Beruflich Dokumente
Kultur Dokumente
Fracture Analysis
Consultants, Inc.
Objectives
2
Part I
1
3/28/2017
Workshop Agenda
3
Part I
2
3/28/2017
5
Part I
3
3/28/2017
7
Part I
Advantages:
Does not (or should not) require in-service inspections
Disadvantages:
Even with built-in conservatism and large scatter factors, cracks
and failure do sometimes occur before the design safe-life is
reached.
Many (most) components are retired unnecessarily
If cracking is discovered in-service, safe-life provides no
information about the relative risk due to the damage.
Does not account for in-service loads that differ from those
considered during design.
8
Part I
4
3/28/2017
C-5A F-111
9
Part I
Comet 1 Aircraft
Was certified as safe-life for
16,000 flights and 10 years in
service
There were two crashes in 1954
due to operational pressurization
higher than that used in design at
that time (8.25 psi vs 5 psi)
Crashes of two Comet1 airplanes
in 1954 led to the adoption of
Fail-Safe design approach
10
Part I
5
3/28/2017
11
Part I
Advantages:
Able to manage the unexpected and mitigating damage if element
failure occurs
Disadvantages:
Over-design will lead to overweight structure
Difficult to accurately predict the failure modes of the structure
12
Part I
6
3/28/2017
13
Part I
14
Part I
7
3/28/2017
15
Part I
Inspectable: Non-inspectable:
Fail-Safe - multiple Safe-Life - with an
load paths or crack initial in-service
arrest mechanisms inspection.
or or
Slow Crack Growth - Slow Crack Growth -
crack will not grow to a cracks will not grow to
critical size between a critical size over the
inspections service life
16
Part I
8
3/28/2017
worse case
crack scenario
time of operation
First inspection at 1/2 the
worst case fatigue life
17
Part I
Advantages:
Healthy structures can be used indefinitely.
Ease of model adaptation for new crack scenarios that
may evolve in-service.
Disadvantages:
Cost of periodic inspections.
Design approach has traditionally been deterministic
and the life prediction estimate is dominated by:
Choice of initial crack characteristics
Assumed load spectrum.
Multi-site and multi-element damage is not considered.
LEFM does not accurately model small cracks.
18
Part I
9
3/28/2017
19
Part I
Introduction to Fracture
Mechanics Analysis
20
Part I
10
3/28/2017
21
Part I
22
Part I
11
3/28/2017
12
3/28/2017
u
r 2
n n n n
anII ( 1)n sin sin( 2) where = G
n 1 2 2 2 2 2 and 34n, plane stress
n
(3n)/(1n), plane strain
r 2 II n n n n
v an ( 1)n cos cos( 2)
n 1 2 2 2 2 2
25
Part I
y
y
Example of Expansion
Along Crack Line, x = r, Mode I x
x (r)
a1
x 4a2 3a3 r 8a4 r 5a5r 3 2
r
a
y 1 3a3 r 5a5r 3 2
r
First (leading), or singular term, a1: contains the stress
intensity factor
Second term, a2: contains the T-stress
Third term, a3: the leading higher order term (note: non-
polynomial!) 26
Part I
13
3/28/2017
K I lim yy 2 r
r 0
K II lim xy 2 r
r 0
K III lim yz 2 r
r 0
The so-called T-stress is the constant stress acting parallel to the crack
direction.
27
Part I
T-stress
3 T
cos K I cos 2 K II sin (1 cos 2 )
1
2r 2 2 2 2
3 T
cos K I 1 sin 2 K II sin 2 K II tan (1 cos 2 )
1
rr
2r 2 2 2 2 2
1 T
r cos KI sin KII (3cos 1) sin2
2 2r 2 2
28
Part I
14
3/28/2017
K III K III
xz sin (22) yz cos
( 2 r ) ( 2 r )
1 1
2
2 2
2
x y z xz 0
1
K III 2r 2
w sin
G 2
uv 0
n
for plane stress,let n
1 n
29
Part I
30
Part I
15
3/28/2017
Concept of K-Dominance:
When is LEFM Applicable?
y
K-Dominant KI
Region ys
yld
yield
2 x
y ys yns
x
rp
D
y ns 3a3 r 5a5 r 3 / 2 ...
