Ce Papers - 2018 - D Angelo - Using Re Calibrated Design S N Curves For Fatigue Assessment of Road Bridges

25097075, 2017, 5-6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/cepa.615 by CAPES, Wiley Online Library on [28/02/2023].
See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Luca D’Angelo, Alain Nussbaumer
Using re-calibrated design S-N curves for fatigue assessment of

road bridges
Abstract
Fatigue design of road bridges according to Eurocode standards can be based on damage accumulation,
provided that information about loading spectrum is available. Characteristic values of fatigue
resistance and load effects as well as partial safety factors are introduced in the design equation in
order to achieve a target reliability level. Partial safety factors are used to obtain the fatigue design
S-N curves from the characteristic S-N curves. New probabilistic framework proposed by authors
for calibration of fatigue partial safety factors allows for re-definition of Eurocode-based design S-N
curves. The new framework improves the confidence in the reliability level represented by the design
S-N curves. In this paper a real case study showing the impact of re-calibration of design S-N curves
on fatigue design of a road bridge is presented. Importance of choice of design S-N curves is assessed
both in terms of allowable traffic volume and of bridge fatigue design life.
1 Introduction
When information about the loading spectrum is available, fatigue design of road bridge components
can be based on damage accumulation. In Eurocode 1993-1-9 [4] the fatigue design based on damage
accumulation has to meet the following criterion:
Ntot
X ni
Dd = ≤ 1.0 (1)
i
Ni
where: ni is the number of cycles corresponding to the design loading stress range γFf · Si · Ni is
the endurance to failure computed with the factored Sc /γMf − N curve. The factored Sc /γMf − N
curve obtained from the characteristic Sc - N curve, is referred to as design S-N curve. For safe life
design method and high consequence of failure, Table 3.1 [4] recommends to use γMf = 1.35, having
assumed γFf = 1.00. Furthermore, when information about load spectrum is not available fatigue
design can be based on CAFL (γFf · Smaxc ≤ C AF L/γMf ), or on constant amplitude equivalent stress
range (γFf · SE,2 ≤ Sc /γMf ).
The definition of the partial resistance factor γMf in Eurocode 1993-1-9 (Section 1.4, pp. 9, [4]) is
ambiguous because γMf is strictly defined for a fatigue strength at 2 · 106 cycles, but it is applied
by extension for fatigue strengths at any number of cycles. Sedlacek et al. addressed the issue of
calibration of fatigue design partial safety factors in [5]. Nevertheless, a proof of rigorous calibration
of Eurocode-based partial safety factors is not available and the real reliability level, β, corresponding
to these values is still under debate. Therefore, the use of Eurocode-based fatigue partial safety factors
© Ernst & Sohn Verlag für Architektur und technische Wissenschaften GmbH & Co. KG, Berlin · ce/papers 1 (2017), No. 5 & 6
https://doi.org/10.1002/cepa.615 wileyonlinelibrary.com/journal/cepa 443
25097075, 2017, 5-6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/cepa.615 by CAPES, Wiley Online Library on [28/02/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
L. D’Angelo et al.: Using re-calibrated design S-N curves for fatigue assessment of road bridges
in combination with Eurocode-based characteristic S-N curves, which are themselves affected by lack
of accuracy in the high cycle fatigue (HCF) region (N > 107 cycles) [3], could lead to extremely
unreliable design of road bridge fatigue deciding components. In [3] the authors have presented a new
probabilistic approach for estimation of S-N curves, referred as ML-MCS1 method, which improves
the accuracy of characteristic S-N curves in HCF region.
The use of ML-MCS approach S-N curves in combination with partial safety factors, which have
been re-calibrated by using the probabilistic framework presented by the authors in [1], allows for
improving the reliability of fatigue checks and design S-N curves. This paper presents a real case
study showing the impact of re-calibration of design S-N curves on fatigue design of road bridges.
2 Re-calibrated fatigue design curves of welded joints
The characteristic S-N curves for two details (welded cover plate, welded in-plane gusset) were
estimated by using ML-MCS approach [3]. Since the ML-MCS approach provides estimate of the
characteristic value of critical damage, d c,0.5 , different than 1.0 for both details (see Table 2 in [1]),
characteristic S-N curves were re-scaled at d c,0.5 = 1.0 (see re-scaling method in [3]) for direct
comparison with Eurocode. ML-MCS-based design S-N curves were obtained from ML-MCS-based
characteristic S-N curves by using re-calibrated fatigue resistance partial safety factors (γMf = 1.37
for the cover plate and γMf = 1.45 for the in-plane gusset, see Table 3 in [1]).
ML-MCS-based design S-N curves for both details are presented in Fig. 1, where also comparison to
Eurocode-based design S-N curves is provided. It is noted that Eurocode-based design S-N curves are
obtained by classifying the cover-plate detail and the in-plate gusset detail respectively as FAT45 and
FAT40, and by using γMf = 1.35 (safe life design - high consequences of failure).
3 Case study
Design curves presented in Section 2 are used for assessing the fatigue design life of the Venoge bridge,
a composite steel concrete bridge on the Swiss A1 highway (see Fig. 2(a)). This case study allows to
evaluate the impact of re-calibration of design S-N curves on fatigue design, when information about
loading spectrum is available and damage accumulation verification criterion is used.
3.1 Venoge highway bridge
The Venoge bridge consists of two identical independent bridges (one per direction) with a 219 meter
length divided in four spans (52-60-55-52 meters). Each bridge was constructed in 1961 with two
girders and then enlarged in 1995 to four girders (see Fig. 2(b)).
1
The proposed approach uses combination of Maximum Likelihood (ML) method and Monte Carlo Simulations (MCS)
method
444
Fig. 1: Comparison of VA design S-N curves: (a) cover plate; (b) in-plane gusset
Fig. 2: Venoge bridge: (a) Ecublens, Switzerland; (b) Section and lane configuration
Two critical details were considered for the fatigue design life assessment:
1. Welded cover-plate, on the lower flange of the enlargement left bridge at first mid-span;
2. In-plane welded gusset, on the lower flange of the enlargement left bridge at first mid-span. This
detail does not exist in reality and is considered for academic purpose only.
Both fatigue critical details are located at the same location on the enlargement left bridge, due to the
position of the slow lane (heavy traffic) location (see Fig. 2(b)). In order to define the stress range
loading spectrum, a Rayleigh probability distribution was fitted to the observed stress range spectrum
at the critical section in the year 2006 [2]. The obtained Rayleigh stress range spectrum, having scale
parameter equal to 5.3, is shown in Fig. 3(a).
445
Fig. 3: Loading model: (a) Stress range spectrum and traffic composition (according to 2006 WIM measure-
ments, see [2]); (b) Comparison of design values of nveh
3.2 Design fatigue life assessment
The design value of allowed number of cycles, nd , was computed, for the two considered critical
details, using following equation:
d c,0.5
nd = Z ∞ (2)
f S (s)
ds
0 N (s)
where d c,0.5 is the characteristic value of critical damage sum, which is equal to 1.0;
f S(s) is the Rayleigh loading stress range spectrum;
Sc
N (s) is the endurance obtained from the factored − N curve (design S-N curve).
γMf
Both ML-MCS-based design S-N curves and Eurocode-based design S-N curves were used for com-
parison. It is recalled that ML-MCS-based design S-N curves (presented in Section 2) were obtained
by using ML-MCS-based characteristic S-N curves and re-calibrated partial safety factors.
The design value of allowed number of heavy vehicles, nveh , was computed as follows :
nd
nveh = X6 (3)
i · pi
i=2
where i is the number of heavy vehicle axles;

