Vous êtes sur la page 1sur 9

Engineering Failure Analysis 56 (2015) 89–97

Contents lists available at ScienceDirect

Engineering Failure Analysis


journal homepage: www.elsevier.com/locate/engfailanal

Failure and repair analysis of a runway beam: Influence of the


standard applied to lifetime prediction
F. Pelayo ⇑, C. Rodríguez, A.F. Canteli
Department of Construction and Manufacturing Engineering, University of Oviedo, Campus de Gijón, Zona Oeste, Edificio 7, 33203 Gijón, Spain

a r t i c l e i n f o a b s t r a c t

Article history: The structural integrity of a welded structure is guaranteed insofar as the integrity of any
Received 27 October 2014 particular welded detail is ensured. This is highlighted in this paper which analyzes the
Received in revised form 3 March 2015 premature failure of a runway beam, occurring only two years after being commissioned.
Accepted 8 April 2015
Initial fractographic analysis and visual in-situ inspection of the beam suggest that the fail-
Available online 17 April 2015
ure was caused by fatigue. The fracture initiated in an apparently innocuous zone where
the beam joins a twin runway beam, the bracing of which is performed by a transverse
Keywords:
truss beam with fillet welds.
Failure
Fatigue
The failure analysis was performed according to two standards specific to runway beam
Cranes design: the former Spanish UNE 76202:1992 standard, applicable at the time the beam was
Strength built, and the current EN 1993-1-9:2005 standard. The main difference between the two
Stresses standards consists in the load spectrum considered for calculations. While UNE
76202:1992 prescribes a defined combination of the most unfavorable loads, EN 1993-1-
9:2005 leaves to the designer’s discretion the choice of more realistic load data directly
derived from in-service observations or measurements, which presumably better reflect
the loads the beam will have to support throughout its service life.
To achieve this goal, once the beam was repaired and operating once again, in-service
strain measurements were performed over several days. These measurements allowed
us to obtain a real load spectrum, which was taken into consideration in fatigue strength
assessment based on the EN 1993-1-9:2005 standard. The results provide a more realistic
fatigue life prediction than that obtained using UNE 76202:1992, a standard no longer
applicable.
Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction

Fatigue failure of runway beams is one of the most significant hazards that needs to be properly and constantly addressed
in order to ensure the reliability of the structural integrity of large bridge crane assemblies [1–5]. The design of joints and
details [6], inappropriate welding solutions applied between the different elements or a non-adequate fatigue strength
assessment can constitute the origin of many failures arising in these structures. Although the state of knowledge about
fatigue in these welding structures has been significantly enhanced [7–10], there still remain a number of safety-related
uncertainties in the design solutions proposed by the standards and even in the different verification methods currently
employed.

⇑ Corresponding author.
E-mail address: fernandezpelayo@uniovi.es (F. Pelayo).

http://dx.doi.org/10.1016/j.engfailanal.2015.04.008
1350-6307/Ó 2015 Elsevier Ltd. All rights reserved.
90 F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97

The aim of this paper is to obtain a technically reliable explanation for a specific case of failure which is relevant for a
number of reasons:

 First, due to the potential repercussion it may have on the respective costs of maintenance and repairing the damage to
such large runway beams, production continuity and even human lives.
 Second, to verify the suitability of design standards specific to fatigue.
 Third, to improve technical knowledge aimed at avoiding crack propagating and the hazard of failure in runway beams in
future designs due to the inappropriate execution of weld details.

