Professional Documents
Culture Documents
Abstract
Distribution factors (DFs) for one typical cross-section as specified in the AASHTO LRFD specification can be varied
when the bridge parameters such as span length, loading lanes and skew are changed. The diversity between design
and actual DFs may be varied as the bridge parameters changed. To study this diversity, this paper presents an evalua-
tion of lateral load DFs for prefabricated hollow slab bridges. The response of the bridge was recorded during the field
test. This field test was divided into two stages: a concentrated force loading test on the prefabricated girder that set-
tled on the bridge supports before the girders were connected transversely and a vehicle loading test after the girders
were connected transversely. The instruments used to record the response of the bridge were strain gauges and dial
indicators. The measured data in the multi-stages of the field test could be used to calibrate the support condition
of the bridge and transverse connection between adjacent girders in the finite element model (FEM) using beam
and plate elements. From the FEM, DFs for this hollow slab bridge were determined and compared with the DFs in
the AASHTO LRFD specification. A parametric study using the calibrated FEM was then used to investigate the effect
of various parameters including span length, skew and bridge deck thickness on the DFs. It was found that AASHTO
LRFD specification is conservative compared with the analysis in the FEM, while this conservatism decreased as the
span length and skew of the hollow slab bridge increased.
Keywords: hollow slab bridge, field test in multi-stage, distribution factor, support condition, transverse connection
© The Author(s) 2020. This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and
the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material
in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material
is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the
permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativeco
mmons.org/licenses/by/4.0/.
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 2 of 13
joint to attract load together. For accurate analysis, trans- bridge response under the vehicle loading test. In addi-
verse connection between adjacent girders and support tion, a parametric study was conducted to investigate the
condition are two important parameters that need to be influence of various parameters on the DFs of this hollow
calibrated in the FEM for this type of bridge. Multi-stage slab bridge including span length, skew and bridge deck
loading tests on the constructing bridge were an effective thickness. Finally, the effect of these parameters on the
way to calibrate these two parameters, respectively. DFs was quantified compared with the AASHTO LRFD
Many studies investigated the effect of secondary ele- specification.
ments such as parapets, sidewalk and diaphragm on the
DFs and concluded that AASHTO LRFD specification 2 Bridge Description (STRUKTUR)
was conservative without considering these secondary A constructing hollow slab bridge, which is located
elements (Mabsout et al. 1997; Conner and Huo 2006; on WeiXu expressway in Henan province, China, was
Namy et al. 2015). Geometric dimensions of the bridge selected for this study. The superstructure was sim-
remained unchanged in these studies. However, geo- ply supported prefabricated concrete hollow slab gird-
metric dimensions of the bridge were the variables of ers, which was shown in Fig. 1. The span of bridge was
the DF calculation formula in AASHTO LRFD specifica- 13 m in length with a 12.412 m clear span. The width
tion. For a certain kind of bridge, geometric dimensions of the bridge was 13.5 m and skewed at 25 . The hollow
could be varied in an appropriate range. As a result, the slab girders had a height of 0.7 m and a width of 0.99 m,
designed DFs for different geometric dimensions were and covered 0.1 m thick layer of cast-in-place reinforced
also changed in a certain range. So the effect of geomet- deck. The bridge girders were prefabricated and the deck
ric dimensions, which were varied in a rational range, on was cast. The connection of structures in the bridge
the DFs should be figured out first before discussing the superstructure is shown in Fig. 2. The concrete of hinge
effect of secondary elements. joints and bridge deck was cast at the same time. The
The aim of this study is to identify the supporting concrete used in the slab girders, hinge joints and bridge
condition and transverse connection of the hollow slab deck had a designed compressive strength of 50 MPa and
bridge in the field tests using multi-stage construction modulus of elasticity of 3.45 × 104MPa. The superstruc-
of full-scale test in situ. First, a concentrated loading ture was composed of 13 girders, in which the leftmost
test was conducted on the prefabricated girder that set- girder was denominated G1 while the rightmost girder
tled on the supports before the girders were connected denominated G13.
transversely. The response of the prefabricated girder
could be used to calibrate the support condition of the 2.1 Concentrated Force Loading Test (PEMBEBANAN)
hollow slab in the FEM. Second, a vehicle loading test 2.1.1 General
was performed on the bridge after the prefabricated gird- This loading phase was performed after the prefabricated
ers were transversely connected. Transverse connection girders were installed and the adjacent girders had not
between adjacent girders could be calibrated with the been connected transversely. Before this loading test,
13000
294 294
inside edge of the parapet 100mm cast deck inside edge of the parapet
700
Fig. 5 Measurement instrument position (units: mm). Note: Di: strain gauges; DNi: deflection gauges.
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 5 of 13
Hinged Bearing Mid Section Roller Bearing eled by the beam element while the bridge deck was mod-
69.77±N1 69.77±N1 40.±N1
G1 eled by the plate elements. In ANSYS program, beam188
G2 elements were used to model the main girder and virtual
LCA G3
G4
transverse beam, shell63 element were used to model
69.77±N1 69.77±N1 40.±N1
G5 bridge deck. For the beam element in MIDAS CIVIL and
1000×11
13000
the bridge supplemented with engineering judgment(Seo
Fig. 6 Loading vehicle measurement (units: mm). et al. 2015). The concrete grade using for the girder, hinge
joint and bridge deck is C50, The modulus of elastic-
ity of the concrete is 3.45 × 104MPa. As specified in the
AASHTO LRFD specification, live load deflection is a
4 Finite‑Element Model service issue, not a strength issue. So in the calculation
In order to assure the correction of the finite element of the FEM, the nonlinearity of the concrete materials
analysis (FEA), two different FEA programs—MIDAS was not considered. Figure 9 shows the FE models of the
CIVIL and ANSYS were conducted. In MIDAS CIVIL, bridge.
1000
7000 LC E
5000 LC D
3000 LC C
1000 LC B
LC A
G1 G2 G3 G4 G5 G6 G7 G8 G9 G10 G11 G12 G13
Fig. 8 Instrument arrangement in the vehicle loading test. Note: Pi: strain point, PNi: deflection point.
MIDAS CIVIL/ANSYS Ktop (kN/mm) 630 770 770 770 770 770 770 770 770 770 770 770 630
Kbottom (kN/mm) 630 770 770 770 770 770 770 770 770 770 770 770 630
MIDAS CIVIL δp(%) 12.97 5.88 5.32 6.46 8.06 9.75 11.37
ρ 1 1 1 1 1 1 1
ANSYS δp(%) 9.32 8.84 6.91 9.17 10.03 10.55 11.36
ρ 1 1 1 1 1 1 1
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 8 of 13
DFi = ωi Σ
13
× Nvehicles
were systematically varied to consider all possible load
ωi (4) combinations. In all cases the vehicles were placed fol-
i=1 lowing the 25 skew.
where, DFi is load distribution factor of ith girder; ωi is Interior and exterior distribution factors are calculated
maximum deflection of ith girder; Nvehicles is the number separately in the AASHTO specification. As such, the
of vehicles on the bridge. maximum distribution factor for the one, two and three
It can be concluded that the girder which suffered loaded lanes for both the interior and exterior girders
the traffic load directly experiences the greatest deflec- were calculated using the maximum deflection of differ-
tion, whereas girders far away from the loading position ent girders. Calculated moment DFs for the interior and
experience less. DFs for exterior girder were 52% greater exterior girders for each load cases are listed in Table 3.
than the DFs for the most heavily loaded interior girder, DFs calculated by the AASHTO LRFD were also listed in
the reason is that in this test there are no railings which Table 3.
usually offered a significant stiffening contribution to the Table 3 shows that using AASHTO LRFD specifications
exterior girders and the exterior girder is unable to dis - resulted in a conservative DFs estimate. The AASHTO
tribute load to an interior member. DFs were much higher than those calculated using the
Figure 13 shows the deflections at mid-span on each calibrated FEM for exterior girders. For one lane loaded,
girder under different loading conditions. It can be the AASHTO LRFD DFs were 1.5 times larger than that
seen from Fig. 13 that the deflection curves were nearly of the FEM DFs for exterior girders and 1.2 times larger
overlapped for the five loading conditions. It can be for interior girders. For multilane loaded, the AASHTO
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 9 of 13
LRFD DFs were 1.4 times larger than that of the FEM was filled with vehicles on the lanes that it can be divided
DFs for exterior girders and were very close to the FEM into, the DFs were more closed to the AASHTO LRFD
DFs for interior girders. DFs for exterior girder were con- specification for this multilane loaded case.
trolling DFs in AASHTO LRFD specification, while the
controlling DFs were interior girder DFs in the FEM. This
is probably caused by the lack of parapets in the FEM, 6 Parametric Study
which could provide stiffness for the exterior girders. A parametric study was performed for hollow slab bridge
DFs for interior girder in AASHTO LRFD specification to investigate the parameters affecting the DFs of vehicle
was 1% to 18% higher than the DFs in FEM. For exterior loads. The effect of parameters on the DFs were quanti-
girder, DFs of AASHTO LRFD specification was 37% to fied and compared with AASHTO LRFD specifications.
48% higher than the DFs in FEM. Span length, bridge skew and bridge deck thickness were
As the loading lane increased from one to two in the three major parameters in this parametric study. A base-
FEM, DFs increased 17% for exterior girders and 34% for line FEM was determined using the dimensions and sup-
interior girder. While as the loading lane increased from porting conditions of the field tested bridge. The baseline
two to three, the DFs increased by 2% and 9% for exte- model was adjusted to study the effect of one parameter
while the other parameters were kept in the baseline
rior and interior girders, respectively. DFs could be dif-
value. While changes in certain parameters (i.e., span
ferent for one lane loaded and two lanes loaded, but less
length) do affect other parameters (i.e., girder stiffness),
difference for two lanes loaded and three lanes loaded. In
the effect was minimal and they could be evaluated inde-
multilane loaded cases, DFs of three loaded lanes were
pendently (Zokaie 2000). Therefore, span length was var-
more closed to the AASHTO LRFD specification com-
ied from 8 to 20 m, which was the specified length of the
pared with the DFs of two loaded lanes. When the bridge
hollow slab bridge, in the FEM. The effect of skew was
investigated by adjusting the skew angle from 0 to 60 .
Table 3 Distribution factors for different number of lanes. The bridge deck thickness was varied from 0.1 m to 0.5 m
by adjusting the thickness of plate element. DFs calcu-
Number of lanes Calculation method DFEXT DFINT
lated from the FEM were compared with those calculated
One lane AASHTO LRFD 0.230 0.202 ones according to the AASHTO LRFD specifications, this
MIDAS 0.155 0.171 ratio of which could reflect the difference between FEM
ANSYS 0.155 0.175 and AASHTO LRFD specifications.
Two lanes AASHTO LRFD 0.265 0.255 DFs for one loaded lane and three loaded lanes were
MIDAS 0.183 0.231 calculated in the Table 4 to investigate the effect of span
ANSYS 0.180 0.233 length. It can be seen from Table 4 that DFs for exte-
Three lanes AASHTO LRFD 0.265 0.255 rior girder increased 10% for one loaded lane and 42%
MIDAS 0.193 0.252 for three loaded lanes, while for interior girder, DFs
ANSYS 0.190 0.252 decreased 30% for one loaded lane and 9% for three
DFEXT distribution factor of exterior girder; DFINT distribution factor of interior loaded lanes. As the DFs calculated by MIDAS CIVIL
girder. and ANSYS were really close, so the DFs in Fig. 15 were
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 11 of 13
Calculation method MIDAS ANSYS MIDAS ANSYS MIDAS ANSYS MIDAS ANSYS MIDAS ANSYS
DFEXT-1 0.136 0.132 0.147 0.147 0.155 0.155 0.155 0.155 0.150 0.147
DFINT-1 0.205 0.209 0.177 0.174 0.171 0.175 0.160 0.163 0.148 0.145
DFEXT-3 0.148 0.145 0.175 0.171 0.193 0.190 0.200 0.197 0.210 0.206
DFINT-3 0.269 0.272 0.261 0.263 0.252 0.255 0.251 0.251 0.246 0.249
DFEXT-i distribution factor of exterior girder for i loaded lane; DFINT-i distribution factor of interior girder for i loaded lane.
MIDAS CIVIL DFEXT 0.194 0.184 0.195 0.195 0.193 0.186 0.188 0.172
DFINT 0.252 0.242 0.253 0.253 0.252 0.254 0.253 0.261
ANSYS DFEXT 0.193 0.183 0.192 0.192 0.190 0.184 0.186 0.170
DFINT 0.254 0.245 0.255 0.255 0.255 0.255 0.256 0.263
Zhao et al. Int J Concr Struct Mater (2020) 14:22 Page 12 of 13
8 Discussion
The parapets have been proved to provide more stiffness
Fig. 16 FEM/AASHTO ratio for skew parameter.
to the exterior girder compared with the interior girders
(Barker 2001; Zhou et al. 2015). As the vehicle loading
test was conducted on the bridge before the parapet was
constructed, the parapets were not modeled in the FEM.
So the effect of the parapets on DFs was not discussed in
this paper.
Acknowledgements
The authors thank Zhongguo John Ma for his assistance during writing this
paper and Wanfeng Liu and Dong Zheng for their assistance during the field
tests.
Authors’ contributions
YZ proposed the aim of the research and designed the field test. GW and RT
conducted the field test and collected the test data. XC and YZ analyzed and
interpreted the test data regarding the live load distribution. YZ was a major
contribution in writing the manuscript. All authors read and approved the
final manuscript.
Funding
Finical support by the key research platform open fund project of China
No. 310821171121. National Science Foundation of China under Grant Nos.
51678061 and 51978063.
Fig. 17 DFs for bridge deck thickness parameter. Availability of data and materials
The datasets used and analyzed during the current study are available from
the corresponding author on reasonable request.
the effects of span length, skew and bridge deck thick- Competing interests
ness on the DFs were also investigated in the parametric The authors declare that they have no competing interests.
study. From the analyses, the following conclusions were
Received: 6 September 2019 Accepted: 31 January 2020
obtained:
Conner, S., & Huo, X. S. (2006). Influence of parapets and aspect ratio on live- Seo, J., Kilaru, C., Phares, B. M., & Lu, P. (2015). Live load distribution factors for
load distribution. Journal of Bridge Engineering, 11, 188–196. a short span timber bridge under heavy agricultural vehicles. Structures
Eom, J., & Nowak, A. S. (2001). Live load distribution for steel girder bridges. Congress, 2015, 2164–2173.
Journal of Bridge Engineering, 6, 489–497. Seo, J., Teja Kilaru, C., Phares, B., & Lu, P. (2017). Agricultural vehicle load distribu-
Harris, D. K. (2010). Assessment of flexural lateral load distribution methodolo- tion for timber bridges. Journal of Bridge Engineering, 22, 04017085.
gies for stringer bridges. Engineering Structures, 32, 3443–3451. Song, S.-T., Chai, Y., & Hida, S. E. (2003). Live-load distribution factors for con-
Harris, D. K., Civitillo, J. M., & Gheitasi, A. (2016). Performance and behavior of crete box-girder bridges. Journal of bridge engineering, 8, 273–280.
hybrid composite beam bridge in Virginia: Live load testing. Journal of Waldron, C. J., Cousins, T. E., Nassar, A. J., & Gomez, J. P. (2005). Demonstration of
Bridge Engineering., 21, 04016022. use of high-performance lightweight concrete in bridge superstructure
Harris, D. K., Cousins, T., Murray, T. M., & Sotelino, E. D. (2008). Field investiga- in Virginia. Journal of performance of Constructed Facilities, 19, 146–154.
tion of a sandwich plate system bridge deck. Journal of Performance of Zhou, Y. J., Ma, Z. J., Zhao, Y., Shi, X. W., & He, S. H. (2015). Improved definition of
Constructed Facilities, 22, 305–315. dynamic load allowance factor for highway bridges. Structural Engineering
Hodson, D. J., Barr, P. J., & Halling, M. W. (2011). Live-load analysis of postten- and Mechanics, 54, 561–577.
sioned box-girder bridges. Journal of Bridge Engineering, 17, 644–651. Zokaie, T. (2000). AASHTO-LRFD live load distribution specifications. Journal of
Mabsout, M. E., Tarhini, K. M., Frederick, G. R., & Kobrosly, M. (1997). Influence of Bridge Engineering, 5, 131–138.
sidewalks and railings on wheel load distribution in steel girder bridges.
Journal of Bridge Engineering, 2, 88–96.
Namy, M., Charron, J.-P., & Massicotte, B. (2015). Structural behavior of bridge Publisher’s Note
decks with cast-in-place and precast concrete barriers: Numerical mod- Springer Nature remains neutral with regard to jurisdictional claims in pub-
eling. Journal of Bridge Engineering, 20, 04015014. lished maps and institutional affiliations.
Schulz, J. L., Commander, B., Goble, G. G., & Frangopol, D. M. (1995). Efficient
field testing and load rating of short-and medium-span bridges. Struc-
tural Engineering Review, 3, 181–194.
PEMBEBANAN
- Dalam jurnal ini, terdapat dua jenis pembebanan yang dianalisis, yaitu pembebanan gaya terkonsentrasi
dan pembebanan kendaraan.
- Pembebanan Gaya Terkonsentrasi: Pada tahap uji beban gaya terkonsentrasi, beban terkonsentrasi
diterapkan pada gording eksterior jembatan menggunakan metode uji lapangan. Beban ini diterapkan
pada tengah setiap jalur jembatan secara lateral. Jumlah dan posisi kendaraan beban divariasikan untuk
mempertimbangkan semua kombinasi beban yang mungkin.
- Pembebanan Kendaraan: Pada tahap uji beban kendaraan, beban kendaraan diterapkan sebagai beban
terkonsentrasi pada FEM jembatan. Beban kendaraan ini merupakan beban yang dihasilkan oleh
kendaraan yang melintasi jembatan. Jumlah dan posisi kendaraan beban divariasikan untuk
mempertimbangkan semua kombinasi beban yang mungkin.
MANFAAT
- Memahami distribusi beban lateral pada jembatan hollow slab: Penelitian ini memberikan pemahaman
yang lebih baik tentang distribusi beban lateral pada jembatan hollow slab. Hal ini penting untuk
memastikan keandalan dan keamanan struktur jembatan.
- Validasi metode analisis numerik: Penelitian ini menggunakan uji lapangan untuk memvalidasi metode
analisis numerik yang digunakan dalam perencanaan jembatan hollow slab. Validasi ini membantu
memastikan bahwa metode analisis yang digunakan memberikan hasil yang akurat dan dapat
diandalkan.
- Peningkatan desain jembatan hollow slab: Dengan pemahaman yang lebih baik tentang distribusi beban
lateral, penelitian ini dapat membantu dalam meningkatkan desain jembatan hollow slab. Hal ini dapat
mengarah pada pengembangan pedoman desain yang lebih efisien dan aman.
- Kontribusi pada penelitian jembatan beton: Penelitian ini memberikan kontribusi pada penelitian yang
lebih luas tentang jembatan beton. Hasil penelitian ini dapat digunakan sebagai referensi untuk
penelitian selanjutnya dalam bidang ini.
