International Journal of Fatigue 128 (2019) 105190

Contents lists available at ScienceDirect

International Journal of Fatigue

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

Fatigue damage analysis of a riveted steel overhead crane support truss T

P. Lehner , M. Krejsa , P. Pařenica , V. Křivý ⁎, J. Brožovský
a a a b, a

a
VSB-Technical University of Ostrava, Faculty of Civil Engineering, Department of Structural Mechanics, Ludvíka Podéště 1875/17, 708 33 Ostrava-Poruba, Czech
Republic
b
VSB-Technical University of Ostrava, Faculty of Civil Engineering, Department of Building Constructions, Ludvíka Podéště 1875/17, 708 33 Ostrava-Poruba, Czech
Republic

A R T I C LE I N FO A B S T R A C T

Keywords: A method is proposed to evaluate the residual life of existing steel structures in terms of fatigue damage. The
Fatigue modelling proposed method is demonstrated on the assessment of selected elements in a nearly 100-year old exterior
Riveted joints riveted crane support truss. The Monte Carlo method is used to generate a time-dependent load based on the
Overhead crane operation of the overhead crane, and detailed numerical modelling based on rainflow counting and the
Rainflow
Palmgren-Miner rule to assess their fatigue resistance and estimate their residual life is evaluated and discussed.
Monte Carlo method
This study provides a novel approach to the reliability assessment of riveted joints stressed by cyclic loading and
stochastic modelling of the load history of existing structures.

1. Introduction resulting reduction in strength due to cyclic, time-varying loading is

necessary [6,7]. This study accordingly focuses on the analysis of a
In the Czech Republic, industrial production has continually in- load-bearing overhead crane support structure subjected to repeated
tensified since the end of the 18th century. Some of the original steel- loading. The modern design of these types of steel structures must be in
riveted structures erected at that time have been continuously operated accordance with applicable standards [8–10], which enable efficient
until the present, and these historical structures still fulfil their original and sustainable design often using the appropriate application of
function. The assessment of the remaining fatigue life of the members of available computing software [11–14]. However, due to the ageing of
these structures is difficult to estimate because such older materials and existing infrastructure worldwide and to the lack of codes for the con-
their characteristics are no longer commonly known, and often, the tinued safety management of older structures over their lifetimes, it
load history has not been recorded. Furthermore, changes in such remains necessary to develop mathematical tools to evaluate structural
structures due to repair or strengthening measures within the last reliability over time taking into consideration any reduction in struc-
hundred years may not be indicated in the drawings. The supporting tural capacity [14]. There have been a number of studies on the sto-
truss of the crane runway serving the steel warehouse of Vitkovice chastic estimation of reliability and subsequent prediction of the re-
Machinery Group, Czech Republic (see Fig. 1), is an example of such a maining life of various load-bearing elements and structures [15–19].
structure. However, there has been to date no generally valid calculation meth-
In many ways, steel bridges exhibit very similar problems to those of odology based on probabilistic modelling that is directly applicable to
such truss structures in industrial facilities [1–3]. The fatigue assess- the analysis of existing load-bearing overhead crane support structures
ment guideline for the estimation of the remaining fatigue life of steel assembled by rivets.
bridges [4] is a useful tool commonly applied for condition assessment A variety of scientific research has been published investigating
that complements bridge management systems. Riveting was the fatigue crack propagation due to dangerous details and varying applied
dominant joining technique for steel structures from the last decades of cyclic loads [20,21]. The objective of this paper is to present a method
the 19th century until the late 1930s, and determining the expected for evaluating fatigue resistance and predicting the residual life of
lifespan of riveted joints, with a focus on fatigue, is quite complicated riveted steel structures that can be applicable to the wider civil en-
due to the general present-day lack of understanding of riveted struc- gineering practice. Project designers in civil engineering typically lack
ture fatigue behaviour [5]. Additionally, more knowledge and experi- sufficient knowledge and experience to apply fracture mechanics in
ence in the field of fatigue crack propagation in steel structures and the static assessments, so an approach based on the traditional Wöhler

⁎
Corresponding author.
E-mail address: [email protected] (V. Křivý).