The present study analyzes the failure of runway beams at a storage building for hot-rolling steel coils built in the 1960s
which occurred on the path of two bridge cranes that simultaneously operate in tandem most of the time. The beams under
study are constituents of a reinforced beam consisting of two twin-runway beams (the second of which is part of another
parallel running bridge crane). The original structure dating from the 1960s (old design, see Fig. 1) was designed as a com-
plex structure in which two transverse trusses respectively join the top and the bottom flanges of the twin beams by means
of fillet welds.
After several incidents related to local damage as well as the detection of visible cracks in the runway beams, particularly
in the central area of the building, in 2008 the company decided to revise their design. On the one hand, the beam web height
was increased up to 1.98 m and, on the other, the support conditions were modified to render them simply supported. See
Fig. 1.
However, the design of the joint between the twin beams was maintained as in the previous one, i.e. by means of two
transversal bracing truss structures respectively connecting the top and bottom of the two main beams via welded base
plates. The base plate of the upper truss was directly welded to the corresponding upper flange, while the base plate of
the bottom truss was welded to the upper face of the corresponding bottom flange.
Less than 3 years after the revised runway design came into service, a significant large vertical crack was discovered in the
central area of the midspan of the new beams leading to ultimate failure (see Fig. 2). Consequently, a detailed analysis of this
failure and a review of the design involved were required.
The failure analysis showed that the crack started in the welded joint between the beam’s bottom flange and the lower
bracing truss. Subsequently, the crack grew due to fatigue, first causing complete failure of the bottom flange and finally
failure of the beam web. When the failure was detected, only the upper flange of the beam was supporting the loads.
Fig. 3 shows a detail of the beam failure and crack initiation site.
The initial in-situ analysis showed that the premature failure of the beam was due to the presence of the welding joint
between the bottom flange and truss. Perhaps this class of welded detail, with a high concentration of stress, was not taken
into account in the 2008 structure design. It was hence necessary to check the structural integrity of the beam, taking into
account the different welding details due to the major influence they have on the fatigue response of a welded structure [11].
In this study, a review of the integrity of the beam is carried out according to two standards currently in use: the Spanish
UNE 76202:1992 standard [12] and EN 1993-1-9:2005 [13]. In particular, an assessment of the fatigue life prediction
according to these two standards –the bases of which are very similar – was carried out.
Both standards prescribe the same procedure for dealing with fatigue in welding structures, using the same FAT classes
for the same welding details [12,13] and the same general expression for fatigue life prediction:
" Drc
#3
N ¼ 2  106 Ym
ð1Þ
Dr  Y f

1960s Old design

2008 Revised design

Fig. 1. Runway beam support conditions and bending moments in the 1960s (old design) and in 2008 (revised design).
F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97 91

View of the crack


path in the web
2m

Fig. 2. Location of the crack and beam failure.

30 mm
Inside view

350 mm

Crack iniaon site in the boom flange

Fig. 3. Details of crack and failure.

where N is the expected number of cycles, Drc is the joint class category, Dr is the real stress range, and Yf and Ym are safety
factors [13].
Both standards also propose the use of a nominal stress range, Dr, based on an equivalent load, Qe:
Dr ¼ DrðQ e Þ ð2Þ
with
Q e ¼ ki  ufat  Rmax ð3Þ

where ki are damage equivalent factors, ufat is a dynamic factor [13], and Rmax is the maximum load per crane wheel.
The bases of these two standards are very similar, except with respect to the consideration of load. The significant differ-
ence between the two standards in the calculation of runway beams is related to ki . In the case of runway beams, actions in
the Spanish regulations refer to the UNE 76201:1988 standard [14], while actions in the EN 1993 standard are based on the
current EN 1991-3 standard (actions induced by cranes and machinery) [15].
On the one hand, UNE 76201:1988 [14] suggests the use of a load combination based merely on the characteristics of the
crane and therefore independent of the load history:
ki ¼ kQ ð4Þ
kQ being constant for a given runway structure.
On the other hand, in EN 1993-1-9, ki relates the loading spectrum to the reference number of cycles, 2  106 [13]:
8  
> P   13
>
> Q ij
Pnij
>
> k1;i ¼ Q imax

< nij

ki ¼ k1;i  k2;i P 13 ð5Þ


>
> nij
>
> k2;i ¼ 2106
>
:
92 F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97

where Qij represents each of the load amplitude values of the spectrum, and nij the number of times the load acts in a certain
period of time. In order to apply this standard properly, the loading spectrum acting on the structure should be determined
with the maximum possible accuracy.
With this aim in mind, in-service strain measurements were performed over several days once the runway was repaired.
These measurements allowed a real load spectrum to be determined, which was then taken into consideration to carry out a
new review of the failure analysis based on EN 1993-1-9:2005. The results show a more realistic fatigue life prediction than
that obtained using the former UNE 76202:1992 standard.