pi is the percentage of vehicles having j axles, according to Fig. 3(a).
446
Hence, the design life of the bridge in years, Td , was computed as follows:
ξ
Td = (4)
nveh
where ξ is the traffic volume, expressed in number of heavy vehicles per year.
4 Results
Results of the design life assessment of the Venoge bridge are presented both in terms of design value
of allowed number of heavy vehicles, nveh , and of bridge design life, Td . The values of nveh and Td
obtained by using ML-MCS-based characteristic S-N curves and re-calibrated partial safety factors
are referred to respectively as ML-MCS-based nveh and ML-MCS-based Td . The comparison between
the ML-MCS-based design value of nveh and the Eurocode-based design value of nveh is presented in
Fig. 3(b) for the two considered details. For the cover-plate detail, the Eurocode approach provides
slightly higher estimate of design value of nveh with respect to the Eurocode approach. However, both
approaches predict a fatigue design life higher than 100 years, for a yearly traffic volume, ξ ≤ 2 · 106 .
For the in-plane gusset detail, the Eurocode provides considerably lower estimate of design value of
nveh and does not achieve the required value of nveh corresponding to ξ ≤ 2 · 106 .
Fig. 4 shows the relationship between the design life in years and the yearly traffic volume, ξ, for
both critical details. With regard to the cover plate detail, for ξ ≤ 2 · 106 , Eurocode approach gives a
design life equal to 142 years while ML-MCS approach gives a design life equal to 118 years. With
regard to the in-plane gusset detail, for ξ ≤ 2 · 106 , the Eurocode gives a design life equal to 78 years,
which is about 60% of the design life predicted with ML-MCS approach (125 years). Finally, for
ξ ≤ 4 · 106 (which is the traffic volume measured by WIM on Venoge bridge in 2006, see Table 1
in [2]) Td decreases, for the cover-plate from 710 years (Eurocode) to 592 years (ML-MCS); for the
in-plane gusset it increases from 390 years (Eurocode) to 627 years (ML-MCS).
5 Discussion and conclusions