https://doi.org/10.1016/j.ijfatigue.2019.105190
Received 14 January 2019; Received in revised form 1 July 2019; Accepted 4 July 2019
Available online 05 July 2019
0142-1123/ © 2019 Elsevier Ltd. All rights reserved.
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 1. Steel truss structure supporting an overhead crane.

concept of the fatigue of steel structures was chosen for further analysis final stage combined the results of the FEM analysis with a 3D model of
[22–24]. Generally, most available commercial static analysis software a selected diagonal member and connection using the ANSYS
makes it possible for steel structure designers to use such advanced Workbench Fatigue Module, and a mathematical evaluation of the re-
numerical models in practice. sults provided predictions of the fatigue crack timing and location, and
This paper accordingly describes the estimation of the structural thus the remaining fatigue life. In this study, a diagonal member was
fatigue resistance of a riveted overhead crane support truss using de- selected for examination and modelling that could be removed from the
tailed numerical models of its structural elements and joints while re- structure for laboratory examination and that was determined to carry
specting the recommended normative procedures based on the Wöhler the greatest axial force among diagonals of similar cross sections.
concept of fatigue, which includes a cumulative damage model based
on well-known Palmgren–Miner rule [25,26]. This study develops and
2.1. 3D FEM numerical model of overhead crane
builds upon information previously published in [27], in which the
basic principle of the proposed method was outlined and a simple ex-
The SCIA software package was used to prepare a basic 3D model
ample application was provided. In this paper, the use of a stochasti-
composed of beam elements for the static analysis of the selected
cally derived annual load is extended to the proposed procedure for the
continuous frame overhead crane supporting truss [35], shown in
assessment of the residual fatigue life of riveted structures, and the
Fig. 2. The frame structure was 150 m long and composed of the six
individual components of the calculations in this method are described.
spans supported by seven truss columns (one of which was the main
double column in the middle of the structure). The spans were 23.87 m
2. Fatigue assessment of existing steel structures and 26.04 m long. The main columns were 4.34 × 1.16 × 11.45 m. The
main horizontal elements were constructed of two L140x30x14 mm
The proposed computational process described in this paper in- with a steel plate P15x430 mm, fixed by rivets onto the top chord. The
cludes an analysis of the fatigue load on an existing load-bearing bottom chord of the horizontal elements was composed of the same
overhead crane support truss. The subject structure is a 3D steel truss angles, but without the plate. The diagonals consisted of two L90x10
built in the 1920s and constructed of mostly riveted joints. In order to elements and two L120x10 elements. At all connections, the connection
conduct an accurate fatigue assessment, it is necessary that the loading plates were P15s.
history, the details of the individual elements and joints, and the ma- In the model, the frame corners were considered to be fixed due to
terial properties be accurately reflected in the analysis. It is then pos- their massive riveted connections, as were the supports of all columns.
sible to determine the load duration on a selected element in the Initial material testing of the steel indicated that it was similar to grade
structure. Then, the estimation of the residual lifetime of the structure S235 [36]. This material classification is only useful for demonstrating
can be performed using methods based on probabilistic or statistical the functionality of the procedures in this paper; it remains necessary to
calculations [28–34]. determine the complete mechanical properties and chemical composi-
The analysis of the overhead crane support structure can be divided tion of the structural materials using samples taken from the structure
into three stages. In the first stage, a static FEM model of the entire to capture the overall fatigue life. Notably, the chosen material classi-
support structure was prepared in the SCIA Engineering software that fication does not affect the distribution of internal forces in the struc-
could reflect the effects of a moving load through multiple load cases as ture, which is important for further calculations using this procedure.
an influence line. In the second stage, special MATLAB code was written An initial analysis was performed to determine the most loaded
and used to generate an estimated load history using stochastic calcu- element of the structure. This calculation included a simulation of the
lations in combination with the derived influence line. Data from the moving overhead crane wheels on the crane track using the time-de-
load history were divided into a spectrum of stress peaks using the pendent load step method. The simulation was conducted using two
rainflow counting method and subsequently used as load cases. The loads of 100 kN at a constant spacing of 4.875 m according to the

2
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 2. FEM model of the steel truss supporting the crane track in SCIA.