CONTOH JEMBATAN :
Engineering Structures
AR T I C LE I N F O A B S T R A C T
Keywords: Efficient and reliable load-carrying capacity evaluation of deficient bridges is essential for bridge maintenance
Bridge load-carrying capacity and management. Based on load-carrying capacity evaluation of an existing bridge, operator can formulate
Static load test strategies and implement countermeasures such as repair, strengthening or traffic regulation. Currently, static
Truck pass-by test load test is used widely for evaluation of bridge load-carrying capacity. The test, however, is often considered
Radar device
expensive and requires massive traffic closure that hinders frequent implementation on highway or bridges in the
Girder bridges
rural area. (KELEMAHAN) In this paper, an efficient method for load-carrying capacity evaluation of girder
Influence line
bridges based on displacement under moving vehicle is proposed. (KELEBIHAN) The method consists of three
main procedures, namely, radar- based dynamic displacement measurement under truck pass-by test, influence
line extraction, and estimation of the displacement under the static load test scenario. Based on these procedures,
evaluation of load-carrying capacity is conducted following the AASHTO manual. The method was verified in
numerical simulations using three-dimensional finite element model of typical girder bridge under vehicle loading.
Effects of vehicle weight, vehicle speed, and road roughness on the accuracy of displacement estimation were
investigated in the numerical simulations. Full-scale implementation of the method was conducted on a simply
supported pre- stressed concrete girder bridge to investigate feasibility of the proposed method and to provide
guidance for practical application. The results reveal that load rating factor of the girders can be reliably evaluated.
The evaluation process was also more effective and economical than the conventional method.
* Corresponding author at: Institute of Advanced Sciences, Yokohama National University, Japan.
E-mail address: dion@ynu.ac.jp (D.M. Siringoringo).
https://doi.org/10.1016/j.engstruct.2020.111645
Received 23 June 2020; Received in revised form 29 September 2020; Accepted 22 November 2020
Available online 16 December 2020
0141-0296/© 2020 Elsevier Ltd. All rights reserved.
Z. Sun et al. Engineering Structures 229 (2021) 111645
Both time and cost-effective field test are necessary in practice for bridge members; DC and DW stand for the dead load effects caused by
load-carrying capacity evaluation of a bridge. For example, traffics on structural components and the wearing surface, respectively. LL and IM
vital link such as highway bridges are too busy that the full bridge stand for the live load effect and its corresponding dynamic effects
closure as required by conventional static load test are often impossible respectively; γDC, γDW and γLL denote the load factors under the different
to implement. On the other hand, budget for maintenance is often load types. Furthermore, the current load-carrying capacity C is defined
limited to perform such testing on the bridges in rural areas and small as:
cities. This condition leaves bridges on rural roads lacking a reliable
C = ϕ cϕ sϕRn (2)
load-carrying capacity rating even when their conditions are visibly
deficient. where ϕc ,ϕs ,ϕ and are the condition factor, system factor, and the LRFD
In this paper, a reliable and efficient load-carrying capacity evalua- resistance factor, respectively, while Rn denotes the nominal resistance
tion method is proposed for girder bridge using moving vehicle. Instead of bridge member.
of conducting the static load test with several trucks, the proposed For an existing bridge with available design documents, C can be
method utilizes a pass-by truck to obtain displacements of the bridge computed based on material properties, section dimensions, reinforce-
girder, and a radar-based device to measure the dynamic displacement. ment, and prestress configurations and other design information.
The paper is organized such that the background on the load-carrying However, despite the rigorous theoretical background, difficulty in
capacity is firstly discussed. Next, the proposed methodologies for determining the load-carrying capacity of a bridge may arise due to
load-carrying capacity using moving vehicle are described in detail. uncertainties in concrete material property, cross-section area, rein-
Numerical simulations and parametric studies using three-dimensional forcement, and unknown deteriorated bridge condition. In such a case,
finite element model are latter presented, followed by full-scale imple- static load test is often conducted to obtain the bridge responses and
mentation on a typical girder bridge. Finally, results of experiments are then compared with the computational results. The comparison is used
discussed in detail, and some suggestions for practical implementation as a basis for updating the load rating factor to provide a more realistic
of the method are provided. (MANFAAT) load-carrying capacity evaluation.
It should be mentioned, however, that a complete static load test is
2. Background for load-carrying capacity evaluation labor-extensive and time consuming. For example, it takes almost one
day to instrument the bridge with various sensors, and another day to
In most countries, routine bridge inspection is conducted at a regu- conduct the test. Generally, four to six trucks are needed for a static load
lated maximum interval. The interval is specified as 2 years, 3 years, and test of a bridge. Traffic closure of all lanes is required, which is very
5 years in United States, China, and Japan, respectively. Bridge operator difficult to implement in reality. In addition, the static load test for a
may also schedule special inspection within the periodic interval to common medium span bridge can be very costly. For example, it is re-
investigate a known or suspected deficiency, such as after a vehicle/ship ported that the cost for such a test could be up to 6% of the bridge
collision. Unfortunately, the widely used condition rating systems usu- replacement cost [10]. These are some reasons that make static load test
ally do not provide direct information on the remaining load-carrying for realistic load capacity evaluation of existing bridges difficult to
capacity of the severely deficient bridges. Such information is essential implement widely.
for bridge operators to formulate strategies and implement counter-
measures such as repair, strengthening or imposing traffic regulation on 3. Methodologies for load-carrying capacity evaluation
the bridge. Therefore, an in-depth investigation on the load-carrying
capacity is often necessary. The investigation is usually conducted In this paper, an efficient method for load-carrying capacity evalu-
with the calculation-based method using finite element (FE) model ation of girder bridges is proposed. The method based is on measure-
made using the design documents. ment of bridge displacement under moving vehicle (PEMBEBANAN).
In the United States, the AASHTO bridge evaluation manual rec- Fig. 1 describes a flowchart of the proposed method, which can be
ommends comparison between the field tests and analytical results to explained as follows. First, a radar-based dynamic displacement
update the bridge load rating factor [6]. Several European countries measurement device is selected, and careful preparation is conducted
have adopted or in the process of adopting new codes for the assessment for measurement loca- tion, test truck weight and speed determination.
of existing bridges such as the UK, Denmark, and Switzerland [3]. The Second, the truck pass- by test is performed to obtain the dynamic
codes incorporate numerous advanced techniques for load rating of displacement. Third, the dynamic displacement data is processed to
existing bridges such as the use of data from weight-in-motion (WIM) calculate the fundamental mode frequency of the bridge. Fourth, a low-
system. Planning, load application, measurement equipment and test pass filter-based technique is employed to obtain the influence line for
procedures for bridge load testing are also described in the guidelines for displacement response of the bridge. Fifth, a simulated static load test is
existing bridges in UK [7]. Similar load test is also required in China [8] conducted on the bridge, andthe displacement results are calculated.
for deficient bridges. The guidelines usually recommend static testing to There are three main procedures involved in the proposed method,
obtain responses and later compare them with the reference values namely, the radar-based dynamic displacement measurement under a
obtained analytically or by finite element model. Afterwards, the dead truck pass-by test, the low-pass filter-based influence line extraction,
load and live load effects are calculated and compared with the resis- and the calculation with simulated static load test scenario. They are
tance factor to obtain estimation of the remaining capacity. In Japan, explained with more details in the following subsections.
although not explicitly required in the guideline, static load test for load
rating is conducted in a special case to provide a reference for load-
carrying capacity evaluation. For example, the static load testing was 3.1. The Radar-based dynamic displacement measurement
performed on Myokyo Bridge, a prestressed concrete box girder bridge,
in 2011 to provide estimate of remaining load-carrying capacity of the In recent years, dynamic displacement is seen as a more attractive
then severely corroded bridge [9]. measurand because it provides information on the deflection as well as
The load rating factor (RF) is adopted based on the load and resis- modal properties of the measured bridge. In this study, a commercial
tance rating factor method, which can be expressed as [6]. interferometric radar device is used for non-contact displacement
C — γ DC — γ DW measurement in the truck pass-by test. The radar system can measure
RF = DC DW
(1) displacement of multiple targets at higher sampling frequency without
γLL (LL + IM) interrupting the traffic. This makes the device more flexible and
In which, C stands for the current load-carrying capacity of the convenient to use than the traditional LVDT sensor or dial gauges. The
2
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig. 1. Flowchart for the proposed load-carrying capacity evaluation method. Note that the procedures in shaded box are proposed in this study.
device employs two principles, namely, the Stepped Frequency 3.2. Low-pass filter-based influence line extraction
Continuous Wave (SF-CW) and the interferometry. The former is to
distinguish multiple targets, and the latter is to allow measurement of Since Moses [13] introduced the idea of WIM to obtain the axles and
high frequency. A more detailed introduction on the techniques and gross weights of vehicles in motion with sensors instrumented on a
application procedures was presented by Su et al. [11]. The device has bridge, the influence line identification has been investigated widely.
been applied for the field truck pass-by bridge tests and proved to be Most of the existing methods utilize bridge response under a moving
efficient and reliable [12]. vehicle instead of the time-consuming step-by-step static test method.
Before the truck pass-by test, a detail testing program should be Many researchers have come proposed numerous techniques to elimi-
developed, which includes testing schedule and scenarios, such as truck nate the dynamic components. McNulty and O’Brien [14] adopted the
speeds and running lanes. For a girder bridge, the test truck runs over all point-by-point process on the strain response graph to obtain the bridge
lanes corresponding to the regulated positions in static load test, and experimental influence line. Manual adjustment is required in this
displacement is measured at the mid-span of the relevant girders. process, however, and this can be subjective and largely depends on the
Preparation for the test includes bridge field survey, preparation of the experience of the user. Wang et al. [15] proposed a fitting method for
test truck and weight measuring. Next, radar device is deployed under extracting the influence line from bridge dynamic response induced by a
the bridge and reflectors are installed at the mid-span. The relative lo- passing vehicle, which describes the influence line and the fluctuation
cations of antenna and reflector targets are measured to calculate di- part with a piecewise polynomial and a series of harmonic sinusoids,
rection of the radar wave propagation. Afterwards, the test is started. To respectively. Sun et al. [16] derived two damage indexes from identified
minimize traffic interruption, the test is generally conducted during at strain influence line from structural health monitoring data, which can
the nighttime when traffic volume is low. During the test, public vehicles indicate the damage existence and damage location, respectively. The
and pedestrians are not allowed to enter the testing area. The mid-span methodology is verified with measurements of a cable-stayed bridge. An
displacement is measured with radar device at high sampling frequency. in-depth review and comparison of the existing influence line identifi-
cation methods was summarized by Zheng et al. [17].
3
Z. Sun et al. Engineering Structures 229 (2021) 111645
Vehicle-induced bridge vertical displacement response y(x,t) is In this paper, a low-pass filter is employed to obtain the quasi-static
schematically illustrated in Fig. 2, The vehicle-induced load is repre- part as the influence line. Initially, the fast Fourier transform (FFT) was
sented by a constant force P that moves at a constant speed v from left to applied to the dynamic displacement data measured by radar device,
right. At the instant of force arrival, the beam is at rest without initial and the fundamental frequency of the bridge is identified. It should be
defection or velocity. The beam is assumed to follow the Euler-Bernoulli noted that although the vehicle weight may affect fundamental fre-
beam assumption with small deflections, and damping is assigned pro- quency of a bridge identified from moving vehicle test as reported in
portional to the vibration velocity [18,19]. The governing equation of [27], such variation was not observed in this case. This is probably
motion of a simply supported beam with uniform cross-section can be because in this case, the vehicle mass is much smaller than total weight
expressed as of the bridge. This information was used to determine the cut-off fre-
4 2
quency for the low-pass filter. The original signal was filtered with the
EI ∂ y(x, t) + m ∂ y(x, t) + 2mω ∂y(x, t) = Pδ(x — vt) (3)
cut-off frequency in the frequency domain; and an inverse Fourier
∂x4 ∂t2 b
∂t transform is performed to obtain the filtered displacement.
In Eq. (3), x denotes the coordinate in the longitudinal direction on
the beam; t is the time of the load moving;y(x, t)is the vertical 3.3. Displacement calculation using simulated static load test scenario
displacement of the beam at time t; L is the total length of the beam; P is
the amplitude of the load; m is the mass density for a unit length; EI is the After obtaining the displacement influence lines, the simulated static
flexural rigidity of beam; v is the velocity of the moving load; and ωb is load test is performed according to the designated truck weights and
the circular frequency of the beam. δ(x —vt) denotes the Dirac’s delta locations in static load test scenarios. The displacement yt is estimated as
function, which shows that the load P is applied only at the contact follows.
location vt on the beam. The equation of motion is subjected to boundary ∑n
mode of the beam vibration. evaluation method, numerical investigation is conducted on a three-
It can be observed in Eq. (6) that the bridge response under moving dimensional girder bridge model. Static load scenario is firstly con-
vehicle includes two parts, the quasi-static component, and the dynamic ducted, which serves as a reference of comparison. Truck pass-by load
component [19]. They are related to the vehicle speed and the bridge scenarios with different truck weight and speeds are then investigated,
natural frequency, respectively. The vehicle induced dynamic and the results provide a guidance for selecting the truck weight and
displacement has been investigated by numerous researchers, focusing speed in the real bridge field tests.
on extraction of bridge fundamental frequency from estimated vehicle
response [20–24], detection of local stiffness loss [12,25], prediction of
bridge dynamic response [26], among others.
4
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig.3. (a) Finite element model of the bridge built in ABAQUS software, (b) Static load test scenario, (c) Location of artificial damage on girder G4 at between 7 m
and 9 m positions by reduction of Young’s modulus corresponding to 20% of stiffness loss.
the bridge model is described below. The deck is made of reinforced Static (mm) 2.31 2.98 2.67 3.12 2.32
concrete with following dimensions: length 25 m, width 8.5 m and
thickness 0.2 m, accommodating two traffic lanes. The concrete cover is
4.3.1. Truck model
0.035 m thick. Main girder consists of five steel I-girder beams in the
A quarter car model of 20-ton vehicle [22] is used in the analysis, as
longitudinal direction with dimensions 1.2 m high and 0.6 m wide.
shown in Fig. 4 (a). The stiffness coefficient and damping coefficient are
Thicknesses of the web and flange members are both 0.03 m. Artificial
damage of is created on girder G4 between 7 m and 9 m positions with 3.8 × 106N/m and 3.4 × 104 N⋅s/m, respectively. The equation of mo-
reduction of Young’s modulus corresponding to 20% of stiffness loss as tion for the quarter car vehicle model traversing a bridge can be
shown in Fig. 3(c). expressed as:
The transverse beam is an x-shaped beam. The top and bottom Mv Ü v + Cv U̇ v + Kv Uv = fv (8)
chords are diagonals with dimensions 100 × 100 × 9.5 and 75 × 75 ×
9.5, respectively. The deck and bearings, I-girder, and transverse beams, where Mv, Cv, Kv are the mass, damping coefficient, and stiffness of the
rebars are modeled by solid element, shell element and truss element, vehicle, respectively. Üv , U̇v , Uv are the acceleration, velocity, and
respectively. There are 11,067 elements and 15,897 nodes in total. The displacement of the vehicle, respectively. The symbol fv denotes the
boundary conditions are specified as simply supported at two ends of the vertical force on the vehicle due to interaction with the bridge at the
five I-shape girders on bearings. Details on the material properties used wheel position.
in the finite element model are provided in Table 1. It is assumed that the vehicle moves at a constant speed on a straight
path. The interaction force with the bridge is expressed as:
4.2. Static load scenario f = fv + Mv g — Cv U̇ v — Kv Uv (9)
In the static load scenario, four 20-ton trucks are placed on the bridge The road surface roughness that affects dynamic response of the
in a stationary position as shown in Fig. 3(b). In the longitudinal di- vehicle-bridge interaction system is represented by random field using a
rection, the load positions are arranged to ensure the largest mid-span stationary Gaussian random process. The road surface roughness is
deflection. In the transverse direction, the positions correspond to the described by the following power spectral density functions [26]:
two traffic lanes. To avoid local stress concentration, the four vehicle ∑N √̅̅̅
weights are applied as the surface forces. Static analysis is performed, r(x) = k=1
2Gd (nk)Δncos(2πnkx — θk) (10)
and the deflections at the mid-span of the five T girders are obtained as In which, Gd(nk) = a
(2πnk )
2 is the power spectral density function. a is
shown in Table 2. The values are 2.31 mm, 2.98 mm, 2.67 mm, 3.12 mm,
selected as 2 × 10 indicating a good road surface condition. Δn is the
—6
and 2.32 mm, respectively. It is noted that the displacement of the
damaged girder T4 is larger than that at the symmetrical location of frequency interval of the road surface wave expressed as Δn = nmax —nmin ,
in which n N
girder T2. max and n min are 1 Hz and 0.05 Hz. Nis the wave number
selected as 20 here andnk is the frequency of wave number k; while θk is
the random phase angle distributed between 0 and 2π. The selected road
4.3. Truck Pass-by load scenarios roughness profile used in the numerical simulation is illustrated in Fig. 4
(b).
In the simulated truck pass-by load scenario, parametric study In the vehicle-bridge interaction analysis, displacement compati-
related to the truck weight and speed is conducted. The results are bility at the wheel contact points should be satisfied. This means that
described in the following sections. vehicle and the bridge move together in the vertical direction, without
zero contact force. A global iteration procedure is used in the compu-
Table 1 tation [26]. The interaction force time history is calculated in MATLAB
Material properties of bridge model. with Newmark’s method and then applied as loads on the ABAQUS
Material Property Value bridge model [28].
Concrete Density (kg/m3) 2400
Young’s modulus (Pa) 2.8 × 1010 4.3.2. Load scenario description
Poisson’s ratio 0.2 The parametric study is performed with different truck weights and
Steel Density (kg/m3) 7850 speeds with the following scenarios. 1) While keeping truck speed
Young’s modulus (Pa) 2.1 × 1011
constant at 5 m/s, the truck weight is varied to 10ton, 20ton, 30ton,
Poisson’s ratio 0.3
5
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig. 4. (a) Quarter-car vehicle model and (b) Road roughness thickness used in finite element numerical simulation.
40ton, and 50 ton. 2) While keeping truck weight at 20 ton, the truck
speed is varied to 2.5 m/s, 5 m/s, 12.5 m/s, and 25 m/s. In each
simulated scenario, the truck runs over the two traffic lanes that
correspond to the truck transverse positions in the static load scenario.
An example of loading case for vehicle runs over traffic lane above girder
T4 is illustrated in Fig. 5.
Fig.6. Displacement time-histories at mid-span of girder T2 for the case of: (a) original response to truck pass-by over girder T2, (b) filtered response to truck pass-by
over girder T2, (c) original response to truck pass-by over girder T4, (d) filtered response to truck pass-by over girder T4.