The new ML-MCS-based framework for calibration of partial safety factors presented by authors
in [1] allows for re-definition of Eurocode design S-N curves. The presented framework improves
consistency in achieving target levels of safety when these S-N curves are used for fatigue design of
road bridges. Authors showed that the partial safety factors should be differentiated in function of the
verification scheme. For example, the use of the ML-MCS framework for calibration of partial safety
factors for the two fatigue sensitive details considered in the Venoge bridge case study shows that the
verification scheme based on CAFL gives higher values of γMf with respect to the verification scheme
based on damage sum (1.80 vs 1.37 for the cover plate, 2.27 vs 1.45 for the in-plane gusset, see Table
3 in [1]).
Fatigue design of the Venoge bridge based on real traffic-induced load spectrum shows significant
differences between the ML-MCS approach and the Eurocode approach in terms of predicted fatigue
lives. When cover-plate detail is considered critical, for a yearly traffic volume equal to 2 · 106 ,
447
Fig. 4: Relationship between design fatigue life and traffic volume, (a) cover-plate; (b) in-plane gusset
ML-MCS approach estimates design life that is 24 years shorter than design life estimated by Eurocode
approach, but both approaches predict a design life that is longer than 100 year. Nevertheless, the
difference is more significant when the in-plane gusset detail is considered critical: for a yearly traffic
volume equal to 2 · 106 the Eurocode approach predicts a design life with does not achieve the 100-year
required value, while the ML-MCS approach predicts a design life equal to 125 years.
References
[1] L. D’Angelo and A. Nussbaumer. “New framework for calibration of partial safety factors for fatigue design”. In:
Journal of Constructional Steel Research (In Review).
[2] L. D’Angelo and A. Nussbaumer. “Reliability based fatigue assessment of existing motorway bridge”. In: Structural
Safety 57 (2015). url: http://dx.doi.org/10.1016/j.strusafe.2015.07.001.
[3] L. D’Angelo and A. Nussbaumer. “Estimation of fatigue S-N curves of welded joints using advanced probabilistic
approach”. In: International Journal of Fatigue 97 (2017), pp. 98–113.
[4] DIN EN 1993-1-9. Eurocode 3: Design of steel structures - Part 1-9: Fatigue. CEN (European Committee for
Standardization), 2005.
[5] G. Sedlacek, H. Eisel, W. Hensen, B. Kühn, and M. Paschen. Leitfaden zum DIN-Fachbericht 103 Stahlbrücken.
Ernst & Sohn, 2003.
448

Ce Papers - 2018 - D Angelo - Using Re Calibrated Design S N Curves For Fatigue Assessment of Road Bridges

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ce Papers - 2018 - D Angelo - Using Re Calibrated Design S N Curves For Fatigue Assessment of Road Bridges

Uploaded by

Copyright:

Available Formats

25097075, 2017, 5-6, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/cepa.615 by CAPES, Wiley Online Library on [28/02/2023].

Using re-calibrated design S-N curves for fatigue assessment of

2 Re-calibrated fatigue design curves of welded joints

3.1 Venoge highway bridge

3.2 Design fatigue life assessment

where i is the number of heavy vehicle axles;

5 Discussion and conclusions

You might also like