wheelbase of the overhead crane. A readily accessible diagonal, B355 in The load can be characterized by a time course over the investigated
the third span as shown in Fig. 2, was selected for evaluation and re- life span. For structures such as crane tracks, the size of the wheel
moved for material testing purposes. However, it is important to note pressure areas, the number of movement repetitions, and the resulting
that the proposed procedure is applicable to any part of the structure internal forces in the target element should be determined. To de-
and makes it possible to determine the residual life of any element. The termine the equivalent fatigue load, dynamic coefficients corresponding
validity of the element selection can, however, be affected by the cal- to the applied load must be implemented. From the point of view the
culated maximum stress, or other assumptions. high-cycle fatigue of the elements of the subject structure, it was pos-
In order to conduct the numerical analysis of the riveted connection sible to use the so-called spectrum stress range in place of the standard
model, the influence line of the moving load on the selected diagonal equivalent fatigue load [8,24]. This spectrum stress range can be ob-
was determined as shown in Fig. 3. This was obtained using the nu- tained by stochastic analysis using the Monte Carlo method by entering
merical model of the structure, geometrical and technological para- random input parameters, as discussed in Section 2.2.1. The result of
meters of the crane under operation, and expert input from the oper- the analysis is the variable amplitude load history of the selected ele-
ating technician, who estimated the working conditions of the crane, ment. This record can then be applied to a detailed model of that
including its characteristics (e.g. distribution of the lifted mass), particular element or connection.
number of crane movements over the selected working cycle (e.g. one
working shift), and number of working shifts per year, etc. A transition
analysis was then used to apply a constant moving load from the 2.2.1. Input parameters for load history estimation
overhead crane wheels. The resulting influence line was then used to The structure under consideration was consistently loaded by the
evaluate the load history in the following calculations. overhead crane, which weighed 25.72 tons, and its cargo, which could
weigh up to 10,000 kg. In the stochastic calculation of the load history,
140 work paths were reported in one shift. Movements with and
2.2. Load modelling and structure behaviour without load were considered separately: half of the load paths were
without load, 40% carried cargo weighing between 0 and 5000 kg, and
To evaluate fatigue damage, it is necessary to obtain a history of the 10% carried cargo weighing between 5000 and 8000 kg. These char-
loading (or load response) of the overhead crane structure for a suffi- acteristics were entered as three uniform distributions within the given
ciently representative period of its operation. Long-term measurements limits. Dynamic coefficients φ1 = 1.10 and φ2 = 1.23, corresponding to
are usually very expensive and often very complicated [37], so simu- the type of drive and the type of overhead crane, respectively, were
lation methods can be used to adequately determine the load history of taken from [8].
critical elements or details [38]. Accordingly, a computational proce- The cab of the overhead crane was capable of shifting from
dure was employed that allowed for the determination of load history y = 1.9 m to y = 16.16 m, and the distribution of this parameter was
or load response in the selected details using the Monte Carlo simula- uniform. Using a simple static analysis including cargo loading, crane
tion method [39]. cab displacement, and actual crane weight, the ranges of reactions at

0.8
0.6
0.4
Force [-]

0.2
0
0 20 40 60 80 100 120 140
-0.2
-0.4
Shift path [m]
Fig. 3. Influence line for selected diagonal B355 generated by simulation of moving overhead crane load.

3
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 4. Amplitude of axial force load history data for the selected diagonal during one work shift.