6
Z. Sun et al. Engineering Structures 229 (2021) 111645
During the static load test, the traffic was closed in both directions.
vehicle locations in the static load test scenario in Fig. 3 (b) are 8.5 m
Four loading truck locations are shown in Fig. 10 (a) and photo of the
and 16.5 m respectively, which correspond to the instantaneous time
loading test is shown in Fig. 10 (b). It is noted that the static test took
1.7 s and 3.3 s in Fig. 6 (a), (b), (c) and (d). The comparison of dis-
about 3 h with total traffic closure, and 12 persons participated in the
placements obtained from the static and moving load with varied truck
test. The traffic closure was implemented under low traffic volume. The
weights at constant speed 5 m/s is illustrated in Fig. 7. In the figure, the
load test was administered by the local traffic administration and
displacement result at mid-span of all five T-girders are obtained based
announced to local people in advance. Each truck weighs around 30 ton
on Eq. (1). The figure demonstrates that the difference between 10-ton
as presented in Table 3. The distance between first (front) axle with the
truck scenario and the static load result ranges from 8.1% to 10.5%.
second (rear) axle is 3.5 m. Whereas the longitudinal and lateral dis-
In contrast, the differences for all other four truck weights are between
tances between the two rear wheels is 1.4 m and 1.8 m, respectively
0.32% and 4.5%. Therefore, selecting the same truck weight as in static
(Fig. 10(a)). The ratio between the test load effect and the design load
load case would provide better accuracy in the proposed method.
effect is 1.01. Digital level Leica DNA-03 is used to measure the
Fig. 8 exhibits comparison of displacements subjected to a 20-ton
displacement, which has a precision of 0.1 mm. The displacement
truck loading running at various speeds of 2.5 m/s, 5 m/s, 12.5 m/s
measurement results at mid-span of the five T girders are summarized in
and 25 m/s, and the corresponding displacement obtained by static
Table 4.
loading. It is evident from the figure that the differences with the static
result are much smaller compared to that in the case of varied truck
5.3. Description of the truck pass-by testing
weights. The difference is more than 2% for the highest speed scenario of
25 m/s, which is still acceptable. It can be concluded that the proposed
In the dynamic test, the truck ran over the two lanes respectively, and
method is very robust with regard to the truck running speed.
the running routes are selected according to the truck position in Fig. 10
(a). The truck speeds were selected as 10 km/h, 20 km/h and 30 km/h
respectively. The reflectors were mounted at mid-span of girder T1 and
T2 respectively, and a schematic view of the measurement with radar
device is illustrated in Fig. 11(a). The distance between the radar device
and reflector on T1 is 14.20 m with an elevation angle of 24◦. Fig. 11(b)
shows the reflector used for the radar device. The truck pass-by test was
conducted for six times with 12 min of traffic closure involving four
persons and one test truck. The testing time was reduced significantly
from a typical duration of 3 h in the conventional static test to only 12
min. Moreover, the required number of persons involved is reduced from
12 to 4 persons. This is certainly more efficient and effective than the
standard static load test.
The bridge displacements were measured, and the fundamental
natural frequency was obtained at 4.41 Hz. A low-pass filter was used to
eliminate the dynamic components in the signal, and the cut-off fre-
quency was selected as 0.5 Hz after sensitivity analysis. Fig. 12(a) and
(b) show the original and filtered displacement time histories recorded
at girder T2 for the case of truck ran over lane 1, respectively. Whereas
the original and filtered displacement time histories recorded at girder
T2 for the case of truck ran over lane 2 are illustrated in Fig. 12(c) and
(d), respectively. Similarly, the filtered displacement signals at girder T1
and T2 under all three different speeds were obtained. As shown by
Fig. 12, the maximum dynamic displacements at the center of the girder
Fig. 8. Comparison of displacements obtained from static and dynamic load T1 and T2 are 1.83 mm and 3.36 mm, respectively. Whereas the
scenarios for various truck speeds. maximum filtered displacements are 1.55 mm and 2.45 mm,
7
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig. 9. Description of object bridge. (a) Photo of Wenxi Bridge, (b) Finite element model of the bridge built in Midas Civil software, (c) T-girder beam cross section,
(f) Bridge deck cross section (unit:mm).
respectively. reduced the stiffness. The similar phenomenon of stiffness loss caused by
The measurements results were compared with finite element model cracks opening was also pointed out in another investigation [31].
of the bridge built in Midas/ Civil Software [29] (Fig. 9 (b)). Comparison In principle, the proposed method utilizes a specific load level to
of the testing results with the finite element model under the same predict behavior of the bridge under heavier loads, which is verified to
loading condition is also shown in Fig. 12. The maximum corresponding be more efficient in time and less labor-intensive. In the application, it is
displacements under the test truck load at the center of the girder T1 and important to consider nonlinearity effects with regard to the extrapo-
T2 are 1.90 mm and 3.27 mm, respectively. The displacements obtained lation. It is recommended that nonlinearity amplification factor of 1.1 is
from the load test are smaller than the results from finite element adopted for PC T-girder bridges. Hence the averaged mid-span
analysis, which is believed due to the fact that the bridge stiffness is displacement under three speeds become 6.66 mm and 7.55 mm for
larger than the designed stiffness, as is the usual case of typical bridges. girder T1 and T2, respectively. These results agree well with the previ-
Nevertheless, the results demonstrate that bridge displacements under ously obtained deflections in Table 4.
loading condition obtained experimentally agree reasonably well with
the estimated displacements in the finite element model.
5.5. Estimation of load rating factor from measurement
5.4. Analysis of measurement results As mentioned previously, a finite element model of Wenxi Bridge was
built in Midas/Civil software [28] as a reference to compare the mea-
Based on the static loading truck configuration shown in Fig. 10 (a), surement results with theoretical values. The model was built using as-
the truck positions were obtained, and the corresponding values on the built drawing of the bridge. Using the finite element model, the mid-
displacement time history were extracted. According to Eq. (1), the mid- span displacements under the static loading scenario were calculated
span displacement on girder T1 and T2 were obtained for three different as 8.80 mm and 9.00 mm for girder T1 and T2, respectively. The live load
truck speeds as presented in Table 5. The table demonstrates that the transverse distribution factors for girder T1 and T2 were 0.656 and 0.523,
results of analysis using truck pass-by test are all smaller than that in the respectively. The load effects and load-carrying capacity were calculated
static load test. The largest difference is 10.3% and 5.3% for girder T1 following Eq.(1) as illustrated in Table 6, where the design live load
and T2, respectively. A possible explanation of this is that the method is scenario was applied according to loading pattern shown in Fig. 13.
based on linear extrapolation of displacement due one pass-by truck According to the carrying capacity C defined by Eq.(2), the condition
while the actual measurement results may include some nonlinearity factor (ϕc), system factor (ϕs ), and the LRFD resistance factor (ϕ) for this
effect. It should also be mentioned that the increase in displacement may bridge are 0.95, 1.0 and 0.9, respectively. For the prestressed concrete
not linearly correlate with the load amplitude increase. In a destructive bridge, γDC = 1.25, γDW = 1.5, γL = 1.75. Moreover, according to the
test of a similar PC girder bridge, the same nonlinearity phenomenon bridge deck surface condition, IM is taken as 0.2. The load rating factor
was observed when the load was applied from zero to the maximum RF for girder T1 and T2 can be calculated as follows.
loading [30]. This loading procedure induced nonlinearity due to
opening of cracks in the damaged section of the girder and eventually
8
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig. 10. Static load test (a) Loading truck configuration (Unit: m) (b) Photo of the test.
Table 3 Table 4
Truck axle weights. Displacement results in the static load test.
Truck No. Axle weight (ton) T girder no. Displacement (mm)
9
Z. Sun et al. Engineering Structures 229 (2021) 111645
Fig 11. Implementation of radar measurement system, and truck loading. (a) Schematic view of truck pass by test (b) Reflector mounted on the T girder.
Fig. 12. Displacement results at T2 mid-span under truck speed of 10 km/h and their comparison with FE model results for: (a) original displacement when truck
moves over lane 1, (b) filtered displacement when truck moves over lane 1, (c) original displacement when truck moves over lane 2, (d) filtered displacement when
truck moves over lane 2.
Table 5
Obtained displacements at mid-span of two girders. Table 6
Speed Girder T1 Girder T2 The calculated parameters for load carrying capacity evaluation (Unit: kN•m).
(km/h)
Displacement Difference Displacement Difference Parameter Girder T1 Girder T2
(mm) with static (mm) with static DC 2323 2212
result result DW 1022 824
10 5.97 10.3% 6.80 5.3% LL 1959 1562
20 6.09 8.6% 6.97 2.8% C 8456 8009
30 6.11 8.3% 6.81 5.0%
10
Z. Sun et al. Engineering Structures 229 (2021) 111645
11
Z. Sun et al. Engineering Structures 229 (2021) 111645
[16] Sun SW, Sun LM, Chen L. Damage detection based on structural response induced by traffic load: methodology and application. Int J Struct Stab Dy 2015;16(4): 1640026.
[17] Zheng X, Yang DH, Yi TH, Li HN. Development of bridge influence line identification methods based on direct measurement data: a comprehensive review and comparison. Eng
Struct 2019;198:109539.
[18] Frýba L. Vibration of solids and structures under moving loads, Vol. 1. Springer Science & Business Media; 2013.
[19] Weaver JR, Timoshenko SP, Young DH. Vibration problem in engineering. New York: John Wiley and Sons; 1990.
[20] Yang YB, Lin CW, Yau JD. Extracting bridge frequencies from the dynamic response of a passing vehicle. J. Sound & Vibr. 2004;272(3–5):471–93.
[21] Lin CW, Yang YB. Use of a passing vehicle to scan the bridge frequencies - an experimental verification. Eng. Struct. 2005;27(13):1865–78.
[22] Sun Z, Nagayama T, Su D, Fujino Y. A damage detection algorithm utilizing dynamic displacement of bridge under moving vehicle. Shock Vib 2016;8454567.
https://doi.org/10.1155/2016/8454567.
[23] Wang H, Nagayama T, Nakasuka J, Zhao B, Su D. Extraction of bridge fundamental frequency from estimated vehicle excitation through a particle filter approach.
J Sound Vib 2018;428:44–58.
[24] Siringoringo DM, Fujino Y. Estimating Bridge Fundamental Frequency from Vibration Response of Instrumented Passing Vehicle: Analytical and Experimental Study. Adv Struct
Eng 2012;15(3):417–33.
[25] O’ Brien EJ, Fitzgerald PC, Malekjafarian A, Sevillano E. Bridge damage detection using vehicle axle-force information. Eng Struct 2017, 153(15): 71–80.
[26] Sun Z, Zou ZL. Towards an efficient method of predicting vehicle-induced response of bridge. Eng Computation 2016;33(7):2067–89.
[27] Yang YB, Cheng MC, Chang KC. Frequency variation in vehicle-bridge interaction systems. Int. J. of Struct. Stability & Dyn. 2013;13(2):1350019.
[28] Abaqus 6.10 [Computer software]. Dassault Systemes Simulia Corp.,Providence, RI.
[29] MIDAS.IT Corporation 2006.
[30] Budette EG, Goodpasture DW. Test to failure of a prestressed concrete bridge. PCI J 1974;19(3):92–103.
[31] Saiidi M, Douglas B, Feng S. Prestress force effect on vibration frequency of concrete bridges. J Struct Eng-ASCE 1994;120(7):2233–41.
KEKURANGAN :
Dalam abstrak jurnal tersebut, terdapat permasalahan yang dihadapi dalam evaluasi kapasitas pemikul beban
jembatan. Permasalahan tersebut antara lain adalah metode pengujian beban statis konvensional yang dianggap
mahal dan membutuhkan banyak tenaga kerja, sehingga sulit diterapkan secara luas. Terutama pada jembatan di
daerah pedesaan, keterbatasan anggaran seringkali menjadi hambatan untuk melakukan pengujian semacam itu.
Akibatnya, jembatan-jembatan di jalan-jalan pedesaan seringkali tidak memiliki penilaian kapasitas pemikul beban
yang dapat diandalkan, meskipun kondisinya terlihat buruk.
KELEBIHAN :
Dalam jurnal ini, diusulkan suatu metode yang efisien untuk evaluasi kapasitas dukung beban jembatan gelagar
berdasarkan perpindahan kendaraan yang bergerak.
PEMBEBANAN :
a. The Radar-based dynamic displacement measurement
Pembebanan dilakukan dengan menggunakan pengukuran perpindahan dinamis menggunakan perangkat
radar berbasis interferometri. Perangkat radar ini digunakan untuk mengukur perpindahan non-kontak pada
saat truk melintas di atas jembatan. Pengukuran perpindahan ini dilakukan dengan menggunakan prinsip
Stepped Frequency Continuous Wave (SF-CW) dan interferometri. Perangkat radar ini memiliki
keunggulan dalam mengukur perpindahan dengan akurasi yang tinggi, dengan ketelitian hingga 0,01 mm
b. Low-pass filter-based influence line extraction
pembebanan dilakukan dengan menggunakan metode low-pass filter untuk mengekstraksi garis pengaruh
(influence line) yang berkaitan dengan respons perpindahan quasi-statis jembatan akibat kendaraan yang
melintas. Metode ini digunakan untuk memisahkan komponen quasi-statis dari respons dinamis jembatan.
c. Displacement calculation using simulated static load test scenario
Pembebanan dilakukan melalui simulasi skenario pengujian beban statis. Dalam pengujian ini, posisi truk
pada pengujian beban statis ditentukan, dan nilai-nilai perpindahan pada waktu tertentu diambil. Hasil
analisis menggunakan metode uji lewat truk ini menunjukkan bahwa perkiraan perpindahan lebih kecil
dibandingkan dengan hasil pengujian beban statis. Perbedaan terbesar adalah 10,3% untuk girder T1 dan
5,3% untuk girder T2. Perbedaan ini mungkin disebabkan oleh metode ini hanya berdasarkan pada
ekstrapolasi linier perpindahan dari satu truk yang melintas, sementara hasil pengukuran aktual mungkin
memasukkan efek nonlinier. Selain itu, peningkatan perpindahan mungkin tidak berkorelasi secara linear
dengan peningkatan amplitudo beban. Meskipun begitu, metode ini tetap memberikan hasil yang dapat
diterima dan dapat digunakan untuk mengevaluasi kapasitas pemikul beban jembatan
12
Z. Sun et al. Engineering Structures 229 (2021) 111645
STRUKTUR :
struktur jembatan yang dievaluasi adalah jembatan girder. Jembatan ini memiliki dek yang terbuat dari beton
bertulang dengan dimensi panjang 25 m, lebar 8,5 m, dan ketebalan 0,2 m. Jembatan ini memiliki lima balok I-
girder baja sebagai girder utama dengan dimensi 1,2 m tinggi dan 0,6 m lebar. Ketebalan anggota web dan flange
adalah 0,03 m. Pada girder G4, dibuat kerusakan buatan dengan mengurangi modulus Young sebesar 20% dari
kekakuan aslinya.
PERENCANAAN STRUKTUR :
Untuk estimasi yang lebih baik, faktor amplifikasi diusulkan untuk mempertimbangkan efek ketidaklinieran pada
beban berat pada jembatan girder yang rusak. Studi ini telah menunjukkan bahwa metode yang diusulkan dapat
menjadi pengganti yang efisien untuk uji beban statis tradisional, terutama untuk jembatan jalan tol atau jembatan
di daerah pedesaan. Di masa depan, lebih banyak pengujian diperlukan untuk mengukur faktor amplifikasi yang
tidak linear untuk jenis jembatan lain seperti jembatan plat, kotak, dan girder truss.
MANFAAT :
• Metode ini efisien dan ekonomis karena menggunakan pengukuran perpindahan dinamis dengan perangkat
radar yang cepat, menggantikan pengujian beban statis yang mahal dan memakan waktu.
• Metode ini memungkinkan pengambilan keputusan yang lebih baik dalam pemeliharaan dan manajemen
jembatan, dengan memberikan evaluasi kapasitas pemikul beban yang akurat untuk merencanakan perbaikan
atau regulasi lalu lintas yang tepat.
• Pengujian kapasitas pemikul beban jembatan menjadi lebih mudah dan cepat dilakukan, tanpa perlu menutup
jalan secara penuh, yang sulit dalam pengujian beban statis konvensional.
• Metode ini cocok untuk jembatan skala menengah di daerah pedesaan, terutama pada jembatan girder beton
prategang, dan memberikan hasil yang dapat diandalkan dalam mengevaluasi pemikul beban.
CONTOH JEMBATAN :
13
JEMBATAN PELENGKUNG (ARCG BRIDGE)
Civil Engineering
A R T I C L E I N F O ABSTRACT
Article history: Concrete bridges are used for both highways and railways roads. They are characterized by their durabil-
Received 13 April 2018 ity, rigidity, economy and beauty. Concrete bridges have many types such as simply supported girder
Revised 29 December 2018 bridges, arch bridges and rigid frame bridges. However, for very large spans, arch bridges are more eco-
Accepted 24 January 2019 nomic in addition to their beauty appearance. In this research, a geometrical structural optimization
Available online 16 February 2019
study for a deck concrete arch bridges using Genetic Algorithms technique is presented. This research
aims mainly to demonstrate a methodology to find the least cost design, in term of material volume,
Keywords: by finding the optimal profile. A Finite Element numerical model is used to represent the arch structure.
Optimization
The MATLAB programing platform is used to develop codes for Genetic Algorithms optimization tech-
Genetic Algorithms
nique and Finite Element analysis method. The resulted design from the optimization process is com-
Finite Element
Arch bridges pared to traditional design and an obvious cost reduction is obtained.
© 2019 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-
nd/4.0/).
1. Introduction cross-section of steel beams [6] and the design of water distribu-
tion networks [7]. Few structural optimization researches investi-
In recent years, Artificial Intelligence (AI) techniques show wide gated the optimal design of arch bridges. In 2008, there was an
applications to facilitate human life. These techniques (such as attempt for bending moment reduction in arch using a specific
Evolutionary Algorithms, Simulated Annealing, Particle Swarm arrangement sequence for hangers [8]. Two years later, an optimal
Optimization, Ant Colony Optimization, Tabu Search, Fuzzy Opti- design of steel truss arch bridges using a Hybrid Genetic Algorithm
mization and Artificial Neural Network-based methods) have been was presented [9]. A comparison between circular and parabolic
applied individually or coupled to solve the life problems [1,2]. arch was presented in the same year and it was observed that
Genetic Algorithms technique, which is one of AI branches, is used the parabolic arch has lower internal forces, more economic than
for solving complex engineering optimization problems. The the circular arch and it is more efficient for open bridges [10].
Genetic Algorithms (GAs) have been started in 1975 by John Hol- Nowadays, Genetic Algorithms are being applied to widespread
land [3]. GAs have been widely used in most engineering fields application in business, scientific and engineering specializations.
such as: chemistry and physics, mechanics, aerodynamics, Finite Element Analysis (FEA) is a numerical technique that can
robotics, networks, architecture and civil engineering. In civil engi- be used to get the accurate solution for complex engineering prob-
neering, GAs have been used in almost all branches; for example: lems. It started since several centuries back as an idea of replacing
reinforced concrete flat slab buildings [4], optimization of tunnel the complicated problem by a simpler one. The main concept of
profile in different ground conditions [5], determining the optimal FEA is considering the origin solution as built up of many small
interconnected sub regions [11]. It has been used for solving many
* Corresponding author at: El-souq St., Next to Civil Registry, Maghagha, El-Minia, structural analysis problems [12]. Today, the Finite Element
Egypt. Method is considered one of the most important analysis tools
E-mail addresses: M.Eid@mu.edu.eg (M.A. Eid), eng.mgs.2013@gmail.com for engineers and scientists.
(M.G. Sayed).
The arch bridges have a main advantage over other crossing
Peer review under responsibility of Ain Shams University.
structures; the economy. In order to increase their economy supe-
rior, some attempts were introduced to find the optimal design for
these bridges. In 1999, the optimality criteria method was used to
develop a recursive relationship for the design variables
Production and hosting by Elsevier
(mainly, the depth at the crown and at the support) considering
https://doi.org/10.1016/j.asej.2019.01.005
2090-4479/© 2019 The Authors. Published by Elsevier B.V. on behalf of Faculty of Engineering, Ain Shams University.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
508 M.Z. Abd Elrehim et al. / Ain Shams Engineering Journal 10 (2019) 507–516
displacement, stresses and minimum depth constraints [13]. In 2.2. Structural system and applied loads
2014, global optimization algorithms were used to minimize the
cost of super structure particularly arches and hangers of network The considered arch bridge consists of two-way directions sep-
arch bridge by optimizing the geometric shape, rise to span ratio, arated by an island with 2.00 m width and sidewalk in each side
cross section of arch, and the number, arrangement and cross sec- with 3.00 m width. Each way has 2 lanes with lane width = 3.00 m.
tional dimensions of hangers [14]. In another attempt, presented in So, the total width of the bridge is 20.00 m as shown in Fig. 2.
