Академический Документы
Профессиональный Документы
Культура Документы
Submitted by-
Abstract:
Fatigue crack initiation in steel structures is one of the most important considerations
facing the infrastructure community. Purely static loading is rarely observed in structural
components. Almost 80% to 95% of all structural failures occur through a fatigue
mechanism. As a result, fatigue analysis has become an early driver in the product
development processes of a growing number of companies.
The traditional approach for determining the fatigue limit for a structure is to establish the
curves (load versus number of cycles to failure) for the materials in the structure. Such an
approach is still used as a design tool in many cases to predict fatigue resistance of
engineering structures although it is generally conservative and no relationship between the
crack length and the cycle number is available.
The computational process to simulate the slow progressive damage in steel over many
load cycles is simple but numerical fatigue life studies usually involve the response of the
structure subjected to a small fraction of the actual loading history. This response then
might be extrapolated over many load cycles using empirical formulae to predict the
likelihood of crack initiation and propagation.
The direct cyclic analysis capability in ABAQUS/Standard provides a computationally
effective modeling technique to obtain the stabilized response of a structure subjected to
periodic loading and is ideally suited to perform low-cycle fatigue calculations on a large
structure. The direct cyclic low-cycle fatigue capability is an extension of the direct cyclic
capability that includes damage accumulation and damage extrapolation. It provides
capabilities to model damage growth in steel structure. In material the cyclic loading leads
to stress reversals and the accumulation of plastic strains, which in turn cause the initiation
and propagation of cracks. The damage initiation and evolution are characterized by the
stabilized accumulated inelastic hysteresis strain energy per cycle. The objective of this
research report is to simulate the fatigue behaviour of structural steel with ABAQUS6.9-2.
Significant performance gains with good accuracy of the direct cyclic low cycle fatigue
capability are clearly demonstrated.
TABLE OF CONTENT
ABSTRACT...
TABLE OF CONTENT.
LIST OF FIGURE
1.0
FATIGUE ANALYSIS .
2.0
LOADING ..... 6
3.0
MATERIAL PROPERTIES... 7
4.0
5.0
6.0
FUTURE STUDY .
7.0
REFERENCES .. 31
29
LIST OF FIGURE
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9
Figure 10
Figure 11
Figure12
Figure 13
Figure 14
Cycling load . 25
Figure 15
Cycling load . 25
Figure 16
Cycling load . 26
Figure 17
Cycling load..
26
Figure 18
Meshing ...
27
Figure 19
27
Figure 20
Piezoelectric sensor .. 29
Figure 21
11
14
19
22
31
The method for calculating fatigue life is sometimes called the Five Box Trick, including
material, loading, and geometry inputs, and analysis and results. The three main inputs for
fatigue life analyses are processed using various life estimation tools depending on whether
the analysis is for crack initiation, total life, or crack growth.
2.0 LOADING
Pun (2001) demonstrated that the proper specification of loading variation is extremely
important to achieve an accurate fatigue life prediction. The loading can be defined in
various manners and whether it is time-based, frequency-based or in the form of some sort
of spectra depends on the type of fatigue analysis. When working with finite element
models the loading can be force, pressure, temperature, displacement, or a number of other
types. The time history used in a fatigue calculation must be a representation of the time
variation in the loading applied in the Finite Element Analysis (FEA). For simple cases,
this implies a force-time history corresponding to a time variation in the point loading used
in the FEA. There are a number of different kinds of loading possible, each one requiring a
different type of time history. For example, RAMP and STEP (Fig. 1) define how and
when the loading is applied during a given step. The following figure shows the difference
between the two.