the crane wheels were calculated. These reactions were used to simu- 3. Fatigue life prediction
late the movement of the overhead crane in combination with the in-
fluence line of the chosen diagonal shown in Fig. 3. Another random 3.1. FEM model of connection
parameter was the location of the crane along the load path. The dis-
tribution of this parameter was also uniform along the 150-m length of For the analysis of the residual life of the riveted diagonal connec-
the crane tracks. The result of the simulation was the amplitude of axial tion, a model of the connection was prepared and evaluated in the
forces [kN] for the chosen period of 250 days, which corresponds to one ANSYS Workbench software. The model included the gusset plate, the
working year. This was then used to produce a variable amplitude load diagonal, and three 24-mm diameter rivets and was constructed of
history corresponding to the selected structural element. The resulting volumetric elements connected by non-linear friction contacts (see
axial forces corresponding to one work shift are shown in Fig. 4. Fig. 6).
The load was introduced as an axial force in the diagonal according
2.3. Sorting by rainflow counting to the generated values in Table 2. The boundary conditions were de-
fined as fixed at the locations of the rivets between the gusset plate and
To analyse the connection details, it is important to determine the the other truss members at the bottom and left edges of the plate. The
specific effects of the load [24], for which a spectrum stress range, axial load was applied as a surface load on the cross section of the
derived from the amplitude of the load history using the rainflow diagonal. The rivets in the model were partially pre-stressed to reflect
counting method [40], can be used. This method was developed in the the method by which riveted connections are assembled. The preload
1960s and has been widely used for the fatigue analysis of machine force was considered to be 1 kN per rivet, which was chosen in part to
parts and other engineering components. The rainflow method reduces provide better convergence of the contact model. The linear elastic
the number of peaks and valleys in the amplitude of the load history material model was used. For the material fatigue characteristics,
and can be used to determine so-called half-cycles, which are then se- though the basic material properties of the steel corresponded to
quenced and sorted. Several available software packages can be used modern S235 steel [36], because the exact S–N curve of the subject
for the method such as the Fatigue Module in ANSYS Workbench [41]. material was not yet available, the S–N curve for riveted joints (shown
However, this procedure is not easy to apply in complex structures and in Fig. 7 and taken from [4,42]), was used as the subject truss was
some input parameters cannot be changed. nearly the same structural system and age as the large number of steel
Therefore, a separate algorithm that uses the rainflow counting structures used to determine this curve [43–46]. The slope m of this S–N
method was developed in this study. This algorithm is able to determine curve is 3 for cycles from 105 to 107 and 5 for cycles from 107 to 109.
the exact values of each load class and create a 3D rainflow matrix (see This choice is considered to be a simplification for easier presentation of
Fig. 5), making it possible to precisely determine the half-cycle values the possibilities of this complex analysis. The S–N curves obtained from
for each load class. the laboratory testing of samples taken from the investigated steel
For improved interpretation of the values from the rainflow matrix structure are expected to nearly match these curves. The curve shown in
and their subsequent implementation into the calculations, the tabular Fig. 7 corresponds to the standard curve of Category 71 [24].
form is convenient; Table 1 contains the mean, range, and count values A fatigue calculation was performed for each load combination
for each load class. The range value indicates the variance within each corresponding to the maximum and minimum axial forces for a given
load class. Individual classes are indicated by letters from A to J and by amplitude. Therefore, each combination was considered a non-pro-
numbers from 1 to 10. Because the analysis deals with axial forces, the portional load. The evaluation of stress components was considered for
values in this table are related only to the load, not the moment. the equivalent von Mises stress. The resulting equivalent alternating
In Table 1 there are 53 non-zero combinations from A1 to I6. From stresses were evaluated for each finite element in the model and the
each combination, it is possible to determine two load cases, making a damage from individual load combinations was calculated. Figs. 8 and
total of 106 load cases. To calculate the bounding values of the axial 9 show the equivalent alternating stresses for Load Case I6 N+ and
force in the selected diagonal, two simple equations were used: Load Case I6 N−. The fatigue damage was obtained by comparing the
stress values from the positive and negative cases.
Ni + = Mean + Range /2 (1)

Ni − = Mean − Range /2 (2)

3.2. Cumulative Palmgren–Miner rule
where Ni + is the positive axial force and Ni- is the negative axial force.
For example, load combination I6 has a mean of 26.06 kN and a When the individual values of the load combinations are known, it
range of 54.28 kN, so according to Eqs. (1) and (2), Load Case I6N+ = is possible to conduct a numerical fatigue analysis of the selected
80.35 kN and Load Case I6N- = −28.22 kN. All resulting positive and construction detail. Thus, a prediction of fatigue damage was calculated
negative load cases are shown in Table 2. for each individual combination. These values are the inverse of the

4
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 5. 3D rainflow matrix load history data for one year.