Inelastic
Region,
Simplified
If rp << D, KI still controls fracture process.
31
Part I
32
Part I
16
3/28/2017
33
Part I
34
Part I
17
3/28/2017
36
Part I
18
3/28/2017
37
Part I
1 3
A quadratic element, (a) the parametric space of the element, (b) the Cartesian space of the
element. The crack tip is at r=0.
3
u N iui 12 ( 1)u1 (1 2 )u2 12 ( 1)u3
i 1
u u2 12 (u3 u1 ) ( 12 (u1 u3 ) u2 ) 2
38
Part I
19
3/28/2017
1 3
2
3
r N i ri l 12 l l ( 12 ) 2
i 1
Its standard, polynomial geometry interpolation scheme.
First, the usual case of mid-side geometry:
2r
1 1
2 l
Then we get the expected, polynomial interpolation:
2
u u1 ( 3u1 4u2 u3 ) r 2(u1 2u2 u3 ) r
l l2
39
Part I
1 3
2
3
r N i ri l 12 l l ( 12 ) 2
i 1
Its standard, polynomial geometry interpolation scheme.
20
3/28/2017
du
Normal strain field: e ( 3u1 4u2 u3 ) 1 4(u1 2u2 u3 ) r
dr l l2
du
-point strain field: e 2(u1 2u2 u3 ) 1 ( 23 u1 2u2 12 u3 ) 1
dr l lr
n
n 1 n n n n
Compare to theoretical field: x r 2 an1 2 ( 1)n cos( 1) ( 1) cos( 3)
2
n 1 2 2 2 2
41
Part I
Displacement field:
Contains a constant value, a linear variation in r, and
the square root variation in r.
Corresponds to the leading terms in the LEFM
expressions for the near crack-tip displacement.
Strain field:
Contains a constant term and a singular term that varies
as r-1/2
Corresponds to the lead term in the LEFM stress and
strain expansions
42
Part I
21
3/28/2017
43
Part I
44
Part I
22
3/28/2017
45
Part I
46
Part I
23
3/28/2017
47
Part I
48
Part I
24
3/28/2017
49
Part I
1/ 2
,v KI r
u cos 1 2n sin 2
2 2 2
1/ 2
KI r
v sin 2 2n cos 2
2 2 2
,u
Note: for plane stress, let n = n/(1+ n)
1/ 2
K II r
u sin 2 2n cos 2
2 2 2
1/ 2
K II r
v cos 1 2n sin 2
Set r = ra-b, and = 180
2 2 2
K I ra b K II ra b
vb va (2 2n ) ub ua (2 2n )
2 2
50
Part I
25
3/28/2017
2 (v b v a )
KI
ra b (2 2n )
2 (ub ua )
ra-b K II
ra b (2 2n )
(w b w a )
K III
2ra b
is the shear modulus, n is Poisson's ratio, r is the distance from the crack tip to the
correlation point, and ui, vi, wi are the x, y, and z displacements at point i
52
Part I
26
3/28/2017
2
K II 4(ub ud ) ue uc
r (2 2n )
53
Part I
54
Part I
27
3/28/2017
J-Integral
55
Part I
56
Part I
28
3/28/2017
57
Part I
Where (r,) are the near crack-front stresses (in cylindrical coordinates),
and u(r,) are the near crack-front displacements
* Irwin, G.R. (1957) Analysis of stresses and strains near the end of a crack traversing a plate, Journal of Applied Mechanics, 24,
58361-
364
Part I
29
3/28/2017
u
x 2(1+ n ) r
2 (
cos q 1- 2n + sin 2 q
2 ) ( )
sin q2 2 - 2n + cos2 q2 0
KI
(3)
uy =
E 2p
(
sin q2 2 - 2n - cos2 q2 ) cos ( 2n -1+ sin )
q
2
2q
2
0
K II
K
uz 2sin q2
III
0 0
* Williams, M.L. (1957) On the stress distribution at the base of a stationary crack, Journal of Applied Mechanics, 24, 109-114
59
Part I
s K u 2(1- n )K
yy x 2(1+ n ) r
1
I I
s xy = K
II and u
y = 2(1- n )K (4)
2p r E 2p II
s yz K III u
z K III
Substituting equations 4 into 1, and evaluating the integral gives the well
know result
1- n 2 2 1- n 2 2 1+ n 2 (5)
G= KI + K II + K III
E E E
30
3/28/2017
61
Part I
energy
a
contour
informally: 0 small
dEcontour dEelastic dE plastic dEnew surface dEheat dEsound dElight
da da da da da da da
31
3/28/2017
J is the energy release for the specified virtual crack extension q. This must
be normalized to get the energy released for a unit amount of new crack area
J J u q
G= = = s ij i -W d 1 j dV Aq (7)
q(s)ds Aq 1 x j
V x
This is an expression for the energy release rate in terms of stresses, strains,
and displacements 64
Part I
32
3/28/2017
J-Integral
The problem with J is that it only gives one
number for the fracture energy release rate
Difficult to segregate the fracture energy release rate
among the three modes of fracture.