2. Revised 2008 runway description and configurations studied

The runway beams under study were 18 meters long and approximately 2 m high operating under simple support
conditions. The beams were connected to another twin runway beam as support for a parallel bridge crane by means of
two bracing trusses which join the upper and lower flanges of the beams by means of fillet welds (see Fig. 4(a)).
The runway beam supported two bridge cranes of approximately 50 t capacity and 120 t weight working simultaneously
most of the time with a maximum speed of 120 m/min. Although both cranes move along the entire glide-path, crane traffic
is concentrated in the central area of the industrial building, precisely where the failure occurred. Moreover, failure occurred
in the midspan of the beam where higher bending moments appear due to the simple support conditions of the revised 2008
runways; as opposed to the 1960s old design (see Fig. 1).
The transport pattern of the cranes is conditioned by the location of the hot-milling steel end-line in the hoisting
operation and the storage park of the coils, both situated in the central area of the building. This means that the two cranes
operate within this area most of the time; this is why this middle track-line is subjected to most of the load cycles.
As a first approach to the failure analysis, two beam configurations were analyzed against fatigue. The first comprises the
2008 revised design in which the failure occurred: configuration (a) (see Fig. 4(a)).
The second is configuration (b), in which the effects of removing the four central connections between the bottom truss
and the evaluated beam (see Fig. 4(b)) require analysis, as the welded joint between the lower bracing truss and the bottom
flange of the beam was apparently responsible for initiating the crack.
Finally, a new design solution consisting in reinforcing the runway beam by means of an inverted T profile is proposed
aimed at reducing the maximum levels of stresses and deflections, without the need to remove the bottom bracing truss:
configuration (c) (see Fig. 4(c)).
Finite element calculations (FEM) were performed using the ABAQUS finite element program to assess the consequences
of the structural changes resulting from the revised design (configuration (a)) and those proposed here (configurations (b)
and (c)) for the runway beam, as well as the effectiveness of the design improvements undertaken. The material properties
were provided by the company as per the original specifications corresponding to a S375-JR steel for the structural compo-
nents of the main beam and a S255-JR steel for the bracing truss. The beam was modeled using 2D dimensional quadratic
shell elements (S8R), while beam elements (B21) were used for the truss. The model for configuration (a) is shown in Fig. 5.
Static validation of the three structural configurations was carried out using the corresponding FE models. All the
configurations meet the static requirements of the two standards used in this study (UNE 76201:1988 and EN 1993-1).
The same FE model was used to check the fatigue requirements. The loading spectrum in the application of the
EN-1993-1-9 standard [13] was obtained in two different ways: by means of the flow of loads estimated by the crane
operators (based on their experience), and using the operational stress spectrum resulting from strain gage measurements
recorded over several days.
A modal analysis of the beam was carried out with the aim of improving the reliability of the numerical model. The result-
ing experimental natural frequencies of the beam were used to verify the FE runway-model. In this case, the experimental
frequencies were in good agreement with the numerical values, so no further modal-updating was necessary. More details
are given in the experimental section.

(a) (b) (c)


Fig. 4. Beam configurations studied: (a) complete bottom truss, (b) bottom central truss removed, and (c) proposed reinforced design.
F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97 93

Fig. 5. Finite element model for the revised 2008 design of the runway beam in which failure occurred.

3. Experimental program

3.1. Experimental set-up and stress estimation

In order to assess the real operational loading conditions the runway beam is subjected to, the strains in the beam were
continuously measured over a period of four days. Two measurements were carried out: in the repaired beam where, addi-
tionally, the middle four modules of the bottom bracing truss were removed (configuration (b)); and in the next contiguous
runway beam, exhibiting the same dimensions and design as the beam that failed (configuration (a)).
Four 350 X unidirectional strain gages (HBM) were used to measure the strains in the truss, while eight 120 X rosette
strain gages (HBM-LY11) were employed in the analysis of the main beams. The strain gages were allocated at accessible
points as close as possible to the mid-section of each beam (where maximum stresses are expected). The detailed locations
of the gages are respectively shown in Figs. 7 and 8. In the analysis, the stresses arising at the bottom of the mid-section of

(a) (b)
Fig. 6. Location of the strain gages in beam configurations (a) and (b).
94 F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97

(a) (b)
Fig. 7. Details of the location of the strain gages in beam configurations (a) and (b).