predicted residual lifetime. Subsequently, all particular damages were respectively, and were then used to determine their residual lifetimes.
summed by the cumulative Palmgren–Miner rule [25,26] to determine From the distribution of fatigue damage in the model, it is possible
the predicted residual lifetime of each finite element in the model. The to identify points where there is a high probability of fatigue crack
cumulative Palmgren–Miner rule can be expressed by: initiation. Damage was particularly observed at the edges of the holes,
n1 n n where stress concentrations are present. In Figs. 10–12, the maximum
D= + 2 + ⋯+ k , damage is indicated for each individual detail.
N1 N2 Nk (3)
To determine the residual lifetimes of each component of the FEM
where D is the resulting cumulative damage, ni is the number of cycles model, the selected S–N curve was used with the cumulative damage to
for each stress range (load combination), and Ni is the number of cycles obtain the results given in Table 3. The most vulnerable part of the
of residual lifetime for each stress range. target detail considering fatigue damage can be observed to be the di-
The resulting cumulative damage was transferred back to the nu- agonal, which has an estimated life span of 249 years, and so a re-
merical model for graphical display, allowing for the evaluation of the maining lifetime of 159 years. In contrast, the rivets and the connection
dangerous parts of the detail considering fatigue damage from a given plate have considerably longer remaining lifetimes.
load. These dangerous parts were generally found to be located at stress
concentrations, especially at the locations of the holes where the rivets
deform. The maximum cumulative fatigue damages to the gusset plate,
diagonals, and rivets were calculated, shown in Figs. 10–12,

Table 1
The tabular record of the rainflow matrix contains the mean, range, and count of axial force values.
Range [kN]

3.19 9.58 15.97 22.35 28.74 35.13 41.51 47.90 54.28 60.67

Mean [kN] −26.49 233 13 1

−15.98 298 378 145 8 1 2
−5.47 14,483 96 5586 1777 32 3
5.04 24,115 834 62 11 17 17 36 4
15.55 16,259 33 111 332 318 419 2719 3025 911 5
26.06 923 5 111 474 5336 1782 37 229 360 6
36.57 146 89 126 239 70 117 80 7
47.08 269 101 31 15 9 1 8
57.59 115 18 2 9
68.10 10
A B C D E F G H I J

5
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Table 2
Load values for individual combinations obtained from the rainflow matrix, arranged in order of descending cycle count.
Load case Cycle count Ni+ [kN] Ni-[kN] Load case Cycle count Ni+ [kN] Ni-[kN]

A4 24,115 8.23 1.85 A9 115 60.79 54.40

A5 16,259 18.74 12.36 C5 111 31.52 −0.42
A3 14,483 −2.28 −8.67 C6 111 42.03 10.09
C3 5586 10.49 −21.44 B8 101 56.66 37.50
E6 5336 54.80 −2.68 B3 96 4.11 −15.05
H5 3025 63.45 −32.35 B7 89 46.15 26.99
G5 2719 57.06 −25.96 G7 80 78.08 −4.94
F6 1782 61.19 −9.07 E7 70 65.31 7.83
D3 1777 16.88 −27.83 C4 62 21.00 −10.93
A6 923 29.25 22.87 G6 37 67.57 −15.45
I5 911 69.83 −38.74 G4 36 46.55 −36.47
B4 834 14.62 −4.54 B5 33 25.13 5.97
D6 474 48.41 3.71 E3 32 23.27 −34.21
F5 419 50.68 −19.58 C8 31 63.05 31.12
B2 378 −6.40 −25.56 B9 18 67.17 48.01
I6 360 80.35 −28.22 E4 17 33.78 −23.70
D5 332 37.90 −6.80 F4 17 40.16 −30.09
E5 318 44.29 −13.19 D8 15 69.44 24.73
A2 298 −12.79 −19.18 B1 13 −16.91 −36.07
A8 269 50.28 43.89 D4 11 27.39 −17.31
D7 239 58.92 14.22 E8 9 75.82 18.34
A1 233 −23.30 −29.69 D2 8 6.37 −38.34
H6 229 73.96 −21.84 B6 5 35.64 16.48
A7 146 39.76 33.38 C9 2 73.56 41.63
C2 145 −0.02 −31.95 E2 1 12.76 −44.72
C7 126 52.54 20.61 F8 1 82.21 11.96
F7 117 71.70 1.45

Fig. 6. Selected diagonal connection (a) detail and (b) numerical model.

1000
ǻınet in N/mm2

100 m=3
71
52 m=5

29

10
1.E+05 1.E+06 2.E+06 5.E+06 1.E+07 1.E+08 1.E+09
N [cycles]
Fig. 7. S–N curve for the fatigue assessment of old riveted steel bridges [4,42].

