Materials of Engineering Department, IPEN, University of Sao Paulo, Avenida Lineu Prestes 2242 , 05508-
900, Sao Paulo – SP, Brazil
lricardo@ipen.br & aandrade@ipen.br
ABSTRACT
This paper presents a finite element model used to simulate the crack opening and closure stress intensity
factors. Three types of specimens (SE(B), SE(T) and M(T)) were modeled by finite element method under
constant amplitude loading to obtain the opening stress intensity factor and the results are compared with
FASTRAN data. A C(T) specimen also was modeled using a finite element code to determine the crack
closure stress intensity factor and the values found are correlated with crack propagation test results. The data
show that the finite element model can be used to reproduce the crack propagation processes.
88
the predicted crack closure results is significant under plane stress conditions. In experiments of
when a traditional plasticity model is used, Jiang crack closure with constant amplitude loading
et at.[9]. Elber´s [2] proposed the following equation for
fatigue crack propagation rates,
Jing et al.[9] shows a CCT specimen submitted a
load ratio, R= 0 and maximum nominal applied ∆a
stress, Smax ,of 160 MPa, was used to show the = C (∆K eff ) n (2)
effect of the element size on the predicted crack ∆N
opening load. For the CCT specimen ,the
corresponding stress intensity factor range, where C and n are constants of the material and
∆K = 0 , was 28.7 MPa m .The cyclic plastic ∆Keff is the effective stress intensity factor range,
which can be calculated by :
zone size ,rp,, based on the stress intensity factor
range can be determined by using the following
equation for plane-stress condition (Bannantine et ∆K eff = ∆Seff πc F (3)
al. [10]):
2
where c is the half crack length, F is the boundary
1 ⎛ ∆K ⎞ correction factor and ∆Seff is the effective stress
rp = ⎜⎜ ⎟⎟ , (1) range.
8π ⎝ σ0 ⎠
Figure 1 shows a panel with a central crack used
by Newman [11]. The aluminum alloy panel
2 NUMERICAL ANALYSIS material was assumed to be elastic perfect plastic
with a tensile ( and compressive ) yield stress, σy,
Since the early 1970s, numerous finite element of 350 MPa and modulus of elasticity of
and finite difference analysis have been conducted 70,000 MPa.
to simulate fatigue crack growth. The aim of these
analysis were to obtain a basic understanding of
the crack growth processes.
89
permits to evaluate crack closure and crack
propagation analysis until failure.
90
Figure 4 shows the variations of the normalized 3 DESCRIPTION OF FEM MODELS
crack opening load with element size using
bilinear material [9]. In the figure, the symbol r Figures 6-10 shows the finite element models,
represents the crack extension and rp denotes the generated using Ansys version 6.0 [16], from the
plastic zone size (rp =0.178 mm for the loading specimens SE(B) (single edge crack plate
case under consideration). Sop is the opening load bending), SE(T) (single edge crack plate tension)
(nominal stress) and Smax is the maximum nominal and M(T) (middle tension) were evaluated [16].
applied load. It should be noted that the element Fatigue crack growth was simulated by releasing
size refers to those elements near the crack tip. of the crack tip node at Kmin, followed by a single
loading cycle Kmin → Kmax → Kmin, Figure 11.