= + K ...................................................................................................(1)
E
E
Where
is strain,
is stress,
E is Young's modulus, and
K and n are constants that depend on the material being considered.
second term, K , accounts for the plastic part, the parameters K and n describing the
E
hardening behaviour of the material. Introducing the yield strength of the material, 0, and
defining a new parameter, , related to K as = K O
E
term on the extreme right side as follows:
n
K = O
E
E
n1
...........................................................................................(2)
= + O
E
E O
................................................................................................(3)
and
can be seen as a yield offset, as shown in Fig. 2.. This comes from the fact
E
that = (1 + ) O / E , when = O .
can also be
found by means of fitting to experimental data, although for some materials, it can be fixed
in order to have the yield offset equal to the accepted value of strain of 0.2%, which means:
( O / E ) = 0.002
Because the computational cost to simulate the slow progress of damage in a material over
many load cycles is prohibitively expensive for all but the simplest models, numerical
fatigue life studies usually involve modeling the response of the structure subjected to a
small fraction of the actual loading history. This response is then extrapolated over many
load cycles using empirical formulae such as the Coffin-Manson relationship (Fig. 3) to
predict the likelihood of crack initiation and propagation. Since this approach is based on a
constant crack/damage growth rate, it may not realistically predict the evolution of the
crack or damage.
10
The solution at each of these points is used to predict the degradation and evolution of
material properties that will take place during the next increment of load cycles, N. The
degraded material properties are then used to compute the solution at the next point in the
load history. This capability can be used to model progressive damage and failure both in
the bulk material and at the material interface. When failure mechanisms both in the bulk
material and at the interfaces are considered simultaneously, the failure occurs first at the
weakest link in the model. The damage initiation and evolution in the bulk material are
characterized by the accumulated inelastic hysteresis strain energy per stabilized cycle, as
illustrated in Fig. 4.
The onset and fatigue delamination growth at the interfaces are characterized by the
relative fracture energy release rate by using the Paris law, as shown in Fig. 6.
12
For linear elastic materials, Griffithsapproach says that a crack extends if the
thermodynamic crack driving force, characterized by the energy release rate G (Fig. 6),
becomes equal or larger than the crack growth resistance, R (Griffith, 1921), whereas the
Irwin (1957) approach postulates that a crack grows when the crack tip stress intensity
factor K reaches a critical value Kc (Fig. 7). The Griffith and Irwin criteria are equivalent
for linear elastic materials, since energy release rate and stress intensity factor are related.
Crack tip conditions are defined by a single parameter, such as stress intensity factor.
Under cyclic constant amplitude stress intensity, the crack growth rate for small plastic
zones at the crack tip is defined by:
where,
K cycles between Kmax Kmin
R = Kmin / Kmax=Smin/Smax
da/dN = crack growth rate per cycle, L(cycle)-1
In field-situations, a history dependent factor is added into the function to account for
previous loading conditions during service-life of the element. The similitude assumption
shown for crack growth rate does not take into account occasional overloads and/or under
loads which will turn the problem into a variable amplitude loading configuration. Since
the stress intensity factor cannot characterize excessive plasticity at crack tip, researchers
proposed that crack growth be a function of J-integral. For fatigue resulting in large-scale
yielding, the J value will be employed and is analogous from monotonic loading.
Fatigue crack propagation curve (log da/dNlog K) may be divided into three stages
which are typical for: short crack growth propagation stage (Region I), long crack
propagation (Region II), and fracture stage (Region III) (Fig. 5). Where the behavior of the
13
linear segment in Region II, also known as the Paris regime, is enforced by the Paris power
law:
14
15
independent of the contour used to evaluate it, so Jtip = Jfar where Jfar is the integral on a
contour in the far-field. This path independence is important, since the energy released at
the crack tip (Jtip) cannot be easily measured, whereas the total energy released during
crack extension in a body (Jfar) can be readily measured.
In homogeneous elastic materials, Jtip is identical to the total energy released in the
specimen per unit crack extension, whereas this is not so in elasticplastic materials due to
the dissipation in the plastic zone which induces the plasticity influence term, Cp, defined
as the total configurationally force due to plasticity, projected on the crack growth
direction. For deformation plasticity, the plasticity influence term Cp vanishes in the
context of deformation plasticity, whereas the crack driving force Jtip vanishes for rigid
plasticity.
