c). The conclusion is that for very low stress ampli-
o found b y static testing. In that case the former should be chosen and be
r, mnce -~~ D 0~~ , say, by the analytical expressions. In tha t case , the val id
liv of the damage curve expressions is limited to that loading range and C is
a t ittin g p ar a m e t e r .
~j , ei d he s i m i l a r l y represented .
h r ou g h the p a i n t n a g r o u p of s p e c i m e n s ar e subjected t o n c v i it s at
1
and ar e then v ,e l ~d t o fcm j l u r e a t d i f f e r e n t s t r e s s l ev e l s 3~
0 ich ot
~2
t iit 1 ttt i tiies ~It 0~ ~0 ~
3
... d e t e r m i n e a p o i n t on t h e damage c u r v e .
10
ihe d am -m g i c urves iOn be separatel y defined for two stage loadings whi h
it nbc - l owhi gh t y p e or of the highlow type. There then arise s the qtact i- UI
if b i b -c,t s of curves are identi cal. Consider for this purpose a load in c pr o
cr a m ~n wh i ii n
1
cy c l e s t a k e p l a c e at o , . Then the amplitude is ra~ sc 1
~
t or t h e dur .mt i n of a very small number of cycles. After that , the amp litud u
i~~ . r~~~. iii t-d to - mm to C 1, and cycling is continued until failure. A diagram i f
u~l mi ~1l to N
1
. the lit et mc - at then th c- segment A A~
1
is insigni fican t and
, ad t he P mc - 0 r cu r - c-
~ xhihit a h y s t e r e s i s phenomenon . On the basis ci e x p e r i c n ~ --
sift- -, .-, I ises with conse quent conclusions for the identity of upward and d o w n s m r l
(i i. Y e o r - . L S .
t ., t 0 , su ~ h e t t t e - t s m u s t be u n c o v e r e d b y e x p e r i m e n t s . This a p p e a l s to he ~
3. A n a l y s i s of Loading Programs
in terms of the damage curves is trivial since the damage curves have been
Let it be assumed that the SN curve is of form (2.6) and ti-mat the damage
(2 . 8) . It follows that
l og ( N n )
2 2
a = a [1 + ( ~ l 1)] (3.1)
2 s log n a
1
Sa l v i n g t o r n we h a v e
2
1s
1 (3.2)
-
n = N n
2 2 1
wher e
i i 0~)
= (3.3)
~1 it
~2 0
rhe quantit .-
_ j_ ( i .~~)
= ~ +
N N
1 2
1s
n 2 n
(i )ls l + 1 (3.5)
1 2
1 S2
~
(~~ .~. ) l 5 -1-
M = l + ~~ L
(3.6)
it is e a s i ly shown t h a t
M~~~~l s s
1 2
when (3.7)
M~~~~l s s
1 2
r e s u l ts
log s / l o g s
2
(
~~~~ ) N
=1
1 2
(3 . 8 I
log s / l o g s
n n 2 ~
(i- )
1
M = 1 +
N N
1 1
N o n that (3.5) and ( 3 . 8 ) are of the form (1.4) which is based on an expon
l
f
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
~~~~ ~~~~~~
under consideration . (In much v , l g I l er language we would say that the two loadings
r e s u l t in t h e same d a m a g e ) . C o n s e q u e n t ly , t h e t h r e e s t a g e l o a d i n g is e q u i v a l e n t
time n .
3
Evidently t h i s p r o c e d u r e c a n be g e n e r a l i z e d t o m o l t i s t a c - l i d logs s i h ,, t m
loading.
14
C o n s i d e r ci t h r e ~ stcige l o a d i n g , F i g . 5 , w i t h damage c u r v e s of s e m i l a g
ty pe. l~i t h t i l t n u t - m t i o n a d o p t e d
n
11
= 11 n 1, (3.9)
12
0 fl ~~~~~
3
Since the damage curves l ove , by hyp othesis , the same functional form as the
n 1s 2
( _i ) i . s i +
~
= 1 (3.10)
0
27
1 53
( - -
~~ ~~
- - ) ~~~2 + ~ -c~ 1 (3. 1 1 I
2 3
15 1-s
2 3
.