6
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 8. Equivalent alternating stress (Δσi) on rivet surface, Load Case I6+.

Fig. 9. Equivalent alternating stress (Δσi) on rivet surface, Load Case I6-.

4. Conclusions that there is substantial remaining fatigue life in the structure. As could
be expected, the rivets showed the highest resistance of the considered
The objective of this study was to propose and evaluate an approach components.
for the fatigue damage analysis of an existing steel truss structure The method of conducting a stochastic analysis based on the Monte
supporting an overhead crane track. To demonstrate the proposed ap- Carlo simulation method with random input parameters offers inter-
proach, a fatigue analysis of the connection details of the diagonal to esting possibilities. In the case of appropriately input data, not only can
the bottom chord of the steel truss structure was presented in this the load history be described, but the future effects of loads can be
paper, considering three elements of this detail: the diagonal, rivets, predicted. The application of the proposed procedure demonstrated
and gusset plate. It should be noted that the procedure detailed in this that the common fatigue design curve for riveted details can be used to
paper relies upon several simplification procedures that were necessary inform a safe estimation of the remaining fatigue life of old riveted steel
to reduce computational complexity, including material assumptions. crane structures still in service today. This investigation has determined
Based on the results of numerical fatigue modelling, it was determined that the maximum stress ranges seldom, if ever, exceed the fatigue limit

7
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Fig. 10. Diagonal cumulative damage corresponding to one year of operation – 0.00401.

Fig. 11. Rivet cumulative damage corresponding to one year of operation – 0.000447.

Fig. 12. Gusset plate cumulative damage corresponding to one year of operation – 0.00312.

8
P. Lehner, et al. International Journal of Fatigue 128 (2019) 105190

Table 3 fatigue life prediction. Probabilistic Eng Mech 2011;26:148–56. https://doi.org/10.