An experimental study on C(T) specimens to examine the influence of plastic deformation
near the crack tip was performed on annealed mild steel with a Youngs modulus of 200
GPa. Loading was simulated by prescribing the load line displacement ll assuming plane
strain conditions. The far-field J-integral values are compared to ABAQUS values
calculated using the virtual crack extension and to experimental values determined from the
area below the load vs. load-line displacement curve (Rice, 1973). Results showed that the
configurational body force arises only due to plastic deformation and is large directly at the
crack tip for homogeneous material. From the calculated plastic zone of 2 mm, the
configurational force rapidly decreases to zero at a distance less than 2 mm. Abaqus
showed that a plastic deformation starts at the back face of the specimen for ll=0.2 mm (a)
and as it increases to 0.343 mm (b) and 0.352 mm (c) show that with increasing loading,
both the crack tip plastic zone and the region of remote plasticity expand. The two regions
16
eventually merge and general yielding ocurrs in the ligament. J-integral Jtip is the scalar
driving force on a crack tip in elasticplastic materials.
After an extensive analysis performed on the behavior of the J-integral for different cases,
it was concluded that the scalar driving force at the crack tip of an elastic-plastic material is
defined as Jtip. The crack driving force, Jtip, is equal to the sum of the global driving force
Jfar and the plasticity influence term Cp, therefore Jtip = Jfar+Cp. It was also found that the Jintegral is path-independent in homogenous elastic bodies due to the fact that
configurational body forces disappear, but in elastic-plastic bodies these forces are present
and depend on the plastic strain gradient. This makes the J-integral path dependent in
incrementally plastic materials.
Courtin (2005) studied the crack opening displacement extrapolation method and the Jintegral approach applied to ABAQUS finite element models. The results obtained by these
various means on CT specimens and cracked round bars are in good agreement with those
found in the literature. The J-integral calculation relies as a good technique to deal with
since no knowledge of the crack-tip field is required. In the heart of the body, as one can
assume a plane strain state, the expression on the crack lips ( = ), where u is the angular
displacement, is Poissons ratio, and r is the radial distance from crack-tip is defined by:
18
energy release rate, G, representing the elastic energy per unit crack surface area required
to extend a crack.
(
where in plane strain, E = E/(1-2). Value of K depends on crack shape, displacement mode
and structure configuration with S being the nominal stress assuming no crack. Units for K
are MPa(m)1/2 and ksi(in)1/2. An arbitrary stress at a distance r from the crack tip and in
plane may be represented by Fig. 10.
19
It must be specified that the use of LEFM is restricted to small plastic zones at crack tip.
Stress intensity factors for different loading types inside the same mode may be added
together by using superposition, whereas K values for different modes (mode I, II, III)
cannot be added.
When comparing plastic zone sizes using the von Mises maximum shear stress criteria, it
may be seen that a larger plastic zone is present in plane stress conditions as seen in Fig.
11(a). This is due to the fact that in plane strain, the normal stress component, z, is zero
and restricts plastic flow. Irwin developed a correction for the plastic zone radius, ry, ahead
of the tip; he argued that plasticity at the tip forced the crack to behave longer than its true
length due to plasticity and redistribution of the stress fields.
During cyclic loading, a plastic zone size will be present and the local stress at the crack tip
will be less than that observed in monotonic loading. For a given cycle, monotonic loading
is referenced by the maximum load and the cyclic characteristic is observed when the load
reaches a minimum in a single cycle, as seen in Fig. 11(b). Compressive stresses inside the
plastic zone radius, ry, are compressive and will decrease to tensile at a distance away from
2ry.
For plane stress conditions, the plastic zone for R>0 is calculated as:
345
6
8 9
7. ,5
where Sy is the yield stress; for plane strain the plastic zone results in a third of the plane
stress size. It may be seen that stress intensity factor governs crack growth and plasticity
zone and the LEFM methods may be applied because ry is much smaller than for
monotonic loading. A fracture toughness criterion, KIc is defined as the K value that for a
given load, crack length, and geometry will cause fracture at the last cycle of fatigue
without an increase in applied load. It is dependent on material, environment metallurgical
conditions, grain orientation and thickness. Therefore, low impurity metals will provide
better KIc behavior. A decrease in temperature will increase the yield strength and reduce
the KIc, wherease an increase in strain rate will inhance crack sensitivity and also reduce
fracture toughness. For it to be considered a true material property, the minimum KIc is
stated as it reaches an asymptotic minimum value, where an increase in thickness will not
affect the KIc, known as plane strain fracture toughness. Low strength ductile materials will
be subjected to plane strain only if they are very thick.