) 1 5 ~ ~ ~~ ]~~~~ 2 4 = 1 s . = ( 1. 12 )
ih r- t ri- t h e Min e r s i- u t I ic - te nt c g i v e n b~
. - -
~~~
15
1s
1s 3
n n n 2 n
M i + + -2
_ ) l -s + 2 ( 3 . 13 )
=
~
~~~~~~
~ ~~~
1 2 1 2
S i m i l a r a n a l y s i s f o r the loglog f o r m s g i v e s t h e r e s u l t s
log s / l o g s
log s / l o g s 3 2
n 2 1 n n
+
~~~~~
] +~~ =1
1 2
3
(3.14)
n n~ n log s / l o g s n log s / l o g s
1 2 1 3 2
N N N N
1 2 1 2
n = g ( y , s) s = 0/0 (3.15)
An a l t e r n a t i v e f o r m is
= (s , n ) (3 .lbu
16
n = n
_ Il l
= g ( r ~~ , s~~) =
n = + n
27 12 2 (3.17)
0
21
= g (y 2, s ) 2 -( (n ,s )
1
~ 22 2
~~~ + n ,
fl = fi
31
Il- ~~s stepho-- step procedure is continued until the SN curve is reached , th us
lhe functions p and for the two types of damage curves considered are
sI
g ( s ,~~ ) = 10
s I
f ( S ,n) =
-
iog n
i/I
=
loglog (3.19)
m- t 5
(s , n ) = - --- -
l~~o g
~
n
-j = it(fl) ( 1.20 )
- . - -- - -- - -
.-.. . --.-- - -
~~~~~~~~~~~~~~~~~~ ~ ~~~~ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
17
W1. t h invm r - , e
= i -i ( - : ) (3.21)
c u r v e at 0 . At the p o i n t t h u s r e a c h e d in t h e sn p l a n e , p r o c e e d a no t h e r do
the sn plane . I t is s e e n t h a t by t h i s p r o c e d u r e t h e l o a d i n g c u r v e C is
This p r o c e d u re of - n s t r u c t i on
~ of the image C of C w i l l now be expressed
G ( s , n) 0 s = 0/0 1,3.22)
The d i f f e r e n t i a l dn is g i v e n b y
dn (lO 4 d~ (3 .23)
ll
- ~~~~~~~~
~
18
O = n (3.24)
f)
fl g ( s , 1) (3.25)
f)
-
dn = -~~~~ ds (3.26)
D
= 1(s,n) (3.27)
n = h(s) ( 3 . 2 8)
= + h ( s ) ]ds (a)
(3.29
(= y( s ,n) (h)
h i i in p e r t in rmed
ui t
.- -- - - _ - - -~~ -- - - - - -~~
_ - -- -- ~~~
- 1.9 -
cln 4~(s,n)ds
Its integration stnrting out from an initial point n 'Cf gives the equation
o' o
ol- the l'\trVl' C', (3.22).
The intersection of (3.22) with the S-N curve defines the stress level 0u
at which failure occurs under the loading program (3.20). The cycle ordinate n
.tn t.hc S-N plane associated with the intersection point does not define the
N [;(o ) (3.30)
u l1
tu Integrate (3.29) until the image of the discontinuity in the s-n plane is
readlE'd. TlH_ jump in o then proceeds on the damage curve which passes through
the end point of integration reached, in the s-n plane. Then the integration
The ittnctions g(y,s) and y(s,n) which appear in (3.29) have been given
for s~mi-Jog and log-log damage curves by (3.18- 19). Therefore, for these
~og n
dn 2n 10 + h'/s)]ds (a) semi-log
s - l
(3.31)
the equations are readily integrated numerically by use of the Runge Kutta method,
n log n (3.33)
(a)
(3. 31~)
As an example we consider the loading program shown in Fig. Ra. The initial
(3.35)
With this value of h' (s) Qqus. (3.34) must be integrated starting out at the
initial point n = 0; s = s .
0
Note that n = 0 corresponds to n = 1, not n = 0.
This is an insignificant correction which is required because of the log repres-
cnUttion.
Integration of the differential equations gives the image C' of the loading
fLmction in s-n plane. This image is shmm in Fig. Bb for the semi-log case,
4
equ. (J.34n) when so= 0.7, n = 10 . It is seen that the image C' in the s-n
0
plane of the linear loading curve C starts out horizontally and then approaches
for l~rge n the second term on the right sides becomes very small in comparison
r.o thE: first terms because of lOn in the denominator. If the second term is
neglected the solution of the differential equations are the damage curves (2.8a),
(2.9:t) respectively. This is also evident from (3.23) since neglect of the. second
22
In o r d e r to c o m p a r e t h e p r e s e n t p r e d i c t i o n t o t h o s e of Miners r u l e it is
n e c e s s a r y to w r i t e M i n e r s r u l e f o r a c o n t i n u o u s loading f u n c t i o n . To do t h i s
JJ
0
0
do= l (3.37)
0 0
0 0
In t e r m s of the nondimensional
~/o S (3 . 3 7 ) assume s the f o r m
s t r e s s s =
ds = 1 (3.38)
si
P
N(s) = 10
( 3 . 39)
1/F
N(s) = s
~1
ds + = 1 (3. -~O l
wh i c h del i n e s t h e r e s i d u a l l i f e t i m e on t h e b a s i s of M i n e r s r u l e .