2015, a research proposed a methodology for optimization the lon- The structural system consists of a bridge slab, secondary
gitudinal shape of open arch bridge using Simultaneous Perturba- beams, main beams, and the arch girder as shown in Fig. 3. Second-
tion Stochastic Approximation (SPSA) algorithm [15]. In 2016, a ary beams are distributed every 2.00 m along the bridge (2 edge
new research presented a new strategy for modular steel construc- and 9 intermediate secondary beams). The main beams, with
tion: Adjustable module for variable depth steel arch bridges [16]. 20.00 m length, distributed at least every 4.00 m in which the
In the same year, some researchers developed a new methodology, bridge slab is divided into a number of one way bays. So, the num-
which evaluates the optimum configuration of network arch bridge ber of main beams is adaptable by the program according to the
schemes (hangers) [17]. span which is defined by the user. The two arch girders support
The current research deals with the concrete arch bridge opti- the bridge from both sides.
mization problem by developing a new analysis tool that com- The dead loads is automatically calculated by the program con-
prises both Finite Analysis and Genetic Algorithms using MATLAB sidering slab thickness, main beam dimensions, secondary beam
programming platform. In this methodology, the arch is modelled dimensions and material properties which are given by the user.
numerically as a continuum structure divided into 2D Constant According to the Egyptian Code of Practice for design loads
Strain Triangular (CST) elements joined together at corners. The (ECP), there is a typical live loading case for design, including the
joints coordinates are the design variables for the optimization dynamic impact factor, shown in Fig. 2 [19]. Using influence line,
process which aims to minimize the arch weight. The developed the worst position for vehicle to get the maximum internal forces
Finite Element Analysis program is used to check the structural on arch girder is the center.
safety issue.
2.3. The structural analysis
2. The numerical model
Herein, FEA model is presented. This model is classified as a two
dimensional case (plane stress). The assumption of plane stress is
2.1. Geometry
applicable for bodies whose dimension is very small in one of the
coordinate directions. Thus, the analysis of thin plates loaded in
The parabolic geometry for the applied arch is presented by the
the plane of the plate can be made using the assumption of plane
equation:
stress. In plane stress distribution, it is assumed that.
f ωx
z¼4 ðl — xÞ ð1Þ rzz ¼ rzx ¼ ryz ¼ 0 ð2Þ
l2
where where (z) represents the direction perpendicular to the plane of the
plate as shown in Fig. 4.
x, z are the coordinates of any point on the arch The main concept in FEA is to divide the structural problem to
l is the span of arch small interconnected elements. The Constant Strain Triangular
f is the rise of the arch at the crown (at l/2) (CST) elements are to be used to form the arch body. The CST ele-
ment is composed of 3 corner nodes; each node has 2 Degrees of
The rise to span ratio for arches varies widely. Most arches Freedom (DOF), u and v, as shown in Fig. 5. The stress-strain rela-
would have a rise to span ratio in the range from 0.16 to 0.20. tion is given by:
The span to crown thickness (t crown) ratio (from existing concrete frg ¼ ½D]feg ð3Þ
arches) can be taken between 70 and 80. The springing thickness
0 1
(t springing) to crown thickness ratio is between 1.55 and 1.72. 1 t 0
Fig. 1 shows the preliminary dimensions for a typical concrete E B C
½D] ¼ @ 2 t 1 A 0 ð4Þ
arch [18]. 1 t —1—t
0 0 2
where
8 9
< ex =
e = strain vector for element = : ey ;
cxy
ex = strain component in x direction.
ey = strain component in y direction.
cxy = the strain on the plane xy.8 9
< rx =
r = stress vector for element = ry
:s ;
xy
rx = stress component in x direction.
ry = stress component in y direction.
sxy = shearing stress on the plane where x value is constant and
y is the direction.
D = the elasticity matrix which expresses the properties of
Fig. 1. Preliminary dimensions for a typical arch. material
M.Z. Abd Elrehim et al. / Ain Shams Engineering Journal 10 (2019) 507–516 509
Table 1
Assumed conditions.
Fig. 8. Deformation.
qffi ffiffiffiffiffiffiffiffi ffiffiffi
Table 2
Comparison of results for the developed numerical model.
Allowable shear stress ¼ 21:79 ω f cu =cc ðkN=m 2Þ ð13Þ
weight with a suitable factor. The best 8 solutions which have less mutation probability (p m) – number of iterations (not less than
weight for the arch girder are selected to be new parents for the 1000 iterations).
next generation [24]. These operators are repeated and new gener- By using a laptop CORE I3 processor and 2 GB RAM, the devel-
ations are created until the global optimum solution is reached. oped model has been applied for 3 examples with different spans
as shown in Table 3.
4. The results After running the model for these examples, with average
4 days running time for each example, the induced optimized
The developed program gives the user the authority to change shape for each case has been shown in Figs. 12–14.
any of the following numerical model parameters: It’s noticed that thickness of arch girder is increased in the
Arch span (not less than 20.00 m) – arch width – secondary lower part of the first and last third while increasing in the upper
beam width – secondary beam depth – main beam width – main part of the middle third of the arch for all cases. A skeletal model is
beam depth – bridge slab thickness – fcu – poisson’s ratio – developed for the investigated cases using SAP program. The
induced bending moment diagrams are shown in Figs. 15–17.
It is clear that the resulted rough optimal design follow the
Table 3 moment diagram with reversed direction. This rough geometry
Input Data of Examples. can be smoothed using curve fitting (for practical considerations)
Example no. 1 2 3 as shown in Figs. 18–20.
Span (m) 20 30 53
Fig. 21 shows the progression of the minimum weight value
Arch width (m) 0.70 0.80 0.80 through 1,000,000 iterations of the optimization process. There
Width of secondary beams (m) 0.25 0.30 0.30 are many clear drops in the weight value in the first 100 iterations
Depth of secondary beams (m) 0.50 0.50 0.60 then the change in weight value occurs at long intervals until
Width of main beams (m) 0.30 0.40 0.40
Depth of main beams (m) 0.60 0.70 0.80
reaches the optimal weight.
Thickness of bridge slab (m) 0.20 The resulted optimal geometry is compared to concrete arch tra-
Concrete strength (fcu ) (kN/m 2 ) 29419.95 ditional design. This comparison is illustrated in Table 4. The rough
Poisson’s ratio 0.25 optimal design has arch weight reduction ranges from 35% to 40%
Mutation probability (P m ) 0.1 compared to traditional design arch weight. After solution enhance-
Number of iterations 1,000,000
ment, the reduction percentage became in range of 30–35%.
Fig. 15. Bending Moment Diagram for skeletal model of span 20.00 m.
Fig. 16. Bending Moment Diagram for skeletal model of span 30.00 m.
Fig. 17. Bending Moment Diagram for skeletal model of span 53.00 m.
Fig. 18. Enhanced arch shape vs traditional shape for 20.00 m span.
Fig. 19. Enhanced arch shape vs traditional shape for 30.00 m span.
M.Z. Abd Elrehim et al. / Ain Shams Engineering Journal 10 (2019) 507–516 515
Fig. 20. Enhanced arch shape vs traditional shape for 53.00 m span.
Table 4
Comparison of optimal and traditional design.
Example no. 1 2 3
[17]
M.Z. Abd Elrehim et al. / Ain Shams Engineering Journal 10 (2019) 507–516 517
KELEBIHAN :
Penelitian untuk mendemonstrasikan metodologi dalam menemukan desain dengan biaya paling rendah, khususnya
dalam hal volume material, dengan menemukan profil yang optimal.
KEKURANGAN :
Kurangnya penelitian tentang optimasi desain jembatan lengkung beton dengan Algoritma Genetika memerlukan
penelitian lebih lanjut untuk mengembangkan metode yang lebih efektif. Meskipun ada beberapa studi sebelumnya di
bidang ini, masih ada ruang untuk pengembangan yang lebih efisien.
PEMBEBANAN :
Beban mati dihitung secara otomatis oleh program dengan mempertimbangkan ketebalan pelat jembatan, dimensi
balok utama, dimensi balok sekunder, dan properti material yang diberikan oleh pengguna. Beban hidup mengacu
pada kasus pembebanan desain yang mencakup faktor dampak dinamis. Menurut Kode Praktik Mesir untuk
perancangan struktur, ada satu kasus beban hidup yang sering digunakan untuk menguji kekuatan struktur.
STRUKTUR :
Struktur jembatan tersebut terdiri dari pelat jembatan, balok sekunder, balok utama, dan balok lengkung. Balok
sekunder terdistribusi setiap 2,00 m sepanjang jembatan, dengan 2 balok tepi dan 9 balok sekunder di antaranya.
Balok utama, dengan panjang 20,00 m, terdistribusi setidaknya setiap 4,00 m, yang membagi pelat jembatan menjadi
beberapa bay satu arah. Ada dua balok lengkung yang mendukung jembatan dari kedua sisi.
PERENCANAAN STRUKTUR :
Sruktur jembatan lengkung beton direncanakan dengan menggunakan metode analisis elemen hingga (finite element
analysis) dan algoritma genetika (genetic algorithms). Proses perencanaan melibatkan pemodelan komputer jembatan
sebagai struktur berkelanjutan yang dibagi menjadi elemen segitiga dengan regangan konstan (Constant Strain
Triangular/CST) dalam program analisis elemen hingga. Variabel desain yang digunakan dalam optimasi adalah
koordinat simpul-simpul elemen segitiga ini, dengan tujuan mengurangi berat jembatan lengkung. Selain itu, program
analisis elemen hingga yang telah dikembangkan juga digunakan untuk memeriksa masalah keamanan struktural
jembatan.
MANFAAT :
- Pengurangan biaya
Penelitian ini menggunakan algoritma genetika untuk mencari desain jembatan lengkung beton yang lebih efisien
secara biaya. Dengan menemukan desain yang lebih optimal, penelitian ini dapat membantu mengurangi biaya
pembangunan jembatan lengkung beton
- Efisiensi struktural
Dengan menggunakan pendekatan optimasi yang canggih, penelitian ini dapat menghasilkan desain struktural
jembatan lengkung beton yang lebih efisien. Desain yang lebih efisien dapat meningkatkan kinerja struktural
jembatan dan mengurangi risiko kegagalan struktural
518 M.Z. Abd Elrehim et al. / Ain Shams Engineering Journal 10 (2019) 507–516
CONTOH JEMBATAN :
A R T I C L E I N F O A B S T R A C T
Keywords: The traditional wind load assessment for long-span bridges relies on assumed models for the wind field and
Structural monitoring aerodynamic coefficients from wind tunnel tests, which usually introduce some uncertainties. Recent studies have
Wind loads shown that large deviations can exist between the predicted and observed wind-induced dynamic response of
Force identification suspension bridges. In studies of the dynamical behavior of bridges, inverse force identification methods can
Suspension bridge
therefore be an interesting tool in the assessment of possible uncertainties involved in the modeling of wind loads.
This paper presents a novel case study of the identification of the dynamic wind loads on the 1310 m long
Hardanger bridge, a suspension bridge equipped with a monitoring system for wind and vibrations. The modal
wind loads are identified from acceleration data using an algorithm for model-based joint input and state esti-
mation. Several data sets with different wind conditions are presented. The wind loads are studied in the time and
frequency domains and are compared to the mean velocity and turbulence characteristics of the wind.
* Corresponding author.
E-mail addresses: oyvind.w.petersen@ntnu.no (Ø.W. Petersen), ole.oiseth@ntnu.no (O. Øiseth), e.lourens@tudelft.nl (E. Lourens).
URL: https://www.ntnu.edu/kt/research/dynamics (Ø.W. Petersen).
https://doi.org/10.1016/j.jweia.2019.104045
Received 8 March 2019; Received in revised form 8 June 2019; Accepted 16 November 2019
Available online xXXX
0167-6105/© 2019 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/ ).
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
2016) and time-domain (Li et al., 2015; Qiao et al., 2016; Aucejo et al.,
2018a; Sun et al., 2015; Lai et al., 2017; Chang et al., 2019; Bernal and
Ussia, 2015) deconvolution approaches with regularization schemes
have been investigated.
Inverse estimation of wind loads on tall buildings has been studied in
wind tunnels (Zhi et al., 2018; Hwang et al., 2011) and in full-scale
conditions (Zhi et al., 2016; Kang et al., 2012; Niu et al., 2015). Recon-
struction of wind loads on small-scale guyed masts have also been re-
ported (Amiri and Bucher, 2017). While an increasing number of
cable-supported bridges are equipped with data monitoring systems
(Ko and Ni, 2005; Wong, 2004), to the best of the authors’ knowledge, no
studies have focused on the inverse estimation of wind loads. Therefore,
it is of interest to test the state-of-the-art force identification methodol-
ogies with full-scale data and to assess their utility as a tool in detailed
studies of wind loads. As will be explained, force identification does not
utilize any specific load model assumptions, which makes the method- Fig. 2. Topography of the area surrounding the bridge.
ology attractive for studies of non-stationary data, wind gusts or
abnormal events, and could help to provide a better understanding of the
mechanisms behind wind loads. Before such detailed studies can be
performed, the performance of the methods must be evaluated to identify
the weak and strong points. Although the inverse methods have been
explored in laboratory experiments, the transition to full-scale bridges
means that different conditions and uncertainties will play an important
role.
In this article, the dynamic wind loads on the Hardanger bridge are
estimated based on acceleration data. The capability of the full-field
estimation of these algorithms for long-span bridges was studied previ-
ous works (Petersen and Øiseth, 2019), showing that the methodology is
feasible, with the main limitation being the number of modal responses
that can be reconstructed.
The rest of this paper is organized as follows. Section 2 presents the Fig. 3. Positions of the sensors on the Hardanger bridge.
bridge case study and the methodology. In Section 3, results from several
data sets with different wind conditions are examined. The loads are installed sensors. Eight digital ultrasonic anemometers A1-A8 (Gill Wind
assessed in the time and frequency domains, and the influence of the Master Pro 3D) are mounted on the hangers along the bridge span.
wind characteristics on the loads are investigated. A short discussion of Twenty triaxial accelerometers (CUSP-3D strong-motion, range 4 g,
the usefulness of the methodology is provided, and conclusions are given SNR 130 dB) are located in the girder and the towers. More detailed
in Section 4. specifications of this system are given in (Fenerci et al., 2017).
K0 — ωj2M0 φj ¼ 0 (2)
Fig. 1. View of the Hardanger bridge from the north end (Photo: Øyvind Wiig where Ω ¼ diagðω1 ; …; ωnm Þ and Ξ ¼ diagðξ1 ; …; ξnm Þ are diagonal
Petersen/NTNU). matrices assigned the still-air natural frequencies (ωj ¼ 2πfj) and
2
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
damping ratios. The measurement vector ðy Þt2Rny , which consists of Kalman-type technique, is used. This method provides minimum-
accelerations and displacements in selected structural DOFs, is given as bð Þforces (p tk ).In addition,
variance estimates of the states (x tk ) and input bð Þ
follows: a dual Kalman filter (DKF) (Azam et al., 2015) and a
time-domain sequential deconvolution method (Bernal and Ussia, 2015)
yðtÞ ¼ Sa u€ðtÞ þ S d uðtÞ (4)
are tested. For more detailed information on the estimation methods, we
refer to the cited works.
where Sa and Sd 2Rnd×nDOF are binary matrices that select the measured
Contrary to classic (forward) modeling of wind loads, no specific
DOFs. By introducing the modal state variable
time-space assumptions are imposed on the wind field or the wind forces.
xðt Þz¼t ½T ðz_Þ t ð ÞR]2nm2, the system equations is converted into state-
T T
Since the force identification is driven directly by the vibration data (and
space form:
the system model), there is no model for the wind field and no functions
x_ ðtÞ¼ AcxðtÞþ BcpðtÞ (5) are given for the aerodynamic coefficients and admittance of the bridge
deck. This is critical since the study is not constrained to cases with
yðtÞ¼ G xðtÞþ J pðtÞ (6) idealized load conditions so that one can study loading from wind fields
c c
that are skew, inhomogeneous, non-stationary or essentially unknown.
0 I
Ac ¼ 2 (7) 2.3. System model and limitations in the methodology
—Ω —2ΞΩ
The still-air modes are solved from a FE model of the bridge (Fig. 4).
0
Bc ¼ (8) The model is tuned via a model updating procedure to match modal
I
properties from operational modal analysis; see (Petersen and Øiseth,
2019) for details. Table 1 lists the modes, where the labels H, V, T and P,
Gc ¼½ SdΦ — SaΦΩ2 —SaΦ2ΩΞ ] (9) respectively, denote horizontal, vertical, torsion and pylon motions. The
identified (almost) still-air damping ratios ξj in Table 1 are adopted for
Jc ¼½ SaΦ ] (10) the state-space model.
In the discretization to an intersample time Δt ¼ 0:1 s, a first-order The system model is truncated to contain only 18 modes, a choice
hold constraint is imposed on the input. This constraint is necessary to explained in the following. In most force identification methods, the
preserve the model accuracy for the very low-frequency range, which is number of unknown forces is theoretically limited by the number of
important for analysis of the wind-induced response. After the addition of linearly independent outputs. The identification of modal forces requires
stochastic noise, this process yields the following model equations: that rankðJÞ ¼ nm in an instantaneous inversion of system (Maes et al.,
2014). Fig. 5 shows the evaluation of the singular values of J for different
xkþ1 ¼ Axk þ Bpk þ vk (11) model sizes. The clearly visible singular value drop-off for the 19-mode
model indicates close-to-linear relationships in the output data, which
yk ¼ Gxk þ Jpk þ wk (12) increases the condition number of this matriX. Therefore, no more than
18 modal forces could be identified without running into severe ill-
where the secondary state xk ¼ xk — Fpk is introduced (Aucejo et al., conditioning in the system inversion. This number of forces is notably
2018b) and the system matrices are given by: less than the upper theoretical limit governed by the aforementioned
rank-criterion, highlighting the importance of designing optimal sensor
A ¼ expðAcΔtÞ (13) locations, in particular for long-span bridges that have many modes that
contribute to the wind-induced total response. The truncated system
B ¼ ðA — IÞAc—1 Bc — F þ AF (14) model accounts for roughly 80% of the measured acceleration data.