The boundary correction factors used for the
mentioned specimens are :
Figure 9: M(T) FEM model Figure 10: M(T) Release node procedure
91
a) SE(B)
⎛ 2 3 4
⎞
⎜ 1,124 − 1,16349 * ⎛⎜ a ⎞⎟ + 7,3168 * ⎛⎜ a ⎞⎟ − 18,7746 * ⎛⎜ a ⎞⎟ − 31,8028 * ⎛⎜ a ⎞⎟ − ⎟
⎜ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎟
⎜ 5 6 7 ⎟
⎜ ⎛ a ⎞ ⎛ a ⎞ ⎛ a ⎞ ⎟
− 33, 2295 * ⎜ ⎟ + 19 ,1286 * ⎜ ⎟ − 4 , 6091 * ⎜ ⎟
⎛ a ⎞ ⎜ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎟
f⎜ ⎟=⎜ ⎟ (4)
⎝ ⎠ ⎜ ⎟
3
W ⎛ ⎛ a ⎞⎞ 2
⎜ ⎜⎜1 − ⎜ W ⎟ ⎟⎟ ⎟
⎜ ⎝ ⎝ ⎠ ⎠ ⎟
⎜ ⎟
⎜⎜ ⎟⎟
⎝ ⎠
b) SE(T)
⎛ 2 3 4
⎞
⎜ 1,124 − 1,16349 * ⎛⎜ a ⎞⎟ + 7,3168 * ⎛⎜ a ⎞⎟ − 18,7746 * ⎛⎜ a ⎞⎟ − 31,8028 * ⎛⎜ a ⎞⎟ − ⎟
⎜ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎟
⎜ 5 6 7 ⎟
⎜ ⎛ ⎞
a ⎛ ⎞
a ⎛ ⎞
a ⎟
− 33, 2295 * ⎜ ⎟ + 19 ,1286 * ⎜ ⎟ − 4 , 6091 * ⎜ ⎟
⎛a⎞ ⎜ ⎝ W⎠ ⎝ W⎠ ⎝ W⎠ ⎟
f⎜ ⎟=⎜ ⎟ (5)
⎝W ⎠ ⎜
3
⎛ ⎛ a ⎞⎞ 2 ⎟
⎜ ⎜
⎜ 1 − ⎜ ⎟ ⎟
⎟ ⎟
⎜ ⎝ ⎝ W ⎠ ⎠ ⎟
⎜ ⎟
⎜ ⎟
⎝ ⎠
c) M(T)
⎛ 2 3 4
⎛a⎞ ⎞
5
⎜ 1,0 − 0,4999 * ⎛⎜ a ⎞ ⎛a⎞ ⎛a⎞ ⎛a⎞
⎟ + 0,4593 * ⎜ ⎟ − 0,3017 * ⎜ ⎟ + 0,2051 * ⎜ ⎟ − 0,0371 * ⎜ ⎟ ⎟
⎛a⎞ ⎜ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎟
f⎜ ⎟=⎜ ⎟ (6)
⎝ ⎠ ⎜
W ⎛a⎞ ⎟
⎜ 1− ⎜ ⎟ ⎟
⎝ ⎝W ⎠ ⎠
92
mentioned in (7) and the boundary correction KBW (7)
factor (8) : Pmax =
⎛a⎞
πa f ⎜ ⎟
⎝W ⎠
⎛ a ⎞
⎜2+ ⎟ ⎡ 3 4
⎤
⎛a⎞ ⎜ W ⎟ * ⎢0,866 + 4,64 * ⎛⎜ a ⎞⎟ − 13,32 * ⎛⎜ a ⎞⎟ + 14,72 * ⎛⎜ a ⎞⎟ − 5,60 * ⎛⎜ a ⎞⎟ ⎥
f⎜ ⎟= (8)
⎝W ⎠ ⎜ 1 − a ⎟ ⎢⎣ ⎝W ⎠ ⎝W ⎠ ⎝W ⎠ ⎝ W ⎠ ⎥⎦
⎜ ⎟
⎝ W ⎠
Figure 12: C(T) FEM model Figure 13: C(T) Release node procedure
The force was divided into 9 steps of loads Pmax constant ∆P= Pmax - Pmin, under R = 0
and 9 steps of loads to Pmin, in each cycle. The and R = 0.5 load ratios. To evaluate the crack
applied loads were increased with crack length in propagation a nonlinear analysis to compute the
order to ensure that the specimen suffered a deformation history cycle by cycle was applied
93
and the Newton-Rapson method was used. Table Between any two increments, if the nodal
2 show an example of the loads steps used to displacement becomes negative the node is closed
compute the opening and crack closure stress. and a node fixity is applied to prevent crack
During each increment of unloading, the crack surface penetration during further unloading.
surface nodal displacements are monitored.
This criterion is a function of the number of During incremental loading, the reaction forces on
increments employed to describe the unloading. It the closed nodes are monitored, and when the
has been observed that an increment on the order reaction force becomes positive the nodal fixity is
of 1% of the maximum stress in the cycle or removed. During loading and unloading, stresses
smaller results in a negligible variation of the and displacements were monitored along the crack
contact forces along the closed crack surface surface. A negative nodal displacement indicated
94
that the crack was closed at this point, and the
displacement was set to zero. A tensile nodal
stress indicated that the crack was opening at this
point, and the nodal restraint was removed,
Solanki et al.[ 18 ].