Elastic-plastic fracture mechanics (EPFM) describe the phenomenon of limited plasticity at
the crack-tip by assuming a slightly enlarged crack, from a to a+ry. The characterizing
parameter in EPFM is known as the J-integral and is a line integral related to energy
balance to describe a cracked component. By defining the J-integral as path independent it
may me implied that using contour I or II will yield the same results, allowing the use of
a contour remote from the crack which contain only elastic loads/displacements (Fig.12).
The elastic and plastic components of the integral are added to form the elastic-plastic Jintegral; J = Je+Jp, with the linear elastic condition, Je = G = K2/E for plane stress.
21
Figure 12. Line contour surrounding a crack tip for J-integral formulation.
A series of parameters have been investigated to relate crack growth under mixed mode
loading conditions. Experimental studies show that cracks with a small plastic zone will be
dictated by the alternating stresses pulling the crack surfaces apart; whereas in high
plasticity materials, crack growth have shown to be dependent also of the stresses parallel
to the crack plane. Energy release rates for planar crack under plane stress for mixed
loading:
Tanaka proposed an equivalent stress intensity factor range, Kq, that when combined with
the above equation yields:
22
210,000
0.3
Plastic
Plastic strain, p
0
246
294
0.0235
0.0474
374
0.0935
437
480
0.1377
0.18
23
24
CT1
120
Load, P(kN)
100
80
60
40
20
0
t(sec)
Figure 14. Cycling load
CT2
120
Load, P(kN)
100
80
60
40
20
0
0.5
1.5
t(sec)
Figure 15. Cycling load
25
2.5
CT3
60
Load, P(kN)
50
40
30
20
10
0
t(sec)
Figure 16. Cycling load
CT4
60
Load, P(kN)
50
40
30
20
10
0
0.5
1.5
t(sec)
Figure 17. Cycling load
26
2.5
Step(14): Jobs
---Create job
---Submit jobs
---Monitor
Step(15): Result
Open odb filethen result
28
both sensors and actuators. As sensors, they produce an electrical signal when they are
physically deformed (strained). As actuators, they physically deform (expand, contract, or
shear) when an electrical charge is applied. Using this property, piezoelectric materials can
be used to measure stress and strain and can also be used to mechanically excite the
structure to propagate stress waves and induce internal vibrations. Inputting a time-varying
electrical signal to any of the actuators/sensors causes a propagating stress wave or
propagating mechanical deformation to emanate from the sensor/actuator and travel
through the material for detection by a plurality of neighboring sensors/actuators.
The piezoelectric materials can be used in two sensing modes active and passive (Fig.
21). In Active Sensing Mode, the actuators are externally excited to generate pre-selected
diagnostic signals and transmit them to neighboring sensors whose response can then be
interpreted in terms of damage location and size or material property changes within the
structure. In Passive Sensing Mode the piezoelectric sensors can be used as continuously
monitored sensors that listen for external impact events.
Figure 21. Active and Passive Sensing modes used by piezoelectric materials (Bear et al.,
2005)
30
7.0 REFERENCES
1) ABAQUS 6.9 Users Manual Low-cycle fatigue analysis using the direct cyclic
31
10) Landes, J. and Begley, J. The effect of specimen geometry on JIc, Fracture
Toughness, Proceedings of the 1971 National Symposium on Fracture Mechanics, Part
II, ASTM STP 514, American Society for Testing and Materials, 1972, pp. 24-39
11) Paris, P. et al. A rational analytic theory of fatigue, The trend in engineering, 1961,
Vol 13, pp. 9-14.
12) Pun, A. Three methods of calculating total life, crack initiation, and crack growth,
Senior Product Manager, MSC.Software Corp. -- Design News, December 16, 2001
13) Ramberg, W., & Osgood, W. R. (1943). Description of stress-strain curves by three
parameters. Technical Note No. 902, National Advisory Committee For Aeronautics,
Washington DC.
14) Simha, N., et al. (2008), J-integral and crack driving force in elasticplastic materials.
Journal of the Mechanics and Physics of Solids, Vol 56, pp 2876-2895.
15) Stephens, R. et al. (2001). Metal Fatigue in Engineering 2nd Edition. John Wiley and
Sons. New York.
32