Connection part lifetimes based on cycles required for fatigue crack initiation. 1016/j.probengmech.2010.11.001.
[16] Kala Z. Sensitivity and reliability analyses of lateral-torsional buckling resistance of
Connection part No. of cycles Predicted life span Remaining lifetimea steel beams. Arch Civ Mech Eng 2015;15:1098–107. https://doi.org/10.1016/j.
acme.2015.03.007.
Diagonal 2.0656E + 07 249 years 159 years [17] Ghosh P, Konečný P, Lehner P, Tikalsky PJ. Probabilistic time-dependent sensitivity
Rivets 1.8557E + 08 2237 years 2147 years analysis of HPC bridge deck exposed to chlorides. Comput Concr 2017;19:305–13.
Gusset plate 2.6628E + 07 321 years 231 years https://doi.org/10.12989/cac.2017.19.3.305.
[18] Králik J. Experimental and numerical analysis of the hermetic tightness of NPP
bubble tower structure. Procedia Eng 2017;190:472–9. https://doi.org/10.1016/j.
a
Remaining life = Predicted life – 90 year structure age.
proeng.2017.05.366.
[19] Przekop A, Rizzi SA. A reduced order method for predicting high cycle fatigue of
for riveted details of this era. Therefore, in this case, the fatigue re- nonlinear structures. Comput Struct 2006;84:1606–18. https://doi.org/10.1016/j.
sistance of the steel truss structure is sufficient. Notably, this suggests compstruc.2006.01.015.
[20] Seitl S, Miarka P, Kala Z, Klusák J. Effect of rivet holes on calibration curves for
that the effects of fatigue damage accumulation in an old structure edge cracks under various loading types in steel bridge structure. Procedia Struct
during its service life are negligible when compared with the effects of Integr 2017;5:697–704. https://doi.org/10.1016/j.prostr.2017.07.046.
corrosion losses on the durability of the structure. The method de- [21] Krejsa M, Seitl S, Brozovsky J, Lehner P. Fatigue damage prediction of short edge
crack under various load: Direct Optimized Probabilistic Calculation. Procedia
monstrated in this study will likely be of considerable use to engineers Struct Integr 2017;5:1283–90. https://doi.org/10.1016/J.PROSTR.2017.07.107.
evaluating the safety and remaining functionality of any ageing steel [22] Schott G, Donat B, Schaper M. The consecutive Wöhler curve approach to damage
structure connected using rivets. accumulation. Fatigue Fract Eng Mater Struct 1996;19:373–85. https://doi.org/10.
1111/j.1460-2695.1996.tb00974.x.
[23] Susmel L. On the use of the modified Wöhler Curve Method to estimate notch fa-
Acknowledgments tigue limits. Mater Test 2006;48:27–35. https://doi.org/10.3139/120.100709.
[24] EN 1993-1-3:2009 BE. Eurocode 3: Design of steel structures—Parts 1–9: Fatigue.
CEN 2011.
Funding. [25] Miner MA. Cumulative damage in fatigue. Am Soc Mech Eng: J Appl Mech
This research was conducted as a part of the research project GACR 1945;12:159–64. https://doi.org/10.1007/978-3-642-99854-6_35.
17-01589S “Advanced computational and probabilistic modelling of [26] Hashin Z. A reinterpretation of the Palmgren-Miner rule for fatigue life prediction. J
Appl Mech 1980;47:324–8. https://doi.org/10.1017/CBO9781107415324.004.
steel structures taking into account fatigue damage” supported by the
[27] Lehnera P, Křivý V, Krejsa M, Pařenica P, Kozák J, Lehner P. Stochastic service life
Czech Grant Agency and thanks to the financial support provided to the prediction of existing steel structure loaded by overhead cranes. Procedia Struct
VSB-Technical University of Ostrava by the Czech Ministry of Integr 2018;13:1539–44. https://doi.org/10.1016/J.PROSTR.2018.12.314.
Education, Youth, and Sports from the budget for the conceptual de- [28] Krejsa M, Brozovsky J, Mikolasek D, Lehner P, Parenica P. Using DOProC method in
reliability assessment of steel elements exposed to fatigue. MATEC Web Conf
velopment of science, research, and innovations for the year 2019. 2017;107:00046. https://doi.org/10.1051/matecconf/201710700046.
[29] Kracík J, Strnadel B. A statistical model for lifespan prediction of large steel
Declarations of Competing Interest structures. Eng Struct 2018;176:20–7. https://doi.org/10.1016/j.engstruct.2018.
08.065.
[30] Maljaars J, Bonet E, Pijpers RJM. Fatigue resistance of the deck plate in steel or-
None. thotropic deck structures. Eng Fract Mech 2018;201:214–28. https://doi.org/10.
1016/j.engfracmech.2018.06.014.
[31] Yang DY, Frangopol DM. Probabilistic optimization framework for inspection/re-
References pair planning of fatigue-critical details using dynamic Bayesian networks. Comput
Struct 2018;198:40–50. https://doi.org/10.1016/j.compstruc.2018.01.006.
[1] Vičan J, Gocál J, Odrobiňák J, Koteš P. Analysis of existing steel railway bridges. [32] Georgioudakis M, Lagaros ND, Papadrakakis M. Probabilistic shape design opti-