4- EXPERIMENTAL PROCEDURE
95
The software from MTS guarantee the
convergence for a Pk according with minimum e 5- RESULTS AND DISCUSSION
computed. Figure 16 shows the crack closure
load. After the determination of crack closure load
the software relate the crack length, material The procedure to determine the crack opening
properties and the width of specimen and crack and closure stress intensity factor is based on Wei
length of the specimen to compute the crack & James [23] that consider the displacement at
closure stress intensity factor. first node behind crack become positive and
perpendicular stress become positive too. To
2 determine crack closure the displacement become
{v ( Pi ) − vi } , Pi < Pk ;
N
1
(v max − vmin )2 ∑
e= q
negative and perpendicular stress in the first node
i =1 behind crack become negative too.
96
Figure 18: Normalization of stress intensity factors SE(B), R = 0.5
97
Figure 21: Normalization of stress intensity factors M(T), R = 0
98
Figure 25 : Post-processing SE(T), R= 0 Figure 26 : Post-processing detail SE(T)
Figure 29: Post-processing C(T), R = 0.1 Figure 30: Post-processing C(T) detail R = 0.1
Figure 31 an experimental result of crack possible see the Paris region. Figure 32 show the
propagation test in a C(T) specimen where is correlation between numerical and experimental
99
results by normalization of Kmax and Kcl with showing a increase of the experimental plastic
crack length and width of specimen, a/W. The zone much more than the numerical one. Is fact
points shown in the graphics are in the same that the crack propagation rate 5.0E-5 m/cycle, do
coordinates in numerical and experimental not represent the real crack propagation rate when
models. The graphic show good correlation compared with the numerical rate used in this
between numerical and experimental results in the work. To elaborate models with real crack
a/W ratio of 0.32 to 0.40. Above 0.40 the propagation rates is necessary spend a lot of time
numerical and experimental results has processing that have a duration bigger than the
divergence. The divergence is related to the test.
plastic zone formed during crack propagation,
100
In the compact tension C(T) simulation were [3] Swedlow, J. L. The Effect and Plastic Flow
used10 increments to loading and 20 to unloading in Cracked Plates. PhD Thesis. California
with objective to increase the precision in the Institute of Technology. Pasadena. USA.
determination of the crack closure stress. To 1965
SE(B), SE(T) and M(T), were tested used 9 [4] Dugdale, D.S. Yielding of Steel Sheets
loading and unloading increments to determine Containing Slits; J. Mech. Phys. Solids. pp.
the stress opening. The work identified that 9 100-104. nº 8. 1960
increments to loading and unloading are best [5] Dahlberg, M. & Stenfors,S.E. Examination
agreement between time processing and quality of of The Fracture Mechanics Approach to
results. Fatigue Design of Nodular Cast Iron. In:
Blom. Fatigue 2002. Vol. 4/5. West Lands.
7 CONCLUSIONS Emas. pp. 2393-2400
[6] Taylor, D; Ciepalowicz, A.J.; Rogers, P. &
The present work presented a methodology to Develukia, J. Prediction of fatigue Failure in
simulate crack propagation processes as well as a Crankshaft Using The Technique Crack
the crack opening and closure mechanisms using Modeling. Fatigue & Fracture of
finite element method. Three kind of aluminum Engineering Materials & Structures. Vol.
alloy specimen with Al-2024-T351, SE(B), SE(T) 20. N.º 1. pp. 13-21. 1997
and M(T) were tested, comparing results with [7] Godefroid, L.B.; Guimaraes, C.R. & Silva,
FASTRAN and Ansys. The C(T) numerical A.P.F.S. Fatigue Crack Propagation and
results were compared with experimental data. Closure Behavior of a Dual-Phase
Constant amplitude loading was used to conduct Steel.COBEM. SP. Brazil. 2003
the numerical and experimental models. [8] Roychowdhury, S. & Dodds Jr, Robert H. A
numerical investigation of 3-D small-scale
It is possible to determine the stress, yielding fatigue crack growth, Eng. Frac.
displacements and later crack opening and closure Mechanics. Vol. 70., pp. 2363-2383, 2003
stress intensity factors covering all load history . [9] Jiang, Y; Feng, M.& Ding, F., A
To avoid chance to have convergence problems Reexamination of Plasticity-Induced Crack
release node in the minimum load was used. Closure in Fatigue Crack Propagation , Int.