7
x 10
8
Movements of the cranes with loads Avilés a Gijón
7 Avilés a Gijón
Gijón a Avilés
6
Parada-descarga-arranque
Stop - lift load - Start
Stress [Pa]

5 Config.Viga
(a)B
Config.Viga
(b)A
4

0
0 20 40 60 80 100 120 140 160 180 200 220
Time [s]

Fig. 8. Von Mises stresses for both beams (a) and (b) in the bottom central section at different time intervals.

the beams (strain gages 2 and 6 in Figs. 6 and 7, respectively) were used to obtain the equivalent coefficients (ki ) needed to
apply EN1993-1-9 [13].
Additionally, four accelerometers (4508B) were used during one of the experiments to perform a modal analysis
identification of the natural frequencies of the repaired beam. The accelerometers were located equally spaced along the
beam to measure in the vertical direction.
Data acquisition was carried out using a CDAQ National Instruments system with strain gages (NI9235 and NI9236) and
acceleration (NI9234) modules. The CDAQ system was connected to a laptop to view and store the data. The sampling
frequency for the strain gages was 10 Hz, whereas 256 Hz was used for measuring the accelerations.
Before the von Mises stresses were calculated from each strain gage, the raw data had to be processed. Signal processing
was performed using MATLAB and consisted mainly in de-noising the raw data in order to remove the high-frequency base
level that is usually present in strain gage measurements. Once the data had been filtered, the von Mises stresses could be
directly determined. Fig. 8 shows a time interval of the measured stresses obtained for the bottom central area of both
beams. From the results, different crane operations could be clearly identified in the time history of stresses (see Fig. 8).

3.2. Experimental loading spectrum

The operational loading spectrum was obtained using the von Mises stresses estimated from the experimental measure-
ments in the beams. The number of cycles at each different stress level was determined using a LabVIEW application devel-
oped for this purpose. The operational spectrum was obtained following the steps below:

(1) The von Mises stresses signal file is uploaded.


(2) A decimation of the signal is applied (if desired).
F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97 95

Table 1
Number of stress peaks for one day of experimental recordings obtained in the beams.

Stress (MPa) Beam configuration (a) Beam configuration (b)


Number of peaks in the strain rosette Number of peaks in the strain rosette
5 6 7 8 1 2 3 4
80 10 7 0 0 0 0 0 0
70 100 64 0 0 6 0 0 0
60 614 697 3 0 88 5 0 0
50 246 306 106 0 231 96 0 0
40 51 78 449 0 253 272 56 20
30 53 37 146 8 295 313 382 235
20 205 588 64 52 265 211 378 350
10 587 234 800 1306 1177 354 367 617

(3) A high pass, low pass and smoothing filters are applied to improve de-noising of the signal.
(4) A de-trend is applied to the signal to remove possible offset and temperature changes (heating or cooling) during
measurement.
(5) Peak counting is performed at the stress levels of interest (in this case, 10–80 MPa).

The stress cycles recorded on one particular day are presented in Table 1. From this table, it can be inferred that beam (b)
works most of the time below 50 MPa. However, the highest values are around 80 MPa, which are below the previously
obtained maximum stress predictions used in the design of the beams.

4. Model validation and fatigue life predictions

4.1. Modal analysis of the beam

The measured accelerations in one of the beams were analyzed using the ARTeMIS MODAL application. The experimental
natural frequencies were identified by means of the Frequency Domain Decomposition (FDD) technique [16]. The numerical
frequencies are presented in Table 2, together with those obtained experimentally. The results show a good correlation
between experimental data and the numerical model, given that the deviation is less than 5%.

4.2. Comparison between expected and measured load cycles in the runway beams

The first step for fatigue life prediction consisted in comparing the original load flow provided by the company (based on
the experience of personnel) with that derived from the experimental strain-based estimations. Due to the particular pat-
terns of the possible loadings under consideration (see Fig. 8), a specific application based on LabView and MatLab was
designed to estimate not only the stress peaks at the different levels used in the original predictions (from 10 to 90 MPa),
but also to take into account the width or shape of the measured stresses (see Section 4.2). The main objective was to dif-
ferentiate the cycles at the specific assigned stress level, e.g. that corresponding to the travelling of the crane along the beam
while transporting a suspended steel coil, from those assigned to an operation of load hoisting while the bridge crane
remains stationary on the beam. This information was used to check the loading information provided by factory personnel
(based on their experience in crane operations).
The number of cycles per year in the two beam configurations (a) and (b) as provided by the company and those obtained
from the experimental measurements are presented in Table 3. The results show a clear underestimation of the real load
cycles to be considered for lifetime prediction if the company’s proposals are used. The differences for runway beams (a)
and (b) amount to approximately 36% and 23%, respectively.
Subsequently, the experimentally obtained cycles were used for fatigue life estimation related to safety and reliability
analysis.