, Some limitations exist when only acceleration data are available, and
F ¼ Ac—1 ðA — IÞ Δt — I Ac—1 Bc (15) the workarounds and effects hereof are discussed in the following. To
obtain a steady state of the filter gain matrices and reduce the compu-
G ¼ Gc (16) tational time, displacement data are generated by numerical integration
of the accelerations. The influences of these displacements are weighted
J ¼ Jc þ GcF (17) downward to rely mostly on the acceleration data. In filtering type ap-
plications, the covariance matrices are often considered tuning parame-
The vectors wk 2 R2nm and vk 2 Rny are zero-mean white noise terms ters for the system; R is set to diagonal with values of 10—6 for the
that model the uncertainties of the states and measurements. Their cor- accelerations and 10—2 for the displacements. This process is similar to
responding covariance matrices are given by: the principle of using dummy measurements for stabilization, as pro-
E wkwT ¼ Q δkl posed in (Naets et al., 2015); the displacement outputs are strongly
l
penalized to lessen their influence while maintaining a stable solution. A
l
E vkv ¼ R δkl
T
(18) —3
3
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
low-frequency estimates, but such data were unfortunately not available. Table 2
Furthermore, the purely static forces (f ¼ 0 Hz) are theoretically Data sets used in the identification of forces.
impossible to reconstruct from acceleration data alone (Maes et al., 2014) Recording Date and start Duration Mean wind Incoming
and are therefore not considered further. no. time velocity (A6) direction
In addition, the output data are low-pass filtered at 0.6 Hz during 1 2015-03-10 e6h e 10–28 m/s East
preprocessing since the contributions above this limit belong to modes 05:59
not accounted for by the system model (Table 1). In summary, only the 2 2015-05-05 e4h e 10–22 m/s West
dynamic force content in the range f 2 0:01; 0:6 ]Hz is considered. 14:53
½
3 2016-01-29 e8h e 15–28 m/s West
Different results could be obtained for other bridges or sensor network 12:30
configurations. 4 2016-12-26 e4h e 10–20 m/s West
18:09
4
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 7. Fourier transform of the identified forces for modes 1–12 in recording 1. The light blue curve is a smoothed FFT, the star signifies the single natural frequency,
the magenta lines signify all natural frequencies, and the green lines signify non-modeled modes. (For interpretation of the references to colour in this figure legend,
the reader is referred to the Web version of this article.)
5
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 8. Fourier transform of the identified forces for modes 1–12 in recording 2. The light blue curve is a smoothed FFT, the star signifies the single natural frequency,
the magenta lines signify all natural frequencies, and the green lines signify non-modeled modes. (For interpretation of the references to colour in this figure legend,
the reader is referred to the Web version of this article.)
Fig. 9. Fourier transform of identified forces for modes 1–12 in recording 3. The light blue curve is a smoothed FFT, the star signifies the single natural frequency, the
magenta lines signify all natural frequencies, and the green lines signify non-modeled modes. (For interpretation of the references to colour in this figure legend, the
reader is referred to the Web version of this article.)
6
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 10. Fourier transform of the identified forces for modes 1–12 in recording 4. The light blue curve is a smoothed FFT, the star signifies the single natural frequency,
the magenta lines signify all natural frequencies, and the green lines signify non-modeled modes. (For interpretation of the references to colour in this figure legend,
the reader is referred to the Web version of this article.)
these forces attempt to correct the unexplained behavior of the output would be to jointly estimate these aerodynamic coefficients with the
data. In this case, four non-modeled modes (i.e., modes not included in inputs and states; however, such complex approaches are considered
the state-space model) from the FE model near 0.24 and 0.4 Hz are future work and are not pursued here.
indicated in the figures. The deflection of these modes is mainly in the
cables with some small movements in the boX girder. Neglect of these 3.2. Influence of wind characteristics
modes likely causes estimation errors around these frequencies, leading
to a blunt peak for the horizontal and torsion modes. Once more, this Correlating the identified loads with the characteristics of the wind is
characteristic points to a limitation in this methodology for cable- of great interest. In this context, anemometer data are useful, although
supported bridges with a high number of modal contributors and/or the measurements could be seen as a sparse sampling of a large non-
the quality of the sensor network. sparse wind field. The limited amount of data poses a challenge. The
Consistently, for all recordings, a small dip is observed in the fre- wind forces on the structure depend on the effective wind field along the
quency domain at the respective natural frequencies for modes V1–V7 whole bridge, but only eight sampling locations are available at
and T1. This dip can be explained by the self-excited effects, which approXimately the elevation of the bridge deck. Although some as-
introduce added damping in the vibration modes. Although these dips sumptions or spatial interpolations can be made, the wind velocity field
may appear to be insignificantly small compared to the total force does not have a known low-rank dynamic structure, so the eight mea-
magnitude, they are important since they decrease the forcing around the surements alone cannot be used reconstruct the actual properties of the
natural frequency, which significantly reduces the modal response. For full field. The use of LIDAR technology is an alternative approach that
most suspension bridges with boX girder decks, the vertical modes potentially could fill this gap (Cheynet et al., 2017). In the present
experience the strongest influence of aerodynamic damping, which is application, the comparison of the wind measurements and identified
consistent with the observations made here. Furthermore, a sharp- forces is therefore of a heuristic type more than an analytical connection.
gradient drop is observed for mode T1 at f¼ 0:36 Hz, which is due to To make use of the anemometer data, the turbulence and mean wind
the aerodynamic stiffness reduction of the torsional motion. However, velocity should be separated while making as few assumptions as
this phenomenon becomes miXed with the two non-modeled modes at necessary. Let U ð Þt be the along-wind velocity with incoming direction
0.4 Hz, and the close spacing of the modes hinders a clear interpretation. ð θÞ t , as defined in Fig. 2, and let W
ð Þt be the vertical velocity. We assume
Self-excited forces are commonly modeled as proportional to the that the wind velocity consists of a deterministic time-varying mean
structural displacements and velocities. These self-excited forces (still in value Uð Þt ; Wð Þt plus a fluctuating component uð Þt ; wð Þt that is
the modal projection space) can actually be obtained from the state es- stochastic:
btimates by p se ðtÞ ¼ ½ΦT Kae Φ b ΦT Cae Φ]xðtÞ. Depending on the modeling
scheme adopted (Chen and Kareem, 2002; UðtÞ¼ UðtÞþ uðtÞ; WðtÞ¼ WðtÞþ wðtÞ (19)
2017), the aerodynamic coefficient matrices Kae and Cae2 CnDOF (ob- We use empirical mode decomposition (EMD) (Xu and Chen, 2004) to
Kavrakovand Morgenthal,
tained from wind tunnel tests) are generally dependent on the mean wind characterize the mean wind velocity from the measurement data. First
velocity (meaning that they will not be time-invariant) in addition to the proposed in (Huang et al., 1998), EMD is a data-driven tool that can be
b
frequency of the moving structure, so the calculation of pseðtÞ becomes a used to decompose non-stationary data into a basis set of intrinsic mode
convolution integral due to memory effects. An alternative method functions (IMFs), denoted by cjðtÞ, through an iterative sifting process;
7
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
see, for example, (Xu and Chen, 2004): mentioned previously, the wind is measured at only eight locations along
X X the span (x ¼ f460; 280; 240; 200; 180; —10; —180; 420g m), so the
UðtÞ¼ Ncðj UÞðtÞþ rð UÞðtÞ; WðtÞ¼ Ncðj W Þ ðtÞþ rðWÞðtÞ (20) quantities in Eq. (22) are not fully known. Despite this limitation,
ðÞ ðÞ ðÞ ðÞ
U t forces.
ing u t and U t the
Here, w tshort-term
represent “averaged
the approXintensity”
imate intensity of the buffet-
is quantified by the
N N
Fig. 11. Measurements from anemometer A6 in recording 1: a) along-wind velocity; b) vertical wind velocity; c) direction.
8
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 12. Comparison of the time evolution in recording 1: a) along-wind turbulence; b) vertical turbulence; c) identified modal forces.
Fig. 13. Measurements from anemometer A6 in recording 2: a) along-wind velocity; b) vertical wind velocity; c) direction.
9
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 14. Comparison of the time evolution in recording 2: a) along-wind turbulence; b) vertical turbulence; c) identified modal forces.
Fig. 15. Measurements from anemometer A6 in recording 3: a) along-wind velocity; b) vertical wind velocity; c) direction.
10
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 16. Comparison of the time evolution in recording 3: a) along-wind turbulence; b) vertical turbulence; c) identified modal forces.
Fig. 17. Measurements from anemometer A6 in recording 4: a) along-wind velocity; b) vertical wind velocity; c) direction.
11
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 18. Comparison of the time evolution in recording 4: a) along-wind turbulence; b) vertical turbulence; c) identified modal forces.
In recording 2 (Fig. 14), a more homogeneous wind field is observed, 3.3. Comparison with other algorithms and design wind load spectrum
although the north side (A6-A8) appears to have slightly more turbu-
lence. Moreover, the influence of the turbulence for t2 0 ½ : 30; 2 : 30] h The results in the previous sections show some potential issues in the
clearly manifests in the loads. Later, some large loads occur seemingly identification of the forces; therefore, it is interesting to compare other
without any large turbulence at the time, for instance, at t ¼ approaches to check whether they are better suited. In this section, two
f 3 : 12; 4 : 42g h. The exact reason for these loads is not known, but a additional inverse methods are considered. The first is a dual Kalman
closer examination of the anemometer time series for A1 and A2 in- filter (Azam et al., 2015), which is distinct due to its stable estimation of
dicates some strong local gusts with duration of less than 5 s, which are the acceleration output only. In this algorithm, the force has a statistical
not easily detected with an averaging window of 60 s. evolution model of the form pkþ1 ¼ pk þ ηk , where the fictitious process
In recording 3 (Fig. 16), the load is almost zero for the first 15 min, ηk is a zero-mean white noise vector with covariance E½ηk ηTl ] ¼ Qp δkl .
when the wind starts to increase. The influence of strong and along-span This covariance is deemed a regularization parameter that controls the
correlated turbulence can be observed at several time instants. At t ¼ 2 : force solution norm. In this case, a tuned value of Qp ¼ 10—1I yields the
20 h, the effect of an along-wind gust is clearly visible in the forces, and best fit to the output data (minimum innovation norm).
again, more pronounced at t ¼4 : 20 h. Aided by strong vertical turbu- Second, we apply a sequential deconvolution reconstruction (SDR) of
lence, almost all the modal forces reach the maximum magnitude at the the inputs (Bernal and Ussia, 2015). While the JIS and DKF are based on
same time. For reference, the maximum horizontal deflection in the mid- one-step recursive estimation, this method utilizes a sliding window for
span reaches almost 8 m, and the estimated stress in the boX girder at this time-domain deconvolution and is therefore computationally efficient.
time reaches a maximum well over 100 MPa, which testifies to the harsh
The size of the window and the rate of advance are, respectively, set to
impact felt by the structure. In the later period (t > 4 : 30 h), some strong 40Δt and 20Δt according to their definitions in (Bernal and Ussia, 2015).
but not fully correlated gusts still occur, for instance, around t ≈ 5 : 30 h, Note that the SDR uses a deterministic system model, thus neglecting the
which still leads to some large forces but not for all modes. stochastic noise (Q; R; S 0).
¼
In the last recording (Fig. 18), quite inhomogeneous wind conditions
The results in Fig. 19 from the time period t 2 ½2 : 30; 3 : 00] in
are observed. Although the data from A1 are missing, the wind at A2-A6 recording 2 display a representative image of the performance. Inter-
(south side) is considerably more turbulent. Correlations in the wind and
estingly, the deterministic method SDR provides estimates that are
forces are evident at the two indicated times t ¼ f 1 : 12; 2 : 35} h. consistent with the JIS, which formally has a stronger capability due to
However, the strong forces occurring at t ¼ 2 : 00 h cannot be explained the consideration of stochastic noise. This outcome is believed to be the
directly by the turbulence in this figure. result of the use of modal forces rather than localized forces. Since a
12
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Fig. 19. Identified forces for modes 1–12 for a 30 min period of recording 2.
Fig. 20. Identified forces for modes 1–12 for a 30 min period of recording 2. The star signifies the single natural frequency, the magenta lines signify all natural
frequencies, and the green lines signify non-modeled modes. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web
version of this article.)
13
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
modal force theoretically is a summation of the forces in all positions, it ~ ae ¼ ΦT Cae Φ and K
~ ae ¼ ΦT Kae Φ. Utilizing z ðω Þ ¼ H
~ð ωÞ pbð ωÞ ,
where C
can more easily smooth model errors, whereas a localized force has in-
Eq. (25) can be written as:
fluence only in its vicinity. The DKF generally gives lower values, an
effect that increases for higher modes for reasons that are not well un- ~ ae þ K
~ ae ÞH
~ ðωÞÞpb ðωÞ
pðωÞ ¼ ðI þ ðiωC (26)
derstood. From Fig. 20, it can be concluded that the difference is largely
due to the low-frequency discrepancy. Another observation is that the where the generalized transfer function is
DKF does not have a dip at the natural frequency (added damping from ~ ðωÞ ¼ ½— ω2 I þ iωð 2ΞΩ — C ~ ae Þ þ ð Ω2 — K
~ aeÞ] —1 . Only wind loads on the
H
self-excitation) but instead has an increased magnitude, which is believe
girder are taken into account due to the lack of data on the wind char-
to be unrealistic. Scaling the covariance Qp changes the overall force
acteristics at the elevation of the main cables. Assuming the spectrum
magnitude but not the frequency-domain characteristics. A robust Sf bf b ð ω Þ2 CnDOF×nDOF of the buffeting forces (Eq. (21)) is available, the
method for choosing this covariance in scenarios with many forces n ×n
generalized force spectrum SppðωÞ2 C m m
can finally be calculated as:
(possibly having different evolutions) is not available at the moment.
Next, a brief comparison is made with the loads predicted from the ~ ae þ K
Spp ðωÞ ¼ ðI þ ðiωC ~ ae ÞH
~ ðωÞÞH ΦT S f bf bðωÞΦðI þ ðiωC
~ ae þ K
~ ae ÞH
~ ðωÞÞ
design wind field model based on a Kaimal spectrum (see (Fenerci and (27)
Øiseth, 2017) for details). Herein, the static coefficients are set to the
following values obtained from wind tunnel tests (Siedziako et al., 2017): Fig. 21 compares the force spectra from the forward prediction and
the inverse estimate using data from the time period2t ½2 : 30; 3 : 00] in
CD ¼ 1:05, C’D ¼ 0, CL ¼ — 0:363, C’L ¼ 2:22, CM ¼ — 0:017, C’M ¼
0:786. The bridge deck has the proportions B ¼ 18:3 m and D ¼ 3:25 m. recording 2, in which the wind is normal to the bridge. Averaging over
In this context, this classic wind load modeling approach is called the this 30 min period, the mean wind velocity is U ¼ 21:2 m/s, and the
“forward analysis”, since its basis is the wind velocity field yielding the turbulence intensities are Iu ¼ σu=U ¼ 0:096 and Iw ¼ σw=U ¼ 0:035.
loads on the bridge. In contrast, the inverse estimation of the wind loads Although the results inevitably vary for different events, the time period
is driven by the measured acceleration data. These two approaches are shown has reasonably ideal (homogeneous and stationary) wind condi-
fundamentally different, but both independently result in the wind load tions and should yield a fair comparison. However, the design wind field
quantification. In the forward analysis the wind loads can be described model has many uncertain features that do not always match the
only stochastically, so spectral densities in the frequency domain are observed wind characteristics at the site (Fenerci et al., 2017); thus, the
employed. The buffeting plus self-excited forces are defined as: presented frequency-domain inquiry is a simple test to get an idea of the
similarity and cannot be considered a strict tool for validation.
fðω Þ ¼ f b ðωÞ þ Cae u_ ðωÞ þ Kae uðωÞ (24) Overall, the degree of (dis)agreement in Fig. 21 is as expected for the
For simplicity, Cae and Kae are taken from quasi-steady theory in the inverse force identification of full-scale bridges in complex and uncertain
following (see e.g. (Kavrakov and Morgenthal, 2017)). The (total) conditions, which has not been extensively explored. For the JIS modal
force estimate, the magnitude appears to match best at the individual
generalized forces, pðωÞ ¼ ΦTfðωÞ, can then be written as:
natural frequency. In the frequency range far below the natural fre-
~ ae z_ ðωÞ þ K
pðω Þ ¼ pb ðωÞ þ C ~ ae zðωÞ (25) quency, the DKF estimate agrees better, and the JIS yields a higher
magnitude. This result could either be due to an estimation error or
Fig. 21. Comparison between the spectral density of the inverse load estimates (averaged by Welch’s method) and the forward model prediction of the wind loads.
The star and the magenta line signify the natural frequency.
14
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
actual behavior. As discussed in Section 2.3, acceleration outputs are not with significant wind-induced excitation, a large and well-designed
always well suited for estimating very-low-frequency forces due to their sensor network is required to reconstruct the total dynamical behavior.
ω2-proportionality, which becomes small for close-to-static motion. The The estimation based on acceleration data also constitutes a limitation in
insensitivity could lead to a (frequency-banded) low signal-to-noise ratio, studies of wind loads, as close-to-static forces are not always well esti-
which is known to be particularly disruptive for inverse problems. mated. This source of uncertainty could be eliminated if strain mea-
Therefore, displacements or strains are more useful for this purpose; surements were added.
however, latent sensor noise is believed to play only a small role here.
The behavior can potentially be explained by some slow fluctuations in Acknowledgments
the mean wind velocity. Such fluctuations are typically not captured in
the forward analysis that considers stationary conditions (constant mean This work was financially supported by the Norwegian Public Roads
wind velocity). Administration.
References
3.4. Evaluation of methodology and usefulness of obtained load
information Amiri, A.K., Bucher, C., 2017. A procedure for in situ wind load reconstruction from
structural response only based on field testing data. J. Wind Eng. Ind. Aerodyn. 167,
75–86.
No real direct validation of the estimated forces is available at this
Aucejo, M., De Smet, O., Deü, J.F., 2018. On a space-time regularization for force
scale, so the uncertainties are difficult to quantify. Promising progress reconstruction problems. Mech. Syst. Signal Process. 118, 549–567.
has been made in understanding the application to suspension bridges, Aucejo, M., De Smet, O., Deü, J., 2018. Practical issues on the applicability of Kalman
filtering for reconstructing mechanical sources in structural dynamics. J. Sound Vib.
but the case study also shows some robustness issues with the estimation
442, 45–70.
algorithms, as well as the need for extensive sensor networks to achieve a Azam, S.E., Chatzi, E., Papadimitriou, C., 2015. A dual Kalman filter approach for state
high level of confidence in the results. Regardless, application robustness estimation via output-only acceleration measurements. Mech. Syst. Signal Process.
is, for this bridge, far better than that for a similar study on floating 60, 866–886.
Bernal, D., Ussia, A., 2015. Sequential deconvolution input reconstruction. Mech. Syst.
bridges (Petersen et al., 2019), where the model uncertainties are far Signal Process. 50, 41–55.
greater. Bietry, J., Delaunay, D., Conti, E., 1995. Comparison of full-scale measurement and
Although the data of the load and response behavior can be useful, an computation of wind effects on a cable-stayed bridge. J. Wind Eng. Ind. Aerodyn. 57,
225–235.
obvious dilemma is that measurements cannot be made before the Caracoglia, L., 2008. Influence of uncertainty in selected aerodynamic and structural
structure is built, at which point it has already been designed based on parameters on the buffeting response of long-span bridges. J. Wind Eng. Ind.
wind tunnel testing and wind field analysis, and major alterations cannot Aerodyn. 96, 327–344.
Chang, X., Yan, Y., Wu, Y., 2019. Study on solving the ill-posed problem of force load
be conveniently made. Thus, the methodology is more suited to research reconstruction. J. Sound Vib. 440, 186–201.
and validation studies on existing bridges or simply to monitoring Chen, X., Kareem, A., 2002. Advances in modeling of aerodynamic forces on bridge decks.
response behavior throughout the lifetime. The body of knowledge J. Eng. Mech. 128, 1193–1205.
Chen, L., Letchford, C., 2005. Proper orthogonal decomposition of two vertical profiles of
gained from the lessons from existing bridges is also a valuable stepping
full-scale nonstationary downburst wind speeds [lzcl]. J. Wind Eng. Ind. Aerodyn. 93,
stone that could facilitate the design of future bridges. 187–216.
The methodology is also promising for studies of the loads in more Cheynet, E., Jakobsen, J.B., Snæbjo € rnsson, J., 2016. Buffeting response of a suspension
unusual events that do not fit the ordinary set of wind modeling as- bridge in complex terrain. Eng. Struct. 128, 474–487.