_ _ _ .~~~~~~~~ _ _ _ .
--
23
N(s )
F 0
= N(s ) _ _ _ _ _
[1
1 in 10 S ~~5 N( s ) (a) se m i l o g
0 1 1
(3 .41)
n = N(s )
r n
0
{s
N(s
-
)
- s J (b) l o g - l og
~ 1 0 1
~ ~
A n a l y s i s h a s a l s o been p e r f o r m e d f o r t h e c a s e of p u r e l i n e a r v a r i a t i o n of
to o b t a i n such intormati on .
A rati 00:1 ! ph en omenol ogicit theor y t o compute I a t igue I i fet m e under irbi t r a r v
cmi ry e a n ti I y t I cli i r e p re se nt . at i o n s .
_ _ _ _ _
-
24
The theory is based on a basic assumption which has been termed the
have s u f f e r e d the same damage tinder different loading programs Th~ pos t uiit .n
in terms of t h e damage c u r v e s .
cycle amplitude ~-.ith number of cycles. In the first case analysis consists of
a s imp le step by step procedure in terms of damage curves. In the second ease
solution of i n i t i a l value problems for first order nonlinear differential equati ons
is required.
Damage c u r v e f a m i l i e s c o n s i d e r e d in t h i s w o r k a r e s t r a i g h t l i n e s in i~~g n ,
or in log n , log a p l a n e s .
on ~~~~~~~~ rule. Ana ly sLs of a few cases of continuous amp ] :tude ~ cin m - m t :o n pr gn m~
_ _ -
--- - . - - ~~~~~
~1
~~~~
25
Ret erences
7. S .M. Marco and W .L. Starkey , A Concept of Fatigue Damage , Trans. ASME 76 ,
627 (1954).
- - - - - - - . -~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
- - _ - - ~~~
-~~~~~~~-~~~
Ui
> 1
0 z
2
Ui
>
)
(
~ I
--
~~~~~~~
-- --
I
I Ui
0
I
Ui I
I
I
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ z
2
F I
b b
~~~~~ ~~~~~~ b
_______
0~~
o_
$ s~
~2 n2
I!
~I I~I n
(a) (b )
_ _ _ _ _ _ _
Fit ; . iA (W OJRVhS
}{YSTFRFSIS FFFFCT
!( ~ 1iA1~
~
_ _ , -
I
J;
c~d
I 0
-J
/
, 0.
I P ~~~~
-.
/
,
I /
I
.
I
. (I) ~~ , a-
N /
~~~~~~
I L ~J
-
~~~~~~
/
.
/ , 8
II
- I / / -J
U)
I / / I
/
I
0 o
.1
0 II~~
1
,
,
ciz a
N
A?
~ 71
,
/
~~~
I
0 (
0 c.J
~
0 0 0
A-
- --_ -
~ 1~~~~
_
II
,
Ui
> 1
I
~ I I
U
~ I I
z/ F
I
(
0
I
I
2
4
- - --- -
- - - - -. - - -- -
_ _
~ ~~ ~~ ~~~~
>
c .0
U -
2 - -
C
U)
~~~~~~~~~~
_ _
1
~ ~~~~
a
I I, 0
b~~~~
_ . --- . -~~~~~ - - -
-- -- - - - ~~~~~~~~~~
- ~~~~~~~~~ _
-
~~~~~~~~~~~~~~~~~
!:Iti
IC
C
!I ..:
.j.
~ ~
~~~~~~~~~~
I
iV !
0
~~ ~
N / /
~~~~~~
s ~~~~~
o
_
_ _
~~~~ ~0
so /V -j
N
_ I/i
0 I A
N N
0 ~~.
0, C/ / i I
i/i I I
z
I lj I()
I/ l W I I
I/ J o I
// I
,
I I
1
/ I I I :
//
I, I I
if I I
I I I
I I I
I I I
I I I I
I I I
I I
L I L..
0
- (0
I I I 1 1
N
0
o o o o
.
- _ .- _- - ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ - _ _