You are on page 1of 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/267792760

NUMERICAL EVALUATION OF DAMAGE


DISTRIBUTION OVER A SLAT TRACK USING
FLIGHT TEST DATA

Article

CITATIONS READS

4 42

5 authors, including:

Antonio Vecchio Patrick Guillaume


Clean Sky Joint Undertaking Vrije Universiteit Brussel
64 PUBLICATIONS 320 CITATIONS 382 PUBLICATIONS 5,584 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

OWI-Lab: Offshore Wind Infrastructure Application Lab View project

Friendcopter View project

All content following this page was uploaded by Patrick Guillaume on 13 February 2015.

The user has requested enhancement of the downloaded file.


NUMERICAL EVALUATION OF DAMAGE DISTRIBUTION OVER A SLAT
TRACK USING FLIGHT TEST DATA.

Antonio Vecchio, Reinhold Carmine, Renaud de Voghel


LMS International, Interleuvenlaan 68, 3001 Heverlee, Belgium, e-mail: antonio.vecchio@lms.be

Geert Van der Linden, ASCO Industries – Belgium,

Patrick Guillaume, Vrije Universiteit Brussel - Brussel, Belgium.

ABSTRACT The results obtained show that there is high potential for
design optimisation of the slat track, since local stress
In aerospace industry, safety is a crucial element leading amplitudes are far below the endurance limit of the material.
aircraft makers to accurately design critical structural
components. Most of the aircraft components are subject to
periodic maintenance consisting of visual inspections aimed 1 INTRODUCTION
at assessing structural health of components and working
safety of the whole aircraft. If damage occurs in a In this work, the results of a numerical method aimed at
component, the visual inspection will allow detecting a crack evaluating the damage distribution over an aircraft
and starting the required maintenance action. component under real-life loads are presented. The
numerical procedure is based on LMS FALANCS software
In the quest for more advanced methods for damage and allows achieving fatigue life prediction under flight test
tolerance, aerospace industry has introduced the safe-life conditions.
and the fail-safe design approaches. Safe life analysis allows
verifying if a structure containing initial damage has sufficient The test article is a slat-track of an Airbus A320. In the frame
safe crack growth life when subject to the expected loads. of normally scheduled qualification test campaign of the
Fail-safe analysis allows verifying that the structure has an aircraft, the component under study was instrumented with
adequate load path redundancy. If the component strain-gages such as to continuously measure superficial
undergoing the highest load fails, there exist other load stresses during typical flight operations. These include
paths with sufficient strength and fatigue life that can safely several manoeuvring cycles (taxing, take-off, climb, cruise,
resist to the applied loads. descent and landing). The purpose is to identify and extract
from the acquired time histories the time segments
In order to perform the safe-life and fail-safe analyses load representing the most critical load conditions for the
and stress-spectra must be identified. This is usually component under analysis.
achieved by measuring the structural stress resulting from
static loads applied to the structure. However, under real An accurate pre-processing is required because of the
flight conditions non-proportional forces are continuously peculiarity of the test article. Slat-tracks are moving quite
generated that act on safety critical components and affect intensively during critical manoeuvres, thus the acting loads
time dependent local stress tensors at any location of the and the point were loads are applied change all the time.
surface of the component. Time histories are accurately pre-processed also by means
of comparative analysis with other measurement channels
In this work, dynamic measurements are used to numerically such as slat setting angle, flight altitude, aircraft pitch angle
evaluate damage distribution at the surface of the slat track and some vibration measurements both on the slat and the
of an Airbus A320. A large amount of in-flight data recorded wing.
during varying flight conditions is processed to identify the Once the most significant time segments with respect to load
most critical load conditions for the component under study. conditions are identified and selected, fatigue life prediction
can be performed.
Taking into account the material property, a numerical
simulation is then performed that allows identifying crack The numerical simulation adopted, requires as input the
initiation locations. In these locations, local stress tensor dynamic loads acting on the component, the geometry of the
histories are computed and prepared for further slat track - as a finite element model - and the cyclic material
investigations of crack propagation. data for the maraging steel.
Output is the damage distribution over the surface of the slat
track. This allows identifying hot spots on the structure in
terms of fatigue and paves the way to further investigations The problem of how to transfer the aerodynamic loads acting
on design optimisation. on the slat to the dynamic loads acting on the slat track as
well as how to transfer measured loads form one location to
Finally, the fatigue software LMS FALANCS delivers as any other can be addressed by numerical simulation. In
output the local stress time histories for crack propagation particular, a finite element analysis is applied that provides
simulations. the transition factors to transfer dynamic loads form the
location where loads were measured to any location of
Load Time Geometry: Finite interest by use of unit load cases solutions. The cyclic
Materials Data
Histories Element Model materials data for the maraging steel of the slat track also
has to be given as input.

The damage calculation is based on the accumulation of the


basic damaging events – the hysteresis loops – for each
Damage
Damage crack opening stress time-history in the critical planes
Calculation
Calculation evaluated by stress tensor transformation in the surface
plane.

Fatigue
Fatigue 2.1 Evaluation of Local Tensor Time Series Stress
Life
Life
For multi-axial loading, the crack initiation locations in a
structure cannot be determined without a fatigue analysis.
Figure 1: Data flow of the strain-life approach
Thus, many potentially critical locations have to be analysed.
Starting point for these analyses is the local stress tensor
2 METHOD OF FATIGUE EVALUATION
e
σ ij (t ) as a function of time, computed according to the
theory of elasticity (pseudo stress). For n internal forces
A full description of the fatigue evaluation methods is out of acting on the component, the history of the local pseudo
the scope of this article; however, in this section a short and stress tensor eσ kl (t ) at a given location can be determined
schematic description of the numerical method adopted in
the present work is provided. For an historical review and a by superposition of the local pseudo stresses resulting from
more detailed exposition on fatigue methods, the reader is n different unit load cases:
n
referred to [1]-[7].
σ kl ( x, t ) = ∑ ckl( i ) ( x) Li (t )
e
(1)
i =1
Fatigue relates to the degradation of a material due to (i)
Here, the stress influence coefficient ckl (x) is the stress
repeated cyclic loading. For metals, this typically means the
tensor at the x resulting from unit load case i. Li(t) is the i-th
initiation of small cracks from active slip bands in grains on
external force as a function of time. For each load history,
the free surface of a component or structure. Micro cracks
one matching unit load case has to be calculated by a linear
eventually link to form large cracks that either break or
static finite element analysis.
severely degrade the performance of the structure.

In most cases, the time that it takes to initiate cracks is


2.2 Damage Evaluation - Strain Life Approach
typically much longer than the time needed for the cracks to
link and damage the component. This means that the useful
life of a component can be nearly the same as the time Two basic approaches have been developed to estimate the
spent to initiate cracks. crack initiation life of components: the stress-life approach
and the strain-life approach to fatigue analysis.
The numerical fatigue evaluation used in this work can be
divided into three major processing steps. These are (fig.1): The strain-life approach was developed in the early 1960’s
as an alternative to the stress-life approach for determining
1. Determination of elastic dynamic stress tensor time fatigue damage, particularly when yielding occurs at critical
history by quasi-static or modal superposition; locations of a component. The basic idea of the strain-life
approach is that on most useful engineering components,
2. Evaluation of local potential crack opening stress time locations of stress concentration, or notches exist at which
histories with the critical plane approach; the local stress state may be above the yield stress of the
material, even if most of the surrounding material remains
3. Damage accumulation with strain-life or stress-life elastic.
approach
The fact that the location where yielding has occurred is
In order to provide the input for the multi-axial fatigue surrounded by material still in elastic state, restrict the
calculation, time histories of interface loads and outer loads deformation of the yielded material. This results in an elastic-
have to be measured or calculated. This can be achieved by plastic deformation mechanism at the notch location. The
multi body simulation or simply by measuring real-life loads notches, in turn, serve as critical locations for fatigue cracks
with strain gages placed on the component’s surface. to initiate because of the higher stresses and strains at these
locations. Therefore, the fatigue behaviour of a material can
be related to strain-controlled material property tests.
2.3 Damage Evaluation – Critical plane approach an appropriate damage parameter. When calculations for all
the planes are completed, the plane with the highest
Applied loads acting on metallic materials generate by calculated damage is then identified as the critical plane, and
nature a multi-axial stress state. An accurate damage it allows evaluating the fatigue life of the structure at the
evaluation should assess the stress state in each of those specified location.
directions such as to determine the most critical condition.
However, in order to make damage evaluation faster and
easier, several multi-axial fatigue criteria have been 2.4 Damage Evaluation, Iterative Evaluation process
introduced in literature that reduce a given multi-axial stress
state to an equivalent single-axial stress condition. A more The high number of finite elements required for an accurate
detailed description of those criteria is given in [1] and [2] stress analysis poses a number of challenges to fatigue
analysis. In particular, processing speed is an significant
In general, fatigue criteria can be classified according to the concern but even more challenging is to achieve good
physical hypothesis adopted, therefore names such as reliability in the result delivered. Good methods should not
stress invariant-based criteria, energy-based criteria, or force users to ‘guess’ critical locations before the actual
stress averages based criteria, are to some extent self- analysis has been completed. Otherwise, the risk is high to
explaining. lead users to make potentially wrong decisions.

In this work, we refer to the critical plane approach, so-called To accelerate numerical fatigue analysis of complete
because the fatigue failure assessment is carried out in the components and subsystems, LMS Durability Technologies
plane where some stress components or a combination of has developed a special approach, which was used and
them attains the highest value [8]-[11]. experimentally checked in a number of research projects.
A possible way to correlate the position of the critical plane The proposed approach consists of three steps: Rainflow
with the principal stress directions is given in [12]-[15], where projector, Rainflow projection filtering and node elimination.
ad-hoc weighting functions are introduced.
The basic idea is to eliminate early in the processing, finite
The critical plane approach was first presented by Findley [8] element nodes that are only slightly damaged. The rainflow
who observed that for fatigue crack initiation at a given projector picks only those few elements out of a large finite
location, damage accumulates independently in different element mesh that best represent the possible interactions
directions (planes) in the material. For multi-axial stress of the loads acting on the component. The original load
states, damage tends to accumulate more on certain planes histories are then applied to the selected best representative
than on other planes. This “preferred” planes for damage elements and this allows performing a quicker analysis of
accumulation are labelled as critical planes. damage as a function of time.

Several practical methods are available that implement the Time segments where little damage is accumulated for each
critical plane approach. However, the basic steps that all of the representative elements are eliminated from the
methods follow are quite similar and are resumed below. original loading history and the maximum amount of damage
The main difference concerns the way each method eliminated at any of the representative elements is stored.
determines the amount of damage on each plane, as this
depends on the choice of a number of damage parameter. In The shortened load history is then applied to all elements in
the LMS FALANCS software, used for this work, two critical the finite element structure. This allows identifying and
plane damage parameters are available: the crack opening eliminating less damaged elements. The remaining elements
parameter and the shear parameter. The former uses normal are finally analysed with the original loading history.
stresses, the latter uses shear stresses.
This automated, conservative approach strongly speeds up
In order to identify the critical plane, multi-axial stress and the analysis. It has proven not to miss any critical location
strain time histories at a particular location must be known or and produces exactly the same numerical result as the full-
estimated. These can be either measured experimentally or unfiltered analysis procedure.
obtained by numerical simulation.

In a second step, the multi-axial stress state is resolved for a 3 INPUT FOR FATIGUE EVALUATION
number of particular planes (i.e. orientations) in the material.
This results in computing normal stresses and strains and
shear stresses and strains acting on each selected plane. 3.1 Finite Element Model
Although a given point belongs to an infinite number of
planes, candidate critical planes are commonly picked up Using a Finite Element Modeller, a highly accurate model the
from a subset of planes spaced by ten degrees. It has been slat track has been created. The FE core model consists of
indeed observed that fatigue damage does not vary 236630 tetrahedral elements generated by using the
significantly in planes spaced less than ten degrees apart. automated mesh generator. On the top of that, beam
The number of candidate planes can be further reduced in elements have been added to correctly represent the rack in
case of simplified stress state (damage calculated on a the inner section of the slat track profile and to model the rod
traction-free surface, 1D or 2D stress state). assembly connected by ball bearings at the side to the
leading edge.
Fatigue damage estimate for the resolved stress and strain
histories is then computed based on a rainflow counting and
3.2 Boundary Conditions for Unit Load Cases The unit load solutions for T=1 N, N=1 and M=1 (where N
and M are expressed in Nmm) have been calculated for five
In-flight measured data were used for this application case. setting angles aSS respectively equal to 0°, 2°, 4°, 6° and
The aircraft was instrumented as for the normal qualification 20°. Noticing that for aSS >7° all rollers are out of the cross
test campaign where several different measurements are section of MB, it can be assumed that the transition matrix
usually acquired in a large variety of different flight between measured loads and aerodynamic loads at the
conditions. From this large database, a subset of measuring leading edge is nearly constant for the whole range of
channels was selected to retrieve the measurements of conditions between 7° and 28° that are included in the
interest to the aim of fatigue assessment on the slat-track. measured data. Therefore, the unit load case simulated for
On the selected channels, all significant flight conditions aSS =20° represents the most critical condition.
were identified and the relative data extracted such as to
represent a reliable base for the fatigue life prediction.
In particular, all loads acting on the slat track were measured 3.3 Skinning technique
at three locations. Strain gage measurements allowed
acquiring and recording inner moment MB and an axial For an efficient stress analysis at the boundaries of a three-
beam force L at these fixed locations of the slat track (figure dimensional component, it is useful to put an extremely thin
2). In addition, the moment Mp at the pinion acting on a rack skin of shell elements on the surface to build up an
at the inner side of the slat track was measured at a variable evaluation group. This drastically reduces result file size and
position depending of the slat setting angle. saves computation time for fatigue evaluation. For the slat
track model, 76344 TRIA3 elements of thickness 0.001 mm
have been generated on the surface.
R2
Aerodynamic R1
N Forces
Mb 3.4 Results for Unit Load Case
Leading Edge
T
L Three unit load cases for five different slat track positions
M
have been built up for the leading edge forces (T, N, M).
Figure 3 shows the resulting stress tensor field for the unit
Mp R3
load case N = 1 N. In retracted positions, for small angles
αSS = 20°
R4 αSS = 0° aSS, the stress field has only significant values in the area
αSS
before the last roller.

Figure 2: kinematics boundary conditions for varying slat track


position aSS

Under working conditions, the slat track is a moving


component whose absolute position can be referred to the
slat setting angle. A dedicated engine that drives a pinion
mechanism actuates the slat motion. As test data were
acquired in different flying condition, measured loads refer to
different slat extensions. This implies different positions were
to apply loads for an accurate numerical simulation based on
unit load cases.

To account for such load variability, a number of different


Figure 3: von Mises Stress Distribution for different slat track
kinematics boundary conditions for the different slat track
positions resulting from unit load N = 1 N, αSS = 0°
positions were simulated such as to cover the total range of
slat track positions between the fully retracted position - slat
setting angle 0° - and the fully extended position it - slat track
setting 20°. Transition matrices have to be then determined
for all different setting angles of the slat to compute the
aerodynamic forces T, N and M at the leading edge of the
slat.

Kinematics boundary conditions are defined by the varying


position of the rollers (R1-R4 in fig. 2), which change relative
location simultaneously with the measured load Mp. The
rollers constitute a set of kinematics constraints, as radial
displacements at roller locations have to be zero while the
pinion carries transversal loads. The group of acting forces is
however acting in the symmetry plane z=0, therefore all in- Figure 4: von Mises Stress Distribution for different slat track
plain reaction forces are taken in consideration with the positions resulting from unit load N = 1 N, αSS = 6°
defined boundary conditions.
setting angles aSS = 0°, 2°, 4°, 6° and 20°. Only one of the
five sub-time histories is active at one time step while the
others are to be set zero. In figure 6 the measurement data
(MB, Mp, L, aSS) for the flight no 7 are displayed.

The resulting leading edge loads (T, M, N aSS=0° and 20°)


are displayed in figure 7 and figure 8, where only six of the
15 created channels are displayed. The channels for setting
angle 2°, 4° and 6° are only non-zero over very short time
periods of extending or retracting slat track.

Figure 5: von Mises Stress Distribution for different slat track


positions resulting from unit load N = 1 N, αSS = 20°

3.5 Load Histories assigned to unit load cases

The transition matrices from the measured loads (MB, Mp, L)


to external leading edge loads (T, M, N) can be numerically
calculated using the finite element mesh. The calculated
transition matrices for the five setting angles have then been
applied to the measurement load histories MB, Mp and L by Figure 6: Measurement data for flight no. 7
using the LMS Channel Combination tool.

The numerical determination of the transition matrices is


achieved quite simply: knowing the values of the stresses at
the location of MB, the resulting moment can be calculated.
The axial beam stresses for the load L and the reaction force
Fp can be easily found in the result file of the Finite Element
calculation. Those values give the inverse of the transition
matrix for each slat track position.

The transition matrix is shown below for two positions:

Retracted aSS = 0°:

 T [ N ]   -0,0072 -0,0315 -0,137   M B [ Nmm ] 


    
 M [ Nmm ]  =  2,3124 -1,0222 113,1   M p [ Nmm ]  Figure 7: leading edge loads (T, M, N) for two slat track positions
 N [ N ]   0,015  
   -0,003 0,43   L[ N ]  aSS = 0° and aSS = 20° flight no. 7
 

Extended αSS = 20°:

 T [ N ]   -0,0029 -0,0318 -0,056   M B [ Nmm ] 


    
 M [ Nmm ]  =  0,9332 -0,9075 87,4   M p [ Nmm ] 
 N [ N ]   0,006 -0,0023 0,266   
     L[ N ] 

The transition matrices have then to be applied to the


measurement load histories (MB, Mp, L) to get the input time
series (T, M, N) for the damage evaluation. This is done for
the flight no. 7 (figures 6-7-8).

Since the setting angle varies over loading time, the load Figure 8: hysteresis filtered leading edge loads (T, M, N) in two
histories are split into five sub-load histories for the five positions flight no. 7
The leading edge loads consist of over 2 millions data points However, we are not interested in all these stress histories,
per flight, which makes damage calculation very time but only in extracting some specific characteristics at a given
consuming. For this reason a multi-axial hysteresis filtering location and in particular minimum and maximum stresses
with a filter width set to 5% of the total range has been as well as maximal stress amplitude on all potentially critical
performed at the load histories. This allowed reducing load planes.
history to 3500 data points.
The LMS FALANCS software can optionally provide in
outputs the maximal stress amplitude at the surface of a
3.6 Material Data finite element model. This allows figuring out a first
assessment on the locations where high stress amplitude
For the 18% nickel-cobalt-molybdenum maraging steel, the may lead to high damage values.
ultimate tensile strength is equal to 1900 MPa and the
Young Modulus is 200000 MPa. For the strain-life concept, Figure 10 shows that high stress amplitudes, up to 157 MPa,
the cyclic stress-strain curve and the strain-life curve have are obtained in the middle section of the slat track where the
been determined using the Uniform Material Law (figure 9). high stress amplitudes are most likely generated by the high
First results of the on-going material tests show, that the values of bending moment attained while the slat track is in
Uniform Material Law results in strain-life curve is in good extended position. In the notch areas where the slat track
accordance to the measured strain-life points. cross section changes considerably, maximal stress
amplitude values up to 90 MPa are calculated.

Ultimate Tensile strength

Constant amplitude test

Figure 9: cyclic stress strain curve and strain life curve for maraging
steel

3.7 Damage Parameters Figure 10: Distribution of the maximal stress amplitude for multi-
axial loading (T, M, N) at the slat track
The damage parameter is defined according to Smith-
Watson-Topper in its original description:
4.2 Damage Distribution
P
SWT
= (σ a +σ m )⋅ε a ⋅E (2)
A complete fatigue evaluation was performed applying the
leading edge load histories to the component under analysis.
A number of hypothesis and approximations are assumed: The result is that over the whole surface of the slat track no
the applicability of the Neuber approximation to calculate hot spot were identified that leads the structure to damage.
elastic-plastic stress-strain behaviour from elastic pseudo This can be depicted from figure 10, where the distribution of
stresses and strains, and the validity of critical plane maximum principle stress amplitude for multi-axial loading at
approach crack opening mode. Finally, in the computation the slat track is shown. Although two hot spot are visible
an endurance limit of 511000 cycles to failure was set. were the stress value attain 157 MPa, no damage occur over
the whole component.

4 RESULTS OF FATIGUE EVALUATION Since this value is under the endurance limit of the material
for the damage distribution, the loads have been magnified
by a factor 10 in order to verify the damage scenario in case
4.1 Maximal Stress Amplitude Distribution of heavier loads. The result can be seen in figure 11, where
damage on the surface is displayed logarithmically (0 means
Stress tensor time-histories are calculated at each node of damage sum = 1, -3 means damage sum = 0.001). In this
the finite element model. Thus, the crack opening stress case, the bending moment acting on the slat track in
time-histories in all potentially critical planes at a given extended position leads the component to damage in two hot
location are available during the fatigue calculation. spots located in central section of the component (circles 1
and 2 in fig. 11). In addition, two other hot spots are 5 LOCAL STRESS TENSOR HISTORIES
identified at the position in the components where a drastic
change of the cross section is attained. These are located in The local time histories are evaluated at the four potential
the front section of the slat track. crack initiation locations represented by the elements
indicated in table I. For all crack initiation locations, the
absolute maximum principle stress direction has nearly
constant angle over the whole measurement time. This
means that the hypothesis of local uni-axial stress state for
2 all four crack initiation locations can be considered valid.
Consequently, the absolute maximum stress can be taken
as local crack opening signal.

1 Element 240020 244722 270124 292131


no.
Position Intermed. Intermed. Beam Beam
Upper lower lower upper
Angle 60 10 90 80
Principal
Stress

Table I: Elements selected for the calculation of the local time


Figure 11: damage distribution for up-scaled leading edge loads by histories at the four crack initiation locations
factor 10

Figure 14: Local Time series at two crack initiation elements for
Figure 12: Detailed view on the hot spots at the middle section flight no. 07

A more detailed view on the middle section is given in figure The absolute maximal principal stress histories look quite
12 and 13. These two locations will be further investigated similar in shape (see figure 14) but not exactly proportional
for crack propagation simulation. To this end, local stress to each other. This is related to the non-proportional loading
tensor histories have to be prepared. of the three input signals.

6 CONCLUSION

A slat track has been examined under real-life loads acting


on the component in normal operating conditions. Using the
strain-life approach, a thorough evaluation of the damage
distribution over the slat track structure has been performed.
Since the slat track is a moving component, special efforts
have been devoted to a dedicated data pre-processing
aimed at transforming measured loads to external leading
edge loads. In addition, an intensive comparative analysis
has been carried out to extract a subset of data that
significantly represent the most critical load conditions for the
component under study. In order to identify critical fatigue
conditions, also the varying boundary conditions related to
the changing slat track positions during real-flight
Figure 13: Detailed view on the hot spots at the middle section manoeuvres have been reviewed. The use of a number of
unit load cases assigned to the load time histories allowed
performing a reliable fatigue analysis. The results of such state - Part II: Numerical simulation and expe-rimental
analysis have shown that in operating flight conditions no assessment through the weight function method, Int. J.
damage is found on the slat track. Moreover, a simulation Fatigue 21 (1999) pp. 89-96
has shown that in order to achieve a significant damage [14] Carpinteri A., Brighenti R., Spagnoli A, A fracture plane
scenario a magnification factor of 10 is required to the load approach in multiaxial high-cycle fatigue of metals,
time histories. This indicates that there is a high potential for Fatigue Fract. Engng Mater. Struct. 23 (2000) pp. 355-
design optimisation of the slat track. Local stress amplitudes 364
even for the extreme test flight are fare below the endurance
limit of the material. [15] Carpinteri A., Spagnoli A., Multiaxial high-cycle fatigue
criterion for hard metals, Int. J. Fatigue, 23 (2001) pp.
135-145
ACKNOWLEDGEMENT [16] Federal Aviation Regulation (FAR) Part 25, Paragraph
25.571 – “Damage Tolerance and Fatigue Evaluation of
This work was carried out in the context of the project IWT Structures”, U.S.A.
000327 “ASCO” (Damage detection, durability calculation [17] LMS INTERNATIONAL. LMS FALANCS manual, Leuven,
and redesign of safety-critical aircraft components). The Belgium, [www.lmsintl.com], 2001.
support of the Flemish Institute for scientific and
technological innovation in industry (IWT) is gratefully [18] Carmine R., de Voghel R., Van der Linden G.,
Guillaume P., Numerical Life Prediction of a Slat Track
acknowledged.
under Test Flight conditions, Proceedings of MSC
Conference, Toulouse 2002.
REFERENCES

[1] Garud Y.S., Multiaxial fatigue: a survey of the state of


the art, J. Testing Evalua-tion 40 (1981) pp. 165-178
[2] You B-R., Lee S-B., A critical review on multiaxial
fatigue assessments of met-als, Int. J. Fatigue 18 (1996)
pp. 235-244
[3] SCHIJVE, J., “Fatigue Life of Structural Components
nd
under Random Loading,” 2 Plantema Memorial
Lecture, International Committee on Aeronautical
Fatigue (ICAF), Stockholm, Sweden, 1969.
[4] Suresh, S., Fatigue of Materials, Cambridge University
Press, New York, 1991.
[5] Kaplan, M.P., Wolff, T.A., Life Extension and Damage
Tolerance of Aircraft, "Fatigue and Fracture", ASM
Handbook vol 19, ASM International, pp 557-565. 1996.
[6] Amzallag, C., Standardization of the rainflow counting
method for fatigue analysis, Int. J. Fatigue, vol 16, pp
287-293, 1994.
[7] Schijve, J., Fatigue crack growth under variable
amplitude loading, "Fatigue and Fracture", ASM
Handbook vol.19, pp 110-132, 1996.
[8] Findley W.N., A theory for the effect of mean stress on
fatigue of metals under combined torsion and axial load
or bending, J. Engng Industry, Trans. ASME 81 (1959)
pp. 301-306
[9] Brown M.W., Miller K.J., A theory for fatigue failure
under multiaxial stress-strain condition, Proc. Inst.
Mech. Engrs 187 (1973) pp. 745-755
[10] Matake T., An explanation on fatigue limit under
combined stress, Bull. JSME 20 (1977) pp. 257-263
[11] McDiarmid D.L., A general criterion for high cycle
multiaxial fatigue failure, Fatigue Fract. Engng Mater.
Struct. 14 (1991) pp. 429-453
[12] Carpinteri A., Macha E., Brighenti R., Spagnoli A.,
Expected fractu-re plane for multiaxial random stress
state - Part I: Theoretical aspects of the weight function
method, Int. J. Fatigue 21 (1999) pp. 83-88
[13] Carpinteri A., Macha E., Brighenti R., Spagnoli A,
Expected fractu-re plane for multiaxial random stress

View publication stats

You might also like