65
Part I
M-Integral
66
Part I
33
3/28/2017
68
Part I
34
3/28/2017
Principal of Superposition
The principal of superposition for elasticity says that if you have one solution
that satisfies the governing equations for elasticity, and you add it to a second
solution that satisfies the governing, the result will also satisfy the governing
equations
Therefore,
(8)
K I = K I(1) + K I(2) , K II = K II(1) + K II(2) , K III = K III(1) + K III(2)
where the (1) solution will be the finite element results, and the (2) will be
an analytical solution
69
Part I
(10)
G = G(1) +G(2) + M (1,2)
where
70
Part I
35
3/28/2017
(12)
G = G(1) +G(2) + M (1,2)
Where in this case
71
Part I
(1)
Solving for The Finite Element K I
For the (2) solution, we use the first term of the Williams expansion, (equations 2
and 3) after setting KI = 1 and KII = KIII = 0. Substituting this into equation 11 and
13, and setting the two expressions for M equal to each other, gives
Where (1) and u(1) come from finite element results, and
(2) and u(2) are
72
Part I
36
3/28/2017
q q 3q
cos 1- sin sin
s xx 2 2 2
s yy cos q 1+ sin q sin 3q
s zz
2 2 2
1
= n (s xx + s yy )
s xy 2p r
q q 3q
(16)
s yz
sin cos cos
2 2 2
s zx
0
0
and
q q
cos 1- 2n + sin 2
u 2 2
x 2(1+ n ) r q 2q
(17)
uy = sin 2 - 2n - cos
E 2p 2 2
uz 0
For the (2) solution, we use the first term of the Williams expansion, (equations 2
and 3) after setting KI = 0, KII = 1 and KIII = 0. Substituting this into equation 11
and 13, and setting the two expressions for M equal to each other, the finite element
KII is
Where (1) and u(1) come from finite element results, and
(2) and u(2) are
74
Part I
37
3/28/2017
q q 3q
sin 2 + cos cos
s xx 2 2 2
s yy
q q 3q
sin cos cos
s zz 2 2 2
1
= n (s xx + s yy )
s xy 2p r (19)
s yz cos q 1- sin q sin 3q
2 2 2
s zx
0
0
and
q q
sin 2 - 2n + cos 2
u 2 2
x 2(1+ n ) r q
2q (20)
uy = cos -1+ 2n + sin
E 2p 2 2
uz 0
For the (2) solution, we use the first term of the Williams expansion, (equations 2
and 3) after setting KI = KII = 0 and KIII = 1. Substituting this into equation 11 and
13, and setting the two expressions for M equal to each other, the finite element KIII
is
Where (1) and u(1) come from finite element results, and
(2) and u(2) are
76
Part I
38
3/28/2017
0
s xx
s yy 0
0 u
s zz 1 0 x 2(1+ n ) r 0
= and u y = 0
s xy 2p r q
sin E 2p
q
(22)
s yz 2 uz sin
2
q
s zx cos
2
77
Part I
78
Part I
39
3/28/2017
T-Stress
79
Part I
Template Cross-Section
40
3/28/2017
FRANC3D
2H
2c ~2% difference max
2W T
Raju-Newman*
* note, the Raju-Newman solution is known to be
slightly inaccurate near the free surface.
81
Part I
Analyses performed by
CCT SEN
2 2
Dawn Phillips NASA
2h
3
1
2h
3
1 Langley Research Center
2a a
2w w
t t
S S
82
Part I
41