Journal of Plasticity, Vol,. 21, 1720-
The methodology to simulate crack opening and 1740,2005
closure mechanisms by numerical or analytical [10] Newman, J.C. Jr. Finite Element Analysis
procedures is used with success in the aeronautic of Fatigue Crack Propagation Including The
and aerospace industry. The paper shows that with Effects of Crack Closure. PhD Thesis. VPI
few a cycles of crack propagation it is possible & SU. Blacksburg. USA. 1974
have characterization in terms qualitative of the [11] Newman, J.C. Jr.; Armen, H. Jr. “ Elastic-
crack opening and closure stress intensity factors Plastic Analysis of Fatigue Crack Under
using finite element method. Cyclic Loading”. AIAA Journal. nº 13. pp.
1017-1023. 1975
The methodology also can support automotive [12] Newman, J.C. Jr. A Finite Element Analysis
application supporting design criteria in of Fatigue Crack Closure. ASTM 490. pp.
characterization of properties in materials that 281-301. 1976
can be used in suspension and chassis [13] Fleck, N.A. & Newman, J.C. Jr. Analysis of
components like suspension components to Crack Closure Under Plane Strain
predict crack opening stress. Conditions. ASTM STP 982. pp. 319-341.
USA. 1988
8 REFERENCES [14] Newman, J.C. Jr. FASTRAN II – A Fatigue
Crack Growth Structural Analysis Program.
NASA TM-10459. 1992
[1] Christensen, R. H. Fatigue Crack Growth
[15] Bannantine,J.A.,Comer,J.J.,Handrock,J.L.,1
Affected by Metal Fragments Wedged
990.Fundamentals of Metal Fatigue
Between Opening Closing Crack Surface.
Analysis.Prentice-Hall,Englewood
Appl. Mater. Res.. nº 2.October. pp. 207-
Cliffis,NJ.
210. 1963
[16] Ansys Inc. Ansys Version 6.0. USA. 2002
[2] Elber, W. Fatigue Crack Propagation. PhD
[17] Liu, J.Z. & Wu, X. R. Study on Fatigue
Thesis. University of New South Wales.
Crack Closure Behavior For Various
Australia. 1968
101
Cracked Geometries. Eng. Frac. Mechanics. closure measurement and analysis. In:
Vol. 57. Nº5. pp. 475-491. 1997. McClung, R.C., Newman, Jr., J.C., (Eds.),
[18] Solanki, K.; Daniewicz, S.R & Newman Jr, Second volume, ASTM STP,vol.1343,
J.C., Finite element analysis of plasticity American Society for Testing and
induced fatigue crack:an overview; Eng. Materials,West Conshohocken,PA,pp.94 –
105, 1999
Frac. Mechanics, Vol. 71, 149 – 171, 2004
[22] Pippan,R.,Riemelmoser,F.O.,Bichler,C.,199
[19] ASTM. Standard Test Method for
9.Measurability of crack closure. Advances
Measurement of Fatigue Crack Growth
in fatigue crack closure measurement and
Rates. E647 –95a. 1995
analysis. In: McClung, R.C., Newman,Jr.,
[20] Donald, J.K., Phlilips, E. P., Analysis of the
J.C. (Eds.),Second volume, ASTM
second ASTM round-robin program on
STP,vol.1343.American Society for Testing
opening-load measurement using the
and Materials,West
adjusted compliance ratio technique.
Conshohocken,PA,pp.41 – 56.
Advances in fatigue crack closure
[23] Wei, L.W. & James, M.N., A Study of
measurement and analysis. In: McClung,
Fatigue Crack Closure in Polycarbonate
R.C., Newman, Jr., J.C.(Eds.), Second
C(T) Specimens. Eng. Frac. Mechanics.,
volume, ASTM STP, vol.1343.American
Vol. 66. pp. 223-242, 2000
Society for Testing and Materials, West
Conshohocken,PA,pp.79-93, 1999.
[21] Graham, S.M., Tregoning, R., & Zhang,
X.J., Evaluation of the adjusted compliance
ratio technique for measuring crack closure
in Ti -6Al -4V.Advances in fatigue crack
102