Cheynet, E., Jakobsen, J.B., Snæbjo € rnsson, J., Reuder, J., Kumer, V., Svardal, B., 2017.
sumptions. The load mechanisms for bridges in highly non-stationary Assessing the potential of a commercial pulsed lidar for wind characterisation at a
downbursts in thunderstorms, for example, are not well documented bridge site. J. Wind Eng. Ind. Aerodyn. 161, 17–26.
(Chen and Letchford, 2005; Solari et al., 2015). In theory, the input and Cigada, A., Falco, M., Zasso, A., 2001. Development of new systems to measure the
aerodynamic forces on section models in wind tunnel testing. J. Wind Eng. Ind.
state estimation could also be extended to parameter estimation to Aerodyn. 89, 725–746.
indicate damage (stiffness reduction), but this task remains difficult Diana, G., Resta, F., Zasso, A., Belloli, M., Rocchi, D., 2004. Forced motion and free
because modal sensitivity to local damage is generally very small for long- motion aeroelastic tests on a new concept dynamometric section model of the
Messina suspension bridge. J. Wind Eng. Ind. Aerodyn. 92, 441–462.
span bridges. Fenerci, A., Øiseth, O., 2017. Measured buffeting response of a long-span suspension
bridge compared with numerical predictions based on design wind spectra. J. Struct.
4. Conclusions Eng. 143, 04017131.
Fenerci, A., Øiseth, O., Rønnquist, A., 2017. Long-term monitoring of wind field
characteristics and dynamic response of a long-span suspension bridge in complex
This article has presented the inverse identification of dynamic wind terrain. Eng. Struct. 147, 269–284.
loads from measured response data, focusing on the Hardanger bridge, a Huang, N.E., Shen, Z., Long, S.R., Wu, M.C., Shih, H.H., Zheng, Q., Yen, N.-C., Tung, C.C.,
Liu, H.H., 1998. The empirical mode decomposition and the Hilbert spectrum for
long-span suspension bridge instrumented with accelerometers. The nonlinear and non-stationary time series analysis. Proc. R. Soc. Lond.: Math. Phys.
methodology does not impose any prior assumptions on the wind Eng. Sci. 454, 903–995.
loading, as loads are estimated directly from acceleration data using an Huang, N.E., Shen, Z., Long, S.R., 1999. A new view of nonlinear water waves: the Hilbert
spectrum. Annu. Rev. Fluid Mech. 31, 417–457.
algorithm for joint input and state estimation. Hwang, J., Kareem, A., Kim, H., 2011. Wind load identification using wind tunnel test
Overall, the frequency-domain characteristics of the estimated loads data by inverse analysis. J. Wind Eng. Ind. Aerodyn. 99, 18–26.
appear to be realistic, except for the effect of some known errors in the Jacquelin, E., Bennani, A., Hamelin, P., 2003. Force reconstruction: analysis and
regularization of a deconvolution problem. J. Sound Vib. 265, 81–107.
reduced-order modal model. The magnitudes of the estimated loads are Jain, A., Jones, N.P., Scanlan, R.H., 1996. Coupled flutter and buffeting analysis of long-
higher than predicted directly from design wind field models. Buffeting span bridges. J. Struct. Eng. 122, 716–725.
and self-excited forces are both distinctly present, but these factors are Jakobsen, J., Tanaka, H., 2003. Modelling uncertainties in prediction of aeroelastic bridge
estimated jointly and cannot currently be separated. When compared behaviour. J. Wind Eng. Ind. Aerodyn. 91, 1485–1498.
Kang, N., Kim, H., Choi, S., Jo, Seongwoo, Hwang, J.-S., Yu, E., 2012. Performance
with the anemometer data, clear simultaneous trends between the time evaluation of TMD under typhoon using system identification and inverse wind load
series of the estimated modal forces and wind turbulence are observed. estimation. Comput. Aided Civ. Infrastruct. Eng. 27, 455–473.
Kavrakov, I., Morgenthal, G., 2017. A comparative assessment of aerodynamic models for
The results suggest that the methodology, if further developed and tested,
buffeting and flutter of long-span bridges. Engineering 3, 823–838.
could be useful in the characterization of wind loads on bridges with Ko, J., Ni, Y., 2005. Technology developments in structural health monitoring of large-
uncertain wind conditions. scale bridges. Eng. Struct. 27, 1715–1725.
Some practical and theoretical limitations lead to issues that should Lai, T., Yi, T.-H., Li, H.-N., Fu, X., 2017. An explicit fourth-order Runge–Kutta method for
dynamic force identification. Int. J. Struct. Stab. Dyn. 1750120.
be addressed. The number of modal forces that can be reconstructed is Larsen, A., Larose, G.L., 2015. Dynamic wind effects on suspension and cable-stayed
limited by the output data. Since long-span bridges have many modes bridges. J. Sound Vib. 334, 2–28.
15
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
Li, K., Liu, J., Han, X., Sun, X., Jiang, C., 2015. A novel approach for distributed dynamic Petersen, Ø.W., Øiseth, O., Lourens, E., 2019. Full-scale identification of the wave forces
load reconstruction by space–time domain decoupling. J. Sound Vib. 348, 137–148. exerted on a floating bridge using inverse methods and directional wave spectrum
Liu, Y., Shepard Jr., W.S., 2005. Dynamic force identification based on enhanced least estimation. Mech. Syst. Signal Process. 120, 708–726.
squares and total least-squares schemes in the frequency domain. J. Sound Vib. 282, Qiao, B., Zhang, X., Gao, J., Chen, X., 2016. Impact-force sparse reconstruction from
37–60. highly incomplete and inaccurate measurements. J. Sound Vib. 376, 72–94.
Lourens, E., 2012. Force Identification in Structural Dynamics. Ph.D. thesis. Katholieke Rezayat, A., Nassiri, V., De Pauw, B., Ertveldt, J., Vanlanduit, S., Guillaume, P., 2016.
Universiteit Leuven - Faculty of Engineering, Leuven (Belgium). Identification of dynamic forces using group-sparsity in frequency domain. Mech.
Lourens, E., Reynders, E., De Roeck, G., Degrande, G., Lombaert, G., 2012. An augmented Syst. Signal Process. 70, 756–768.
Kalman filter for force identification in structural dynamics. Mech. Syst. Signal Scanlan, R., 1978. The action of flexible bridges under wind, II: buffeting theory. J. Sound
Process. 27, 446–460. Vib. 60, 201–211.
Lourens, E., Papadimitriou, C., Gillijns, S., Reynders, E., De Roeck, G., Lombaert, G., 2012. Scanlan, R., 1978. The action of flexible bridges under wind, I: flutter theory. J. Sound
Joint input-response estimation for structural systems based on reduced-order models Vib. 60, 187–199.
and vibration data from a limited number of sensors. Mech. Syst. Signal Process. 29, Siedziako, B., Øiseth, O., Rønnquist, A., 2017. An enhanced forced vibration rig for wind
310–327. tunnel testing of bridge deck section models in arbitrary motion. J. Wind Eng. Ind.
Lystad, T.M., Fenerci, A., Øiseth, O., 2018. Evaluation of mast measurements and wind Aerodyn. 164, 152–163.
tunnel terrain models to describe spatially variable wind field characteristics for long- Solari, G., De Gaetano, P., Repetto, M.P., 2015. Thunderstorm response spectrum:
span bridge design. J. Wind Eng. Ind. Aerodyn. 179, 558–573. fundamentals and case study. J. Wind Eng. Ind. Aerodyn. 143, 62–77.
Ma, C.K., Chang, J.M., Lin, D.C., 2003. Input forces estimation of beam structures by an Song, W., 2018. Generalized minimum variance unbiased joint input-state estimation and
inverse method. J. Sound Vib. 259, 387–407. its unscented scheme for dynamic systems with direct feedthrough. Mech. Syst.
Macdonald, J.H., 2003. Evaluation of buffeting predictions of a cable-stayed bridge from Signal Process. 99, 886–920.
full-scale measurements. J. Wind Eng. Ind. Aerodyn. 91, 1465–1483. Sun, H., Feng, D., Liu, Y., Feng, M.Q., 2015. Statistical regularization for identification of
Maes, K., Lourens, E., Van Nimmen, K., Reynders, E., De Roeck, G., Lombaert, G., 2014. structural parameters and external loadings using state space models. Comput. Aided
Design of sensor networks for instantaneous inversion of modally reduced order Civ. Infrastruct. Eng. 30, 843–858.
models in structural dynamics. Mech. Syst. Signal Process. 52, 628–644. Wang, H., Hu, R., Xie, J., Tong, T., Li, A., 2012. Comparative study on buffeting
Maes, K., Gillijns, S., Lombaert, G., 2018. A smoothing algorithm for joint input-state performance of Sutong Bridge based on design and measured spectrum. J. Bridge
estimation in structural dynamics. Mech. Syst. Signal Process. 98, 292–309. Eng. 18, 587–600.
Naets, F., Cuadrado, J., Desmet, W., 2015. Stable force identification in structural Wong, K., 2004. Instrumentation and health monitoring of cable-supported bridges.
dynamics using Kalman filtering and dummy-measurements. Mech. Syst. Signal Struct. Control Health Monit. 11, 91–124.
Process. 50, 235–248. Xu, Y., Chen, J., 2004. Characterizing nonstationary wind speed using empirical mode
Niu, Y., Fritzen, C., Jung, H., Buethe, I., Ni, Y.-Q., Wang, Y.-W., 2015. Online decomposition. J. Struct. Eng. 130, 912–920.
simultaneous reconstruction of wind load and structural responses: theory and Zhi, L., Li, Q., Fang, M., 2016. Identification of wind loads and estimation of structural
application to Canton tower. Comput. Aided Civ. Infrastruct. Eng. 30, 666–681. responses of super-tall buildings by an inverse method. Comput. Aided Civ.
Petersen, Ø.W., Øiseth, O., 2019. Finite element model updating of a long span Infrastruct. Eng. 31, 966–982.
suspension bridge. In: Rupakhety, R., Olafsson, S., Bessason, B. (Eds.), Proceedings of Zhi, L., Yu, P., Li, Q.-S., Chen, B., Fang, M., 2018. Identification of wind loads on super-tall
the International Conference on Earthquake Engineering and Structural Dynamics. buildings by Kalman filter. Comput. Struct. 208, 105–117.
Springer International Publishing, pp. 335–344. Zhu, L., Xu, Y., Zhang, F., Xiang, H., 2002. Tsing Ma bridge deck under skew winds - Part
I: aerodynamic coefficients. J. Wind Eng. Ind. Aerodyn. 90, 781–805
16
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
KEKURANGAN :
Penilaian beban angin pada jembatan bentang panjang biasanya didasarkan pada model asumsi
mengenai medan angin serta koefisien aerodinamis yang diperoleh dari hasil pengujian di
terowongan angin. Namun, pendekatan ini seringkali menghadapi ketidakpastian yang signifikan.
Hasil penelitian terkini menunjukkan adanya perbedaan besar antara respons dinamis yang diprediksi
oleh model dengan respons yang diamati pada jembatan gantung akibat pengaruh angin.
KELEBIHAN :
Dalam artikel ini, beban angin dinamis pada jembatan Hardanger diestimasi berdasarkan data
percepatan. Kemampuan estimasi lapangan penuh dari algoritma-algoritma ini untuk jembatan
rentang panjang telah dipelajari sebelumnya, menunjukkan bahwa metodologi ini dapat dilakukan,
dengan batasan utama adalah jumlah respons modal yang dapat direkonstruksi.
PEMBEBANAN :
➢ Pembebanan jembatan adalah proses menentukan gaya-gaya yang bekerja pada struktur
jembatan. Dalam jurnal ini, terdapat dua pendekatan yang digunakan untuk pembebanan
jembatan: pendekatan analisis maju (forward analysis) dan pendekatan identifikasi balik
(inverse identification).
➢ Pendekatan analisis maju (forward analysis) didasarkan pada model angin yang menghasilkan
beban pada jembatan. Model ini menggunakan spektrum kecepatan angin dan koefisien
aerodinamika untuk menghitung beban angin pada jembatan. Pendekatan ini sering digunakan
dalam desain jembatan dan menghasilkan beban angin yang dinyatakan secara stokastik
dalam domain frekuensi
➢ Pendekatan identifikasi balik (inverse identification) adalah pendekatan yang digunakan
dalam jurnal ini. Pendekatan ini tidak mengasumsikan model beban tertentu, tetapi
menggunakan data percepatan yang diukur pada jembatan untuk mengestimasi gaya angin
yang berkontribusi pada respons jembatan. Metode identifikasi gaya angin ini melibatkan
pemodelan invers menggunakan data percepatan yang diukur dan model sistem yang
diketahui. Dalam metode ini, gaya angin diperoleh melalui estimasi vektor gaya modal
berdasarkan data percepatan yang diukur
STRUKTUR :
➢ Struktur jembatan yang dibahas dalam jurnal ini adalah Jembatan Hardanger di Norwegia.
Jembatan ini merupakan jembatan gantung dengan bentang utama sepanjang 1310 meter.
Jembatan ini terletak di daerah pegunungan dengan kondisi lingkungan yang kompleks.
➢ Jembatan Hardanger memiliki fitur bentang utama yang panjang dan ramping, sehingga
sensitif terhadap angin yang datang dari daerah pesisir di sebelah barat atau daerah
pegunungan di sebelah timur. Untuk memantau respons dinamis jembatan, sistem pemantauan
struktural telah dioperasikan sejak tahun 2013. Sistem ini terdiri dari anemometer ultrasonik
digital yang dipasang di hanger jembatan dan akselerometer triaksial yang terletak di balok
dan menara jembatan
PERENCANAAN STRUKTUR :
Tidak ditemukan
MANFAAT :
➢ Memahami karakteristik medan angin
Penelitian ini memberikan pemahaman yang lebih baik tentang karakteristik medan angin di
sekitar jembatan Hardanger di daerah pegunungan yang kompleks.
17
Ø.W. Petersen et al. Journal of Wind Engineering & Industrial Aerodynamics 196 (2020) 104045
CONTOH JEMBATAN :
18
JEMBATAN CABLE STAYED
Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct
Keywords: Accidental or extreme events can induce abnormal loads on structures contributing to a local damage of primary
Blast loading components or to the progressive collapse of the structure. Although hazards, which are usually difficult to be
Cable-stayed bridge predicted, increase the complexity in the design, the need to determine the consequences on structures of these
Dynamic analysis extreme events remains significant. This study aims to evaluate the damage induced by an accidental load such
Collapse mechanism as blast loading to a cable-stayed bridge. In the first part of the paper, through non-linear dynamic analyses, it
Blast fragility
was possible to evaluate the structural response of three configurations of cable-stayed bridges, then to figure out
the possible immediate and forthcoming damages considering different loading parameters and position. Finally,
a displacement-controlled static non-linear pushover analysis of the structure has been performed to evaluate the
limit state of the structure for a random position of blast loading and to predict the possible damage state during
an accidental event such as blast loading. The results indicate significant effects of the structural configuration of
the bridge and stay patterns. The Fan cable-stayed bridge is found to be the most effective among the considered
configurations in reducing the dynamic effect induced by the blast loading on the structure. In addition, the deck
is found to be less critical in terms of blast load mitigation close to the abutment and the pylon.
⁎ Corresponding author.
E-mail address: paolo.zampieri@dicea.unipd.it (P. Zampieri).
https://doi.org/10.1016/j.engstruct.2020.110739
Received 28 August 2019; Received in revised form 27 April 2020; Accepted 1 May 2020
01 41-0296 /© 202 0E lse
v ie
rL td.Allrightsreserved.
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
of the cable damage on the dynamic response of suspended bridges has relationship between load intensity and structural response, then be-
been investigated in [15–17]. In addition, a study of the dispropor- tween structural response and the possibility of the collapse of the
tionate collapse mechanism of an existing cable-stayed bridge with the bridge. Finally, an investigation of the likelihood of exceedance of the
aim to improve the robustness of the structure after a severe seismic structural capacity of the bridge’s structural components subjected to
event is found in [18]. Finally, a new system of cable-stayed bridges has blast loading will be presented. In the first part of the study, only the
emerged in the recent three decades. The under-deck cable-stayed cases of blast loading taking place on the deck will be considered.
bridges (UDCSB) present a better structural behavior, an economical Indeed, since vehicles are considered as the primary transportation
advantage and arguably strong aesthetic characteristics than the con- mean of the explosive item, the blast loading occurring is more sus-
ventional cable-stayed bridges [19]. A recent study showed that, the ceptible to impact the bridge deck than other structural elements. The
dynamic response of such bridges is influenced by the load speed, the document will be subdivided into three sections besides the introduc-
depth of the deck, as well as the structural section [20] and revealed tion and conclusion. A description of the bridge will be presented
how important an investigation of this innovative arrangement of the firstly, followed by the numerical modeling, numerical analysis, and
bridge under blast threats should become a matter of concern. discussion of the results and finally, a section on the blast fragility will
In a very small number of previous studies, structural safety against be presented. In the companion study, the authors will focus on actions
blast loading that can lead to progressive collapse has been assessed in a on the pylon and piers. Satisfactory results from the analyses show that
probability-based approach. Even though the studies are mainly fo- the Fan configuration presents a better resistance in the case of blast
cused on earthquake engineering, it is observed nowadays a significant loading event even though it displays the most rigid structure. Among
advance in the case of blast loading [21–25]. Many studies to define other things, the possibility of total structural failure would arise be-
fragility curves have been conducted either on structural elements cause the dynamic action of the load causes not only excessive damage
[26–28] or on buildings [29–32]. The study of fragility in the case of to the deck but also to the stay cables adjacent to the impact area.
bridges subjected to blast loading is almost non-existent in the scientific
literature. Indeed, many authors have rather addressed their researches
to study the structural response or performance of bridges [33,34]. In 2. FEM analysis
particular, Hashimi et al. [35], Tang and Hao [25] presented a sa-
tisfactory research on the structural performance of the bridge’s com- 2.1. Bridge benchmark Description
ponents for a cable-stayed bridge subjected to blast loading throughout
a dynamic analysis. The first authors focused on the variation of the The cable-stayed bridge is nowadays widely used around the world
blast loading location on the bridge deck and studied also the cable loss owing to the great advantages it displays for long-span bridges. A
mechanism after a blast event while the second document is con- generic cable-stayed bridge consists on deck, pylons and inclined cables
centrated on the influence of the size of the blast loading source, the known as stays, which serve as internal support and provide additional
stand-off distance and presented also the effectiveness of FRP vertical stiffness to the deck. In this study, three different typologies of
strengthening techniques after blast loading. cable-stayed bridge namely fan, semi-fan and harp cable-stayed bridges
The review reveals that abnormal loads can lead to the critical da- have been considered. They differ mainly in the pattern of the cable
mage of structural components that could reach adjacent structural stays. A generic bridge under this study indeed consists of two lateral
elements and ultimately cause a collapse of the structure. Based on this spans of 99.2 m long and one central span of 204.6 m long. The bridge
observation, the authors have found a clear motivation to focus their deck is made of 1.15 m deep and 13.0 m wide steel orthotropic boX
work on the study of the structural performance of cable-stayed bridges girder, which includes three traffic lanes and two footways. It consists
subjected to blast loads. However, the stay patterns might play an im- of a closed hexagonal steel boX section with 20 mm thick flanges and
portant role in the diffusion of the damage between the bridge’s com- webs. In addition, trapezoidal stiffeners are installed on all the section
ponents and should be investigated in depth. Thus, the purpose of this of the deck spaced 350 mm from center to center while vertical stif-
research is to establish direct or indirect links between the collapse feners each 2750 mm were provided in the deck section to limit ex-
capacity of the structural elements of various cable-stayed bridges cessive displacement and increase the buckling resistance. Rectangular
(Harp, Fan, and Semi-fan stay patterns) subjected to abnormal loads closed boX sections have also been considered for the pylon, the
such as blast loading and the possible collapse of the bridge. To reach transverse beam, and the base pylon. The pylon is considered as the
our objectives, the structural performance of three different cable- vertical element above the steel deck whereas the base pylon is taken as
stayed patterns will be assessed thoroughly and only the dynamic effect a vertical element below the steel deck. Two transverse beams are used
deriving from a blast loading (neglecting the air blast released and its to connect the vertical elements and form with the vertical element the
consequences on cars’ glazing and people to simplify the study) will be “H” shaped pylon. The geometry and geometrical characteristics of the
considered. In particular, intensive static and dynamic analyses have structural elements are given in Fig. 1 and Table 1.
been carried out on various cable-stayed bridges in order to determine a
Fig. 1. Geometry and Section of structural elements of the cable-stayed bridge. a) Closed boX section for pylon, transverse beam and base pylon b) Deck section.