Procedia Eng 2016;156:507–14. https://doi.org/10.1016/j.proeng.2016.08.328. mization of structural components under fatigue. Comput Struct 2017;182:252–66.
[2] Hobbacher AF. Remaining life assessment, repair and strengthening of existing steel https://doi.org/10.1016/j.compstruc.2016.12.008.
structures. In: Proc. int. conf. adv. weld. sci. technol. constr. energy transp. AWST [33] Zhang D, Hong J, Ma Y, Chen L. A probability method for prediction on High Cycle
2010, held Conj. with 63rd Annu. Assem. IIW 2010; 2010. p. 63–7. Fatigue of blades caused by aerodynamic loads. Adv Eng Softw 2011;42:1059–73.
[3] Ye XW, Su YH, Han JP. A state-of-the-art review on fatigue life assessment of steel https://doi.org/10.1016/j.advengsoft.2011.07.010.
bridges. Math Probl Eng 2014;2014:1–13. https://doi.org/10.1155/2014/956473. [34] Zhang YH, Wang LJ. Stochastic analysis of fatigue crack growth for metallic
[4] Kühn B, Lukić M, Nussbaumer A, Günther H, Helmerich R, Herion R, et al. New JRC- structures. Int J Fract 1994;66:197–203. https://doi.org/10.1007/BF00042583.
ECCS-Joint Report on Assessment of existing steel structures: recommendations for [35] Scia Engineering, online documentation n.d. < https://www.scia.net/en/support/
estimation of remaining fatigue life. Steel Constr 2008;1:94–5. https://doi.org/10. downloads/scia-engineer-manuals-tutorials > .
1002/stco.200890018. [36] Seitl S, Miarka P, Janosik P. Comparison of selected mechanical properties of old
[5] Åkesson B. Fatigue life of riveted steel bridges. London: CRC Press; 2010. steel and S235 and S355 grades. Model. Mech. 23 and 24 May 2019, Ostrava, Czech
doi:10.1201/b10429. Republic; 2019.
[6] Anderson T. Fracture mechanics—fundamentals and application. Boca Raton: [37] Casciati F, Fuggini C. Monitoring a steel building using GPS sensors. Smart Struct
Taylor & Francis Group; 2005. Syst 2011;7:349–63. https://doi.org/10.12989/sss.2011.7.5.349.
[7] Carpinteri A. Handbook of fatigue crack propagation in metallic structures. Elsevier; [38] Punurai W, Davis P. Prediction of asbestos cement water pipe aging and pipe
1994https://doi.org/10.1016/C2009-0-13210-5. prioritization using Monte Carlo simulation. Eng J 2017;21:1–13. https://doi.org/
[8] EN 1991-3 July. Eurocode 1. Actions on structures—Part 3: Actions induced by 10.4186/ej.2017.21.2.1.
cranes and machinery. CEN 2006; 3. https://doi.org/10.1029/JA088iA10p07887. [39] Metropolis N, Ulam S. The Monte Carlo Method. J Am Stat Assoc 1949;44:335–41.
[9] EN 1993-1-1. Eurocode 3. Design of steel structures—General rules and rules for https://doi.org/10.2307/2280232.
buildings. CEN 2005. https://doi.org/10.1017/CBO9781107415324.004. [40] Amzallag C, Gerey JP, Robert JL, Bahuaud J. Standardization of the rainflow
[10] BS EN 1993-1-8:2005. Eurocode 3. Design of steel structures—Parts 1–8: Design of counting method for fatigue analysis. Int J Fatigue 1994;16:287–93. https://doi.
joints. CEN 2005; 8:135. org/10.1016/0142-1123(94)90343-3.
[11] Sadowski AJ, Rotter JM, Reinke T, Ummenhofer T. Statistical analysis of the ma- [41] ANSYS. ANSYS Meshing User’s Guide. ANSYS User Guid 2016; 15317:514.
terial properties of selected structural carbon steels. Struct Saf 2015;53:26–35. [42] Kühn B. Assessment of existing steel structures - recommendations for estimation of
https://doi.org/10.1016/j.strusafe.2014.12.002. the remaining fatigue life. Procedia Eng 2013;66:3–11. https://doi.org/10.1016/j.
[12] Zhang R, Mahadevan S. Fatigue reliability analysis using nondestructive inspection. proeng.2013.12.057.
J Struct Eng 2001;127:957–65. [43] Fisher JW, Mertz DR, Zhong A. NCHRP Report 267: Steel Bridge Members Under
[13] Wahab MMA, Ashcroft IA, Crocombe AD, Smith PA. Numerical prediction of fatigue Variable Amplitude Long Life Fatigue Loading; 1983. p. 26.
crack propagation lifetime in adhesively bonded structures. Int J Fatigue [44] Fischer JW. Fatigue and fracture in steel bridges. New York: Wiley; 1984. p. 315.
2002;24:705–9. https://doi.org/10.1016/S0142-1123(01)00173-6. [45] Wilson WM, Thomas FP. Fatigue tests of riveted joints: a report. University of
[14] Torres MA, Ruiz SE. Structural reliability evaluation considering capacity de- Illinois at Urbana Champaign, College of Egnineering, Engineering Experiment
gradation over time. Eng Struct 2007;29:2183–92. https://doi.org/10.1016/j. Station; 1938. p. 173–95.
engstruct.2006.11.014. [46] Moses F, Schilling CG, Raju KS. Fatigue evaluation procedures for steel bridges.
[15] Xiang Y, Liu Y. Application of inverse first-order reliability method for probabilistic Washington DC: NCHRP TRB Research Program Report; 1987.