Table 2
Numerical and experimental natural frequencies.

Frequency Mode
1 2 3
Experimental (Hz) 10.91 33.17 42.78
Numerical (Hz) 11.13 31.77 44.63
Error (%) 2.02 4.22 4.32
96 F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97

Table 3
Comparison between expected and experimentally mea-
sured load cycles.

Cycles per year


Configuration (a) Configuration (b)
Expected 185,740 202,840
Measured 291,265 262,882

4.3. Fatigue life prediction

4.3.1. Fatigue life prediction using nominal stresses


As a final step, the operational stresses were used to obtain a more reliable lifetime prediction of the runway beams as per
the method proposed in the EN-1993-19:2005 standard. The results are compared with those obtained using the former UNE
76201:1988 standard. The same values of the safety factors Yf = Ym = 1 were applied in expression (1) for all configurations.

4.3.2. Fatigue life prediction using the number of stress peaks


As can be seen in Fig. 9, according to both EN-1993-1-9 and UNE 76202 standards, the welded joint fatigue classes of the
different welded joints are: FAT-35 for configuration (a), FAT-45 for configuration (b), and FAT-35 or FAT-125 for configura-
tion (c). However, the significant difference between the two standards, as previously explained, lies in the consideration of
load. Table 4 shows the fatigue life predictions according to both standards for all beam configurations.
When using UNE 76202, the same equivalent load should be considered for all configurations. As this load is very high
(Qe = 170 kN), shorter life predictions were obtained for configurations (a) and (b). The fatigue life for configuration (c)
depends on the reinforcement of the inverted T profile web length. As can be seen, with a 400 mm web length in the T rein-
forcement, the fatigue life prediction only increases up to 8 years or 17 years if, in addition to the reinforcement, the joint
between the lower truss and the beam (FAT-45) is removed solely in the mid-section of the beam.
More realistic fatigue life predictions were obtained using the equivalent load, as proposed by EN-1993-1-9, obtained
from the experimental strain measurements in configurations (a) and (b) or from the company predictions in configuration

a
Fig. 9. Welded joint fatigue classes for the different beam configurations analyzed.

Table 4
Fatigue life prediction based on the nominal stress method.

Beam configuration Detail category UNE 76202 EN-1993-9


Qe Dr N Life Qe Dr N Life
(kN) (MPa) (Cycles) (Years) (kN) (MPa) (Cycles) (Years)
(a) 35 170 74 211,611 0.73 86 33 1,155,136 3
(b) 45 170 74 449,751 1.54 60 23 6,528,345 17
(c) t = 200 mm 35 170 41 1,244,178 4.27 101 29 3.51E6 12
(c) t = 200 mm 125 170 53 26,238,103 90 101 39 65.8E6 226
(c) t = 300 mm 35 170 37 1,692,891 5.81 101 26 4.88E6 16.7
(c) t = 300 mm 125 170 52 27,305,804 93.7 101 36 83.7E6 287
(c) t = 400 mm 35 170 33 2,386,120 8.2 101 23 7.05E6 24.2
(c) t = 400 mm 45 170 33 5,071,375 17.4 101 23 14.98E6 51.4
(c) t = 400 mm 125 170 50 29,103,830 99.9 101 35 91.1E6 312
F. Pelayo et al. / Engineering Failure Analysis 56 (2015) 89–97 97

(c) proposed in this paper. Thus, in accordance with EN-1993-9 predictions, the actual failure of the original beam had
occurred after 3 years of service life. Moreover, by removing the four central segments of the truss (configuration (b)), the
fatigue life of the beam can be extended up to six times. However, this option does not seem to be the most appropriate,
because there is a risk of lateral instability in the beam once a portion of the lower truss has been removed. If a T reinforce-
ment is used, however, a web length of 400 mm would be needed to extend the beam fatigue life up to 24 years or 51 years if,
in addition to the reinforcement, the joint between the lower truss and the beam (FAT-45) is removed solely in the mid-sec-
tion of the beam.
The fatigue life predictions for the previous EN-1993-1-9 standard were obtained taking into account the number of
stress peaks. To further improve the calculations, the rainflow method was applied as per ASTM E 1049-85 [17], using
the rainflow MATLAB algorithm package developed by Adam Nieslony. The stresses measured via the strain gages were
directly applied in the rainflow algorithm to obtain the stress amplitudes and their corresponding number of cycles. The
damage was obtained by means of expression [2]:
X     !
N Dr 3
D¼  ð6Þ
2  106 Dr c

Finally, if the cumulative damage is divided by the total lifetime, the fatigue life prediction for the measured stresses can
be estimated, resulting in 4.1 and 24.8 years for beam configurations (a) and (b), respectively.

5. Conclusions

A preliminary analysis shows that the runway beam failure reported in this paper should be assigned to the amplified
fatigue loading acting on the welded joint located in the mid-section of the beam, more precisely on its bottom flange.
In order to analyze and improve the design of the beam, different repair options were studied taking into account two
standards: the Spanish UNE 76202 standard and Eurocode EN-1993-1-9.
Though in the fatigue life prediction analysis of the runway beam both standards met safety requirements, the results
based on EN-1993-1-9 were found to be more reliable than those obtained using UNE 76202 due to the more realistic load
history proposed in the former. Furthermore, the fatigue life estimation can be significantly improved if an operational stress
spectrum is applied.
According to both standards, a fatigue life of the beam of up to 50 years requires reinforcing the web of the beam by the
addition of an inverted T profile.
As not all the possible welding details that may arise and may be determinant in practical fatigue design are comprehen-
sively covered by the standards, the latter do not always ensure a safe fatigue life for the intended period.
A reliable load spectrum obtained by direct measurement is recommended as a safer alternative to the sometimes impre-
cise load information provided by factory experience.
Replacing the original runway design due to the presence of expected cracks once the predicted lifetime has been sur-
passed by a new design that involves the elimination of redundancy in the structural solution requires a careful review of
the new design conditions introduced, which are not always perceived.
Notwithstanding a safe static design, fatigue design requires thorough analysis of the structural solution adopted in the
new design. In particular, seemingly accessory details, such as a simple new welding joint to connect runway beams to a
bracing truss, may become determinant for fatigue life.

References

[1] Mueller JE. Lessons from crane runways. Eng J AISC, Chicago, IL 1965;2(1):16–20.
[2] Demo DA, Fisher JW. Analysis of fatigue of welded crane runway girders. J Struct Div ASCE, May, Reston, VA 1976.
[3] Reemsnyder HS, Demo DA. Fatigue cracking in welded crane runway girders; causes and repair procedures. Iron Steel Eng, Pittsburgh, PA 1978.
[4] Kuwamura H, Hanzawa M. Inspection and repair of fatigue cracks in crane runway girders. J Struct Eng 1987;113(11):2181–94.
[5] Ricker DT. Tips for avoiding crane runway problems. Eng J, AISC, Chicago, IL 1982;19(4).
[6] Fisher JM, Van de Pas JP. New fatigue provisions for the design of crane runway girders. Eng J AISC, Chicago, IL 2002;39(2):65–73.
[7] AISE. Fatigue life analysis of crane runway girders. Iron Steel Eng 1997;73(7). Pittsburgh, PA.
[8] Fan X, Xu G, Wang A. Evaluation method of remaining fatigue life for crane based on the acquisition of the equivalent load spectrum by the artificial
neural network. Jixie Gongcheng Xuebao (CJME) 2011;47(20):69–74.
[9] Caglayan O, Ozakgul K, Tezer O, Uzgider E. Fatigue life prediction of existing crane runway girders. J Constr Steel Res 2010;66(10):1164–73.
[10] Euler M, Kuhlmann U. Crane runways–Fatigue evaluation of crane rail welds using local concepts. Int J Fatigue 2011;33(8):1118–26.
[11] Instituyo Técnico de la Estructura en Acero (ITEA). Tomo 14: Diseño en Fatiga; 1999.
[12] UNE 76202:1992. Basis for the fatigue design.
[13] EN 1993-1-9:2005. Eurocode 3: design of steel structures – Part 1–9: fatigue.
[14] EN 76201:1988. Steel constructions. Rolling ways for bridge cranes. Basis of calculation.
[15] EN 1991-3:2006. Eurocode 1: actions of structures – Part 3: actions induced by cranes and machinery.
[16] Brincker R, Zhang L, Andersen P. Output-only modal analysis by frequency domain decomposition. Smart Mater Struct 2001;10:441–5.
[17] ASTM E1049-1985 Standard practices for cycle counting in fatigue analysis.

Vous aimerez peut-être aussi