2
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Table 1
Dimensions of the structural elements of the cable-stayed bridge.
3
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 3. Geometrical characteristic of the bridges. a) Harp bridge; b) Fan bridge; c) Semi-Fan bridge; d) Longitudinal profile of the bridge; e) Transversal section of the
pylon from the left to the right Harp, Fan and Semi-fan system.
4
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Table 2 for equivalent weight lower than 500 kg as shown in Table 3. Referring
Material characteristics of the structural members. to the strain analysis, the main difference is displayed for W = 1000 kg.
Steel grade Structural Constitutive Yield stress Failure
Indeed, in this particular case, the maximum strain reached the plastic
element model [MPa] strain strain limit for Harp configuration while for other stays patterns, it
remains below the plastic strain limit. Finally, the structural component
S420 Deck Elasto-plastic 420 5%
subjected to the blast loading reaches quickly the allowable stresses just
S355 Pylon, Elasto-plastic 355 4%
Transverse beam
after detonation. Indeed, in the case studied, all the equivalent weights
SWRH72B Cable Elastic 1023 4.8‰ of TNT considered succeed in creating stress reaching the stress limit at
a very short time step (see Fig. 4).
However, in this case, the Harp configuration seems to dissipate the
Moreover, C = 40.4 and q = 5 for low strength steel while C = 3500 dynamic action better than the other two because the stress fluctuates
and q = 5 for high strength steel will be considered as defined in [41]. less. These fluctuations are visible only on the negative phase of the
d action as shown in Fig. 4 and may accelerate secondary damage such as
f yd
= 1.0 + ⎛ ε ̇⎞
1/q
fatigue that would result from a large variation in the internal stresses
fyd ⎝C ⎠ (4) of the element.
Finally, with a random stand-off distance, the displacement, stress
and strain variation in the section close to a blast loading increase by
2.5. Load analysis increasing the dimension of the blast loading source. Indeed, the da-
maging effects in the bridge structural response induced after blast
In this paper, blast loads are considered as the leading variable load loading increase as the equivalent weight of TNT increases.
whereas traffic loads are taken as accompanying actions. The self- It is noteworthy that assessing the internal strain state of the section
weight is directly provided from the software and it is a function of the has a critical issue. Indeed, following the blast event, the admissible
material density and the characteristics of the section. A uniformly stress is quickly reached, but its consequence is critical if the state of
distributed load G2 = 48.7 kN/m is applied along the deck which re- internal strain is higher than the maximum plastic strain. Although the
presents asphalt layer, safety barrier, and waterproofing layer. The pre- strain state seems high for the three configurations studied with most of
tension force on the stays is calculated throughout an optimization them exceeding the yielding strain, the worst situation is reached for
process in order to compensate 95% of the permanent loads with the the equivalent mass of the TNT equal to or greater than 1500 kg.
maximum displacement limited to 10 mm. Traffic loads are defined Therefore, the maximum plastic strain is exceeded with the possible
according to European norms [42]. To reach our goal, load combina- breakage of the deck at this position.
tions for accidental design situations as defined in [43] were used Later, two other positions of occurrence of blast loading along the
where the partial coefficients are given within the code. deck at x = 68.2 m and x = 201.5 m (Mid-span) were studied (Fig. 5
and Fig. 6). It is observed that the displacement at time history t = 0 ms
3. Numerical analysis and discussion is 192 mm at the mid-span (Fig. 6a) which are from traffic and per-
manent loads. In addition, the displacements following blast loading
Owing to the lack of existing experimental models of such kind of are more relevant for an event occurring in mid-span as the outcome of
bridges, it was not able to compare the preliminary static analysis to comparing the results of Fig. 5a and Fig. 6a for any equivalent TNT
experimental results. The first stage of analysis consisted of determining weight.
the actual stress of the bridge in the traffic condition. Then, a series of In terms of strain, it has been observed that for a blast loading oc-
dynamic non-linear analyses have been carried out on the three dif- curring in the mid-span, the consequences are much more significant,
ferent typologies of cable-stayed bridges under study. The nonlinear because, before the stress generated by a blast load, the deck already
dynamic analysis of the blast action is performed on bridge models in has a significant level of stress and strain due to permanent and traffic
the stressed and deformed shape from the permanent and traffic loads. loads. As a matter of fact, in this particular case, from equivalent loads
It should be added that the implicit approach of the FEM software greater than 500 kg, the plastic strain limit would be reached as shown
package was used in the analysis. However, with this technique, the in Fig. 6b. When the event takes place at x = 68.2 m, the internal strain
onset of the element deletion is always difficult to implement within the of the deck section is higher (Fig. 5b) than if the action takes place at
finite element analysis. The use of an explicit solver method may x = 6.2 m. However, it would require actions with an equivalent
overcome the situation even though displaying disproportionate results weight greater than 1000 kg to cause serious damages.
compared to the real situation. Initially, a direct comparison is made Finally, the stress conditions are quite similar in that the blast event,
between the results of displacements, strains, and stresses obtained for which strongly contributes after the deflagration, to the achievement of
the three types of configurations. Then, the most unfavorable config- yielding stress in a short time. The only difference is observed in the
uration in terms of stress is studied individually, varying different lo- considerable fluctuation of the stress, especially for the action taking
cations of the impact with the deck. place in the mid-span for low-intensity loads as shown in Fig. 6c.
A detailed analysis of the deck revealed the possibility of a gradual
3.1. Stress analysis collapse that could be caused by the dynamic action of a blast load. In
the case of a cable-stayed bridge, progressive collapse can be caused by
From the first analyses and considering the global reference system the loss of two or more cables [4]. For an extreme event such as blast
of the bridge defined in Fig. 3, the following observations are made. For loading, the loss of cables may be due either to the consequences of an
the blast loading occurring at x = 6.2 m from the abutment, and the air blast or to the reflected pressure near the stays. Other reasons could
stand-off distance R = 1 (see Fig. 4), the maximum displacements be the reaching of the excessive tension limit of the cable or finally to
reached by the deck in a short time history after the blast detonation is the collapse of the deck that would cause the cables near the failure
summarized in Table 3. The results show that the displacement reaches zone to break. In this investigation, the last two cases were considered.
intolerable levels for an equivalent weight greater than 750 kg. In The first case studied considers the Fan configuration where the
particular, the maximum displacement of the deck is observed for the blast loading occurs close to the abutment. The results show a limited
Harp configuration regardless of the size of the blast loading source. It displacement of the deck at the impact area, a steady condition of the
is also found that the difference in the structural response is obvious for deck when moving far from the impact area, a torsional buckling mode
the weight larger than 750 kg while the displacement is quite similar similar to the case of Fig. 7a. A higher dynamic sensitivity in response
5
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 4. Structural response of the three cable configuration in the case of blast at X = 6.2 m. From the left to the right displacement, Strain and Stress of the bridge’s
deck.
Table 3 to blast action is observed for Harp and Semi-fan bridges as displayed in
Maximum vertical displacement of the deck under a blast loading at stand-off Fig. 7b&c. However, in the Fan configuration, the blast action close to
R = 1 m. the abutment could result in the loss of a single cable, especially the one
Equivalent TNT weight Cable-stayed bridge configurations affected by the torsion of the deck (Fig. 7a).
Harp Fan Semi-fan The analysis also revealed that in the case of Harp bridges (Fig. 7b&
d), the dynamic action creates a torsional buckling of the deck, which
W = 250 kg — 2678 mm — 1774 mm — 2109 mm
W = 500 kg — 2297 mm — 1489 mm — 1728 mm
may result in a reduction of its structural capacity. In these cases, two
W = 750 kg — 1989 mm — 1286 mm — 1508 mm cables are likely to be damaged for the impact at x = 6.2 m (Fig. 7b)
W = 1000 kg — 1421 mm — 1126 mm — 1298 mm whereas three cables are likely to be damaged for the impact at
W = 1500 kg — 406 mm — 331 mm — 385 mm x = 68.2 m (Fig. 7d), particularly because they will have reached their
ultimate stress. This is less relevant because the action lasts for a short
Fig. 5. Structural response of the Harp pattern with blast at X = 62.8 m and R = 1 m. a) Displacement b) Strain c) Stress.
6
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 6. Structural response of the Harp pattern with blast at X = 201.5 m and R = 1 m. a) Displacement b) Strain c) Stress.
period and the damage will only be confirmed by repeated similar ac- 4. Fragility analysis
tions. In the case where the action is caused by a large mass (Fig. 7e),
the consequences are more severe because a total breakage of the deck From the results of the analysis performed so far, it is clear that the
element around the impact area is observed causing a loss of the four consequence of blast loading can be extreme for the deck and therefore
cables adjacent to the detonation zone. Finally, besides the global in- for the whole bridge through the progressive collapse mechanism. For
stability of the deck observed after a blast event, an evaluation of the this reason, it is also important to associate the structural response of
internal distribution of stresses presents a possible local instability of the bridge deck to the likelihood of exceedance of capacity thresholds of
the section of the deck around the impact zone. In particular, for a the deck elements during a blast event.
significant blast loading, a combination of positive and negative stress
was observed throughout the section as shown in Fig. 8b. For less re-
levant blast loading, the upper part of the deck is likely to be in tension 4.1. Methodology
whereas the lower part being compressed Fig. 8a.
The common way to study the structural performance under a non-
Fig. 7. Blast loading analysis in cable-stayed bridge. a) W = 500 kg, b) W = 500 kg, c) W = 500 kg, d) W = 500 kg, e) W = 1500 kg.
7
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 9. Geometrical characteristics of the bridges. From the left to the right: Inverted Y-shaped profile; H-shaped profile, A-shaped profile.
8
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 10. Flowchart for the capacity and demand assessment of cable-stayed bridges.
Using the first-order reliability approach that takes into account the structural component of the bridge. In each case, the structural capacity
first and second moments of individual distributions, Eq. (6) is simpli- is assessed on a stock of bridges that vary in dimensions and the cross-
fied into Eq. (7). section of the structural elements. The same is also true for the struc-
tural demand, which is evaluated through nonlinear dynamic analysis
⎡ ln(μ̄D /μ̄C ) ⎤ for different values of scaled distance, taken in this case as intensity
PFj = Φ ⎢ 2
σ ln D + σ2ln C ⎥ (7) measure.
⎣ ⎦
Where μ̄D = median estimate of the structural demand 4.4. Structural capacity
andμ̄C = median estimate of the structural capacity.σ 2ln D = lognormal
standard deviation of the structural demand andσ2ln C = lognormal In the capacity study, more than thirty models were studied, varying
standard deviation of the structural capacity. Cornell et al., [54] de-
the geometry of the bridge, the characteristics of the material, the pylon
rived the relationship between the estimate of the median and the in-
profiles (see Fig. 9), and the cross-sectional area of the deck. Then, a
tensity of the actions as shown in Eq. (8).
series of nonlinear pushover analyses of the bridge in the vertical di-
μ̄ = aI b or ln(μ̄ ) = ln a + b ln(I j ) (8) rection were performed, which corresponds indeed to the principal
D j D
direction of the damage of the deck after a blast event. From the ana-
The methodology adopted by the authors for the derivation of the lyses, ideal capacity curves were derived from the various pushover
fragility curves is outlined in the flowchart (see Fig. 10). It uses the performed leading to the definition of an envelope, which describes the
component-based approach in which the fragility curve of the system set of capacity obtained from the bridge deck. Then, through a statis-
results from the unconditional combination of the fragility of each tical tool it was possible to highlight the mean capacity curve of each
9
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 11. Nonlinear static analysis of the cable-stayed bridges. a) Envelope of capacity curves of the bridges’ deck; b) Bilinear of the Mean Capacity curve.
Table 6
Statistic properties for structural parameters of cable-stayed bridges.
μ̄C σC μδ
cable stays patterns (Fig. 11a and Fig. 11b). The set of performance 4.5. Structural demand
displays a hardening behavior with a displacement ductility (Eq. (9))
varying from 2.6 to 4.2. Afterward, the same process was done for the In the analysis of the demand, a set of various blast intensities were
other two stays patterns and the statistical parameters of mean per- considered with the possibility of occurring at truly random positions
formance were collected and presented in Table 6. along the longitudinal and transversal direction of the bridge deck
The Harp configuration presents the most flexible in all with the (Fig. 13). It is worth recording that the hazard uncertainties namely the
highest ductility. On the other hand, the Fan configuration, which had a explosive mass and the stand-off distance and highlighted in Section 4.1
good structural response in terms of vertical displacement, as presented will follow a normal distribution (Fig. 13) as advised in [26]. Finally,
in the previous section, is the most rigid and reduces the structure's the longitudinal position was limited to half of the bridge due to the
ability to achieve disproportionate vertical displacements. symmetry and in the transversal direction, the random location of the
blast event was considered. It has been considered a wide range of in-
δv,u
μδ = tensity measure by varying the equivalent TNT weight and stand-off
δ v,y (9) distance.
The weight of an explosive is generally estimated by taking into
Where δv,u and δv,y are the mean value of ultimate and yielding vertical
account a relevant attack scenario, which would involve a vehicle or
displacement after a nonlinear pushover in vertical direction.
improvised explosive device carried by one or more individuals. The
stand-off distance in the analysis of the robustness of the bridge deck is
4.4.1. Limit states and damage index taken accordingly to the number of explosives considered. It is taken as
In the fragility analysis, structural damage after a hazard is a the distance between the position of the center of the weapon and its
function of both intrinsic and extrinsic characteristics of the structure. vertical projection on the deck, which is the shortest path to reach the
In general, it is possible to study the undergone damage by classifying deck, thus causing the most intense impact on the deck. Therefore, the
them according to the damage index [55]. This index includes the weight of the explosive is taken in the range 0–10000 kg and the stand-
structural response observed and the definition of limit states. The limit off distance will be taken in the range 0.2–4 m.
states often derive from experimental campaigns or computational re- The results from the analysis have been displayed in the lognormal
sources. Regardless of the method used, the damage indexes are often scale as advised in [44] and the mean value of the vertical displacement
defined by means of curvature ductility, rotation ductility, displace- has been expressed in terms of scaled distance Z and displayed in
ment ductility, etc. Indeed, several authors [5,26,39] considered two Fig. 14. The results show that the lognormal distribution of the vertical
parameters as failure criterion under accidental loads: the rotation displacement follows the linear regression with a low dispersion rate.
angle (θ) of the element’s end support and the displacement ductility The regression line for each bridge model confirms the outcome that the
(μδ) [26]. Harp bridge model displays larger vertical displacement than the two
The pushover analysis performed allowed defining the limit state for other bridge models with a higher dispersion of data. Finally, the ne-
the study as presented in Fig. 12. In particular, five regions were gative slope is mainly due to the fact that the impact on the bridge deck
adopted as highlighted in [56]: A – B: Elastic Range, B – C: Immediate is more severe by decreasing the scale distance, which is obtained either
10
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Table 7
Limit state for cable-stayed bridges under Blast loading [56]
Limit State (LS) Threshold values of ductility Structural performance registered
by decreasing the stand-off distance or by increasing the weight the bridge not to reach the limit states at the same trend as the other two
TNT. configurations.
11
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
Fig. 14. Prediction of blast demand in cable-stayed bridges: a) Harp bridge model; b) Fan bridge model; c) Semi-fan bridge model.
have excessive displacements leading to the collapse of the deck and TNT, by 45% for a mass of 500 kg (Fig. 15).
subsequently of the bridge, the TNT load must be around 1000 kg. It is On the other hand, the variability is not as important for other
important to note that a quantity of TNT lower than those mentioned cable-stayed bridge configurations. Indeed, considering the same stand-
above will produce similar or advanced damage for Harp and Semi-fan off distances as defined above, there is observed in a Harp configura-
configurations. tion, a reduction in the probability of collapse of 20% for a mass of
Contrary to the influence of the mass of TNT, which is less im- 1000 kg and 30% for a mass of 500 kg. This reduction will be in the
portant from what emerges in the fragility curves, the stand-off distance order of 25% and 37% respectively for the 1000 kg and 500 kg masses
has a more significant impact. It is important to highlight that the of the Semi-fan configuration.
fragility results indicate that as the detonation occurs far from the deck, To calculate the limit state probability of exceedance as presented, a
the probability of a possible failure of the bridge deck becomes lower. random distribution of blast loading scenario was considered. In addi-
Indeed, by isolating two particular cases of stand-off distance tion, the quantity of TNT ranging from 0 − 10000 kg was considered
R = 0.4 m and R = 0.8 m in a Fan configuration, we have the prob- but in reality, it is almost impossible to have an attack with a very large
ability of collapse of the deck is reduced by 30% for a mass of 1000 kg quantity of explosives because the transport would require significant
12
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
logistical means that would reveal a suspicion on the part of the in- relationship in a lognormal distribution describes a negative corre-
telligence services. Thus, for masses varying from 50 kg − 300 kg, lation between the two parameters involved. Thus, the con-
which still represents quantities that would go unnoticed, it is very sequences following a blast loading action are more serious when
likely that there is no risk of collapse because the deck will succeed in the scaled distance is reduced.
withstand the loads caused by the explosion.
• In this study, the authors also presented the need for an internal
strain study of the sections. Indeed, in a case where stress analysis
5. Conclusion reveals the ease with which bridge sections reach their yielding and
those even for small masses, it is important to know if the action at
The structural efficiency of cable-stayed bridges considering three the limit will succeed in creating a collapse of the section.
different typologies, namely (Harp, Fan, and Semi-fan bridges).
Initially, a study of the structural response of bridges subjected to blast
• The study highlighted the feasibility and importance of the potential
contribution of the derivation of the fragility curves of cable-stayed
loadings of different intensities was carried out. Then, a study to bridges. Through a probabilistic approach, the performance-based
highlight the consequences that a blast event could have on the possi- design for blast resistant structures are made and the likelihood of
bility of progressive collapse. Finally, an analysis taking into account collapse for structural elements of the bridge is presented although
the probabilistic parameters was done on the structural components in this practice is commonly used on structures subjected to such ha-
order to find the likely failure. In the study, a series of non-linear dy- zards as (earthquake, flood, hurricane, wind, etc…).
namic analyses were used to find the structural demand of bridges
considering different intensity measures while in the case of structural
• In the common case of a fragility analysis for an earthquake, gen-
erally the structural elements at high risk of exposure are piers,
capacity, a pushover analysis was required to vary the geometry of the abutments and bearing. On the other hand, in the case of a blast
bridge and the cross-section of the structural components. From all event, the vulnerability of the deck must also be taken into account
these investigations, the following findings emerged: to the extent that this structural element remains highly exposed to
action. Finally, the major difficulty that lies in the probabilistic
• The Harp configuration presents a better configuration in terms of
structural response under the action of blast loading. In addition,
analysis of this type of event consists in determining the hazard
function since blast loadings are considered as a low occurrence
excessive displacement can be reached in the case of a large quantity probability event.
of TNT used and the blast loading easily causes partial or total
yielding of the bridge section closest to the point of detonation.
• Nonlinear dynamic analysis of a wide variety of bridges has estab- Declaration of Competing Interest
lished a relationship between structural demand following an ex-
treme Hazard such as blast loading and scaled distance. This The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to
13
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
influence the work reported in this paper. suspension bridges. J Sound Vib 2014;333(11):2404–21. https://doi.org/10.1016/
j.jsv.2014.01.027.
[14] Comanducci G, Ubertini F, Materazzi AL. Structural health monitoring of suspen-
References sion bridges with features affected by changing wind speed. J Wind Eng Ind
Aerodyn 2015;141:12–26. https://doi.org/10.1016/j.jweia.2015.02.007.
[1] Jenkins BM, Gersten LN. Protecting Public Surface Transportation Against [15] Materazzi AL, Ubertini F. Eigenproperties of suspension bridges with damage. J
Terrorism and Serious Crime : Continuing Research on Best Security Practices. Sound Vib 2011;330(26):6420–34. https://doi.org/10.1016/j.jsv.2011.08.007.
2001;(September):124. [16] Lonetti P, Pascuzzo A. A numerical study on the structural integrity of self-anchored
[2] Kumar A, Matsagar V. Blast Fragility and Sensitivity Analyses of Steel Moment cable-stayed suspension bridges. Frat ed Integrita Strutt 2016;10(38):359–76.
Frames with Plan Irregularities. Int J Steel Struct 2018;18(5):1684–98. https://doi. https://doi.org/10.3221/IGF-ESIS.38.46.
org/10.1007/s13296-018-0077-z. [17] Lonetti P, Pascuzzo A. Vulnerability and failure analysis of hybrid cable-stayed
[3] Son J, Lee HJ. Performance of cable-stayed bridge pylons subjected to blast loading. suspension bridges subjected to damage mechanisms. Eng Fail Anal
Eng Struct 2011;33(4):1133–48. https://doi.org/10.1016/j.engstruct.2010.12.031. 2014;45:470–95. https://doi.org/10.1016/j.engfailanal.2014.07.002.
[4] Wolff M, Starossek U. Cable-loss analyses and collapse behavior of cable-stayed [18] Domaneschi M, Cimellaro GP, Scutiero G. Disproportionate collapse of a cable-
bridges. 2013:423-423. http://doi.org/10.1201/b10430-326. stayed bridge. Bridge Eng 2019;172(1):13–26. https://doi.org/10.1680/jbren.18.
[5] Formisano A, Landolfo R, Mazzolani FM. Robustness assessment approaches for 00031.
steel framed structures under catastrophic events. Comput Struct 2015;147:216–28. [19] Ruiz-Teran AM, Apracio AC. Response of under-deck cable-stayed bridges to the
https://doi.org/10.1016/j.compstruc.2014.09.010 . accidental breakage of stay cables. Eng Struct 2009;31:1425–34. https://doi.org/
[6] Elsanadedy HM, Almusallam TH, Alharbi YR, Al-Salloum YA, Abbas H. Progressive 10.1016/j.engstruct.2009.02.027.
collapse potential of a typical steel building due to blast attacks. J Constr Steel Res [20] Madrazo-Aguirre F, Ruiz-Teran AM, Wadee MA. Dynamic behaviour of steel–con-
2014;101:143–57. https://doi.org/10.1016/j.jcsr.2014.05.005. crete composite under-deck cable-stayed bridges under the action of moving loads.
[7] Olmati P, Brando F, Gkoumas K. Robustness Assessment of a Steel Truss Bridge Eng Struct 2015;103:260–74. https://doi.org/10.1016/j.engstruct.2015.09.014 .
2013;2013:250–61. https://doi.org/10.1061/9780784412848.023 . [21] Sanaz R, Armen DK. A stochastic ground motion model with separable temporal and
[8] Abruzzo J, Matta A, Panariello G. Study of Mitigation Strategies for Progressive spectral nonstationarities. Earthq Eng Struct Dyn 2012;41(11):1549–68. https://
Collapse of a Reinforced Concrete Commercial Building. J Perform Constr Facil doi.org/10.1002/eqe.
2006;20(4):384–90. https://doi.org/10.1061/(asce)0887-3828(2006)20:4(384). [22] Nassr AA, Razaqpur AG, Tait MJ, Campidelli M, Foo S. Strength and stability of steel
[9] Song BI, Sezen H. E Xperimental and analytical progressive collapse assessment of a beam columns under blast load. Int J Impact Eng 2013;55:34 –48. https://doi.org/
steel frame building. Eng Struct 2013;56:664–72. https://doi.org/10.1016/j. 10.1016/j.ijimpeng.2012.11.010.
engstruct.2013.05.050. [23] Mosoarca M, Anastasiadis A, Formisano A. New factors for the seismic vulnerability
[10] Abdelwahed B. A Review on Building Progressive Collapse, Survey and Discussion. assessment of reinforced concrete buildings. Int J Sustain Mater Struct Syst
Case Stud Constr Mater 2019;11:e00264https://doi.org/10.1016/j.cscm.2019. 2016;2(3/4):222. https://doi.org/10.1504/ijsmss.2016.078708.
e00264. [24] Lange D, Devaney S, Usmani A. An application of the PEER performance based
[11] Adam JM, Parisi F, Sagaseta J, Lu X. Research and practice on progressive collapse earthquake engineering framework to structures in fire. Eng Struct
and robustness of building structures in the 21st century. Eng Struct 2014;66:100–15. https://doi.org/10.1016/j.engstruct.2014.01.052 .
2018;173(March):122–49. https://doi.org/10.1016/j.engstruct.2018.06.082. [25] Tang EKC, Hao H. Numerical simulation of a cable-stayed bridge response to blast
[12] Das R, Pandey AD, Soumya Mahesh MJ, Saini P, Anvesh S. Progressive Collapse of a loads, Part I: Model development and response calculations. Eng Struct
Cable Stayed Bridge. Procedia Eng 2016;144:132–9. https://doi.org/10.1016/j. 2010;32(10):3180–92. https://doi.org/10.1016/j.engstruct.2010.06.007 .
proeng.2016.05.016. [26] Olmati P, Petrini F, Gkoumas K. Fragility analysis for the Performance-Based Design
[13] Ubertini F. Effects of cables damage on vertical and torsional eigenproperties of of cladding wall panels subjected to blast load. Eng Struct 2014;78:112 –20. https://
doi.org/10.1016/j.engstruct.2014.06.004.
14
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
[27] Hao H, Stewart MG, Li Z-X, Shi Y. RC Column Failure Probabilities to Blast Loads. [42] European Standard. Part 2: Traffic loads on bridges. Eurocode 1 Actions Struct.
Int J Prot Struct 2010;1(4):571–91. https://doi.org/10.1260/2041-4196.1.4.571. 2003;3(1991-2:2003).
[28] Momeni M, Hadianfard MA, Bedon C, Baghlani A. Numerical damage evaluation [43] CEN. EN 1990: 2002 - Basis of structural design. 2002:1-87.
assessment of blast loaded steel columns with similar section properties. Structures [44] Pang Y, Wu X, Shen G, Yuan W. Seismic Fragility Analysis of Cable-Stayed Bridges
2018;2019(20):189–203. https://doi.org/10.1016/j.istruc.2019.04.002. Considering Different Sources of Uncertainties. J Bridg Eng 2013;19(4):04013015.
[29] Olmati P, Petrini F, Bontempi F. Numerical analyses for the structural assessment of https://doi.org/10.1061/(asce)be.1943-5592.0000565.
steel buildings under explosions. Struct Eng Mech 2013;45(6):803–19. https://doi. [45] Stefanidou SP, Kappos AJ. Bridge-Specific Fragility Analysis: When Is It Really
org/10.12989/sem.2013.45.6.803. Necessary? 2019;Vol 17. https://doi.org/10.1007/s10518-018-00525-9.
[30] Khan S, Saha SK, Matsagar VA, Hoffmeister B. Fragility of Steel Frame Buildings [46] Xie Y, Zhang J, DesRoches R, Padgett JE. Seismic fragilities of single-column
under Blast Load. J Perform Constr Facil 2017;31(4):04017019. https://doi.org/10. highway bridges with rocking column-footing. Earthq Eng Struct Dyn
1061/(asce)cf.1943-5509.0001016. 2019;48(7):843–64. https://doi.org/10.1002/eqe.3164.
[31] Kelliher D, Sutton-Swaby K. Stochastic representation of blast load damage in a [47] van de Lindt JW, Taggart M. Fragility Analysis Methodology for Performance-Based
reinforced concrete building. Struct Saf 2012;34(1):407 –17. https://doi.org/10. Analysis of Wood-Frame Buildings for Flood. Nat Hazards Rev 2009;10(3):113–23.
1016/j.strusafe.2011.08.001. https://doi.org/10.1061/(asce)1527-6988(2009)10:3(113).
[32] Ding Y, Song X, Zhu HT. Probabilistic progressive collapse analysis of steel-concrete [48] Herbin AH, Barbato M. Fragility curves for building envelope components subject to
composite floor systems. J Constr Steel Res 2017;129:129–40. https://doi.org/10. windborne debris impact. J Wind Eng Ind Aerodyn 2012;107–108:285–98. https://
1016/j.jcsr.2016.11.009. doi.org/10.1016/j.jweia.2012.05.005.
[33] Andreou M, Kotsoglou A, Pantazopoulou S. Modelling Blast Effects on a Reinforced [49] Casciati F, Cimellaro GP, Domaneschi M. Seismic reliability of a cable-stayed bridge
Concrete Bridge. Adv Civ Eng 2016 2016. https://doi.org/10.1155/2016/4167329. retrofitted with hysteretic devices. Comput Struct 2008;86:1769–81. https://doi.
[34] Hao H, Tang EKC. Numerical simulation of a cable-stayed bridge response to blast org/10.1016/j.compstruc.2008.01.012.
loads, Part II: Damage prediction and FRP strengthening. Eng Struct [50] Perotti F, Domaneschi M, De Grandis S. The numerical computation of seismic
2010;32(10):3193–205. https://doi.org/10.1016/j.engstruct.2010.06.006. fragility of base-isolatedNuclear Power Plants buildings. Nucl Eng and Des
[35] Hashemi SK, Bradford MA, Valipour HR. Dynamic response of cable-stayed bridge 2013;262:189–200. https://doi.org/10.1016/j.nucengdes.2013.04.029.
under blast load. Eng Struct 2016;127:719–36. https://doi.org/10.1016/j. [51] De Grandis S, Domaneschi M, Perotti F. A numerical procedure for computing the
engstruct.2016.08.038. fragility of NPP components under random seismic excitation. Nucl Eng and Des
[36] Primer FEMA. for Design of Commercial Buildings to Mitigate Terrorist Attacks. 2009;239:2491–9. https://doi.org/10.1016/j.nucengdes.2009.06.027 .
Risk Manag Ser 2003. [52] Simões da Silva L, Rebelo C, Nethercot D, Marques L, Simões R, Vila Real PMM.
[37] Deng RB, Jin XL. Numerical simulation of bridge damage under blast loads. WSEAS Statistical evaluation of the lateral-torsional buckling resistance of steel I-beams,
Trans Comput 2009;8(9):1564–74. Part 2: Variability of steel properties. J Constr Steel Res 2009;65(4):832–49.
[38] Calculation of Blast Loads for Application to Structural Components. Administrative https://doi.org/10.1016/j.jcsr.2008.07.017.
Arrangement No JRC 32253-2011 with DG-HOMEActivity A5 - Blast Simulation [53] Olmati P, Vamvatsikos D, Stewart MG. Safety factor for structural elements sub-
Technology Development.; 2013. http://doi.org/10.2788/61866. jected to impulsive blast loads. Int J Impact Eng 2017;106:249–58. https://doi.org/
[39] Verma SV. Blast Resistant Design of Structure. Int J Res Eng Technol 10.1016/j.ijimpeng.2017.04.009.
2016;04(25):64–9. https://doi.org/10.15623/ijret.2015.0425010 . [54] Cornell CA, Jalayer F, Hamburger RO, Foutch DA. Management Agency Steel
[40] Longitudinal Seismic Behavior of a Single-Tower Cable-Stayed Bridge Subjected to Moment Frame Guidelines. J Struct Eng 2002;2002(128):526–33. https://doi.org/
Near-Field Earthquakes. Shock Vib. 2017;2017. http://doi.org/10.1155/2017/ 10.1061/(ASCE)0733-9445(2002)128:4(526).
1675982. [55] Ding Y, Song X, Zhu HT. Probabilistic progressive collapse analysis of steel frame
[41] Paik JK, Thayamballi AK. A concise introduction to the idealized structural unit structures against blast loads. Eng Struct 2017;147:679–91. https://doi.org/10.
method for nonlinear analysis of large plated structures and its application. Thin- 1016/j.engstruct.2017.05.063.
Walled Struct 2003;41(4):329–55. https://doi.org/10.1016/S0263-8231(02) [56] FEMA. Prestandard and Commentary for the Seismic Rehabilitation of Buildings.
00113-1. Rehabil Requir. 2000;(1):1–518.
15
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2
020
)11
073
9
KEKURANGAN :
Kejadian tak terduga atau peristiwa ekstrem bisa menyebabkan beban yang tidak normal pada struktur, yang dapat
menyebabkan kerusakan pada komponen utama atau bahkan keruntuhan bertahap dari struktur itu sendiri. Meskipun
bahaya-bahaya ini biasanya sulit diprediksi, pentingnya untuk menentukan dampaknya pada struktur akibat peristiwa-
peristiwa ekstrem tersebut tetap besar.
KELEBIHAN :
- Penelitian ini bertujuan untuk mengevaluasi kerusakan yang disebabkan oleh beban tak terduga seperti ledakan
pada jembatan kabel. Penelitian ini mengisi kekosongan dalam literatur ilmiah terkait dengan analisis struktural
jembatan kabel tahan ledakan. Meskipun ada beberapa penelitian sebelumnya yang telah dilakukan dalam bidang
keamanan struktural terhadap pembebanan ledakan, penelitian ini fokus pada analisis struktural jembatan kabel
tahan ledakan dengan mempertimbangkan variasi beban ledakan yang berbeda, seperti lokasi dan intensitas beban
ledakan.
- Dengan adanya penelitian ini, para perancang dan insinyur struktur dapat memperoleh pemahaman yang lebih baik
tentang perilaku struktural jembatan kabel tahan ledakan dan kemampuannya dalam menghadapi pembebanan
ledakan. Hal ini dapat membantu dalam perancangan yang lebih efektif dan evaluasi keamanan jembatan kabel
tahan ledakan
PEMBEBANAN :
- Sistem pembebanan pada jurnal ini adalah pembebanan ledakan. Penelitian ini menggunakan analisis numerik
untuk memodelkan dan menganalisis respons struktural jembatan kabel tahan ledakan terhadap beban ledakan
yang berbeda.
- Pada analisis ini, peneliti menggunakan model elemen balok serat 1D untuk merepresentasikan elemen struktural
jembatan. Model ini mempertimbangkan variasi dalam penampang struktural dan stiabilitas torsional. Di
samping itu, diaphragm digunakan untuk meningkatkan kekakuan elemen struktural.
- Pembebanan ledakan dipertimbangkan pada dek jembatan. Beban ledakan ini diterapkan pada berbagai lokasi
dan intensitas, dan dianalisis dampaknya terhadap pergeseran, regangan, dan tegangan jembatan. Hasil analisis
menunjukkan bahwa pergeseran, regangan, dan tegangan jembatan dipengaruhi oleh lokasi dan intensitas beban
ledakan. Pada beberapa kasus, pergeseran dan regangan mencapai tingkat yang tidak dapat ditoleransi, terutama
pada beban ledakan dengan bobot setara lebih dari 750 kg. Selain itu, analisis juga menunjukkan bahwa
konfigurasi kabel-kabel pengikat pada jembatan mempengaruhi respons struktural terhadap beban ledakan
STRUKTUR :
- Struktur jembatan yang diteliti adalah jembatan kabel tahan ledakan. Jembatan ini merupakan jenis jembatan
kabel-tetap yang terdiri dari dek, pylon, dan kabel-kabel miring yang disebut sebagai stay. Jembatan ini memiliki
penampang balok kotak berbahan baja dengan ketinggian 1,15 m dan lebar 13,0 m. Di samping itu, jembatan ini
dilengkapi dengan penegak-penegak vertikal untuk meningkatkan kekakuan struktural.
- Struktur jembatan ini dimodelkan menggunakan elemen balok serat 1D dalam analisis numerik. Model ini
memperhitungkan variasi dalam penampang struktural dan stabilitas torsional. Selain itu, diaphragm digunakan
untuk meningkatkan kekakuan elemen struktural.
- Jembatan ini juga memiliki tiga konfigurasi kabel-kabel pengikat yang berbeda, yaitu konfigurasi Fan, Semi-Fan,
dan Harp. Konfigurasi ini mempengaruhi respons struktural jembatan terhadap beban ledakan.
PERENCANAAN STRUKTUR :
- Perencanaan struktur jembatan dilakukan dengan menggunakan analisis numerik dan pendekatan berbasis
kinerja. Perencanaan ini melibatkan pemodelan struktur jembatan menggunakan elemen balok serat 1D dan
melakukan analisis dinamik nonlinier untuk mengevaluasi respons struktural terhadap beban ledakan.
- Dalam perencanaan ini, variasi dalam geometri jembatan, karakteristik material, profil pylon, dan luas
16
C.D. Tetougueni, et al. E
ngin
eerin
gStru
ctu
res2
19(2020
)110739
penampang dek dipertimbangkan. Analisis pushover nonlinier dilakukan untuk menentukan kapasitas struktural
jembatan dalam arah vertikal. Dari analisis ini, kurva kapasitas ideal diperoleh dan digunakan untuk menentukan
batas kapasitas struktural jembatan
- Selain itu, perencanaan struktur jembatan juga melibatkan analisis fragilitas. Analisis ini bertujuan untuk
menentukan kemungkinan terjadinya kegagalan struktural jembatan selama kejadian ledakan. Metode fragilitas
digunakan untuk mengintegrasikan pengaruh ketidakpastian variabel dalam menentukan kemungkinan melebihi
batas kapasitas struktural.
MANFAAT :
- Meningkatkan pemahaman tentang respons struktural jembatan kabel tahan ledakan: Penelitian ini memberikan
pemahaman yang lebih baik tentang bagaimana jembatan kabel tahan ledakan merespons pembebanan ledakan.
Hal ini dapat membantu para perancang dan insinyur struktur dalam memahami perilaku struktural jembatan
tersebut dan mengoptimalkan desainnya.
- Meningkatkan keamanan jembatan kabel tahan ledakan: Dengan memahami respons struktural jembatan terhadap
pembebanan ledakan, penelitian ini dapat membantu dalam perancangan yang lebih efektif dan evaluasi
keamanan jembatan kabel tahan ledakan. Hal ini dapat mengurangi risiko kegagalan struktural dan meningkatkan
keamanan jembatan tersebut.
CONTOH JEMBATAN
17