Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

Engineering Structures 229 (2021) 111645

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Load-carrying capacity evaluation of girder bridge using moving vehicle


Zhen Sun a, Dionysius M. Siringoringo b, *, Yozo Fujino b
a
State Key Laboratory of Safety and Health for In-Service Long Span Bridges, Jiangsu Transportation Institute, 2200 Chengxin Street, Jiangning kexueyuan, Nanjing,
China
b
Institute of Advanced Sciences, Yokohama National University, 79-1 Tokiwadai, Hodogaya-ku, Yokohama, Kanagawa 240-8501, Japan

A R T I C L E 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
Girder bridges
expensive and requires massive traffic closure that hinders frequent implementation on highway or bridges in the
Influence line rural area. In this paper, an efficient method for load-carrying capacity evaluation of girder bridges based on
displacement under moving vehicle is proposed. 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.

1. Introduction condition assessment of real bridges. Due to the variation in material


properties, boundary conditions and composite action, load test is often
Deterioration of bridge structures is a widespread problem in many performed to update the finite element (FE) model used in the
developed and developing countries. This is mainly due to ever- calculation-based method. For example, Ding et al. [2] combined model
increasing traffic volume, poor quality of material, poorly designed updating with field test data in a nonlinear finite element analysis to
structure, harsh environment condition, and inadequate maintenance. predict the behavior of a slab-girder bridge. Nonlinear analysis is per­
Bridge deteriorations limit the service life and pose challenges for formed on the updated model to evaluate the bridge load-carrying ca­
maintenance of infrastructure network. Generally, to prevent deterio­ pacity, which verified its effectiveness.
ration on various levels, periodic inspection is specified in the codes of A number of researchers have proposed various methods to evaluate
most countries. Periodic inspection usually includes visual inspection the bridge load-carrying capacity. Casas and Gomez [3] proposed a
and/or non-destructive techniques (NDTs). From inspections, conditions simple method based on the reliability theory to consider the random
of bridges are rated at different levels according to the deficiency status. nature of the traffic load to calculate the target proof load in tests for
An important result of inspection that concerns many bridge opera­ bridge assessment. Olaszek et al. [4] showed that diagnostic load test
tors or authorities is the load-carrying capacity. Bridge load-carrying can contribute to the bridge capacity assessment with examples of three
capacity evaluation can be carried out by calculation-based methods, types of bridges, which are reinforced concrete, prestressed concrete,
where inspection results on local damages are incorporated to achieve and steel bridge. An in-depth overview of the state-of-the-art and future
realistic model of the bridge. Brownjohn et al. [1] described a sensitivity- research trends with regard to load test of concrete bridges is provided in
analysis-based finite element model updating method and applied it to [5].

* 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 , γ DW and γ LL 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 = ϕc ϕs ϕRn (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. 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. Fig. 1 describes a
In the United States, the AASHTO bridge evaluation manual rec­
flowchart of the proposed method, which can be explained as follows.
ommends comparison between the field tests and analytical results to
First, a radar-based dynamic displacement measurement device is
update the bridge load rating factor [6]. Several European countries
selected, and careful preparation is conducted for measurement loca­
have adopted or in the process of adopting new codes for the assessment
tion, test truck weight and speed determination. Second, the truck pass-
of existing bridges such as the UK, Denmark, and Switzerland [3]. The
by test is performed to obtain the dynamic displacement. Third, the
codes incorporate numerous advanced techniques for load rating of
dynamic displacement data is processed to calculate the fundamental
existing bridges such as the use of data from weight-in-motion (WIM)
mode frequency of the bridge. Fourth, a low-pass filter-based technique
system. Planning, load application, measurement equipment and test
is employed to obtain the influence line for displacement response of the
procedures for bridge load testing are also described in the guidelines for
bridge. Fifth, a simulated static load test is conducted on the bridge, and
existing bridges in UK [7]. Similar load test is also required in China [8]
the 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
measurement in the truck pass-by test. The radar system can measure
C − γ DC DC − γDW DW
RF = (1) displacement of multiple targets at higher sampling frequency without
γ LL (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

Fig. 2. A simply supported beam under a moving load at a constant speed.

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
quency for the low-pass filter. The original signal was filtered with the
∂4 y(x, t) ∂2 y(x, t) ∂y(x, t) cut-off frequency in the frequency domain; and an inverse Fourier
EI +m + 2mωb = Pδ(x − vt) (3)
∂x4 ∂t2 ∂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
conditions with regard to displacement and rotation: yt = (Fi × di ) (7)
i=1
y(0, t) = 0; y(l, t) = 0; (4.a)
In which, n denotes the total number of loads in the simulated static
⃒ ⃒
∂2 y(x, t)⃒⃒ ∂2 y(x, t)⃒⃒ load test. Fi denotes the amplitude of the i-th vehicle load. di denotes the
⃒ = 0; =0 (4.b) value on the influence line corresponding to the position of the i-th
∂x 2
x=0 ∂x2 ⃒x=l
vehicle load. The bridge displacement response yc is calculated under
The closed-form solution for Eq. (3) can be obtained for initial designated truck weights and locations in the finite element model based
conditions on the design plan. Afterwards, the factor yyc is utilized to update the
t

∂y(x, t)⃒⃒ calculated load rating factor RF for the bridge capacity evaluation.
y(x, 0) = 0; and =0 (5)
∂t ⃒t=0 The dead and live load effects under the design load and the load-
carrying capacity at the concerned cross-section is calculated, and the
as follows: [18,19]. load rating factor can be obtained following Eq. (1). In this equation, the
( ) RF value larger than one indicates that the bridge design load-carrying
3∑∞ sin iπLx ( ) capacity is satisfied. Whereas the value smaller than one indicates that
2PL iπvt
y(x, t) = − {
mπ i=1 i (iπa) − (vL)
2 2 2 2 } sin
L the bridge design load-carrying capacity is less than required, and that
( ) (6) strategies for repair, strengthening or truck weight limitation counter­
4 ∑ ∞ sin iπLx measures are advisable.
2PL v
+ { } sin(ωi t)
mπ 3 a i=1 i3 (iπa)2 − (vL)2
4. Numerical simulations by finite element analysis
√̅̅̅ √̅̅̅
In which, a = EI and ωi = i Lπ2 EI
2 2
is the circular frequency at the i-th
m m In order to verify reliability of the proposed load-carrying capacity
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.

4.1. Description of finite element model of the bridge


Table 2
Displacement results at mid-span of five T-girders.
A finite element bridge model was developed for simulation using
finite element software ABAQUS [28] as shown by Fig. 3(a). Details of Load scenario T1 T2 T3 T4 T5

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 × 106 N/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(2πnk x − θk ) (10)
and the deflections at the mid-span of the five T girders are obtained as In which, Gd (nk ) = (2πan is the power spectral density function. a is
2
shown in Table 2. The values are 2.31 mm, 2.98 mm, 2.67 mm, 3.12 mm, k)

and 2.32 mm, respectively. It is noted that the displacement of the selected as 2 × 10− 6 indicating a good road surface condition. Δn is the
damaged girder T4 is larger than that at the symmetrical location of frequency interval of the road surface wave expressed as Δn = nmaxN− nmin ,
girder T2. in which nmax and nmin 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.

4.4. Results of numerical simulation

In the dynamic analysis, displacement time histories at the mid-span


of the five girders are obtained after convergence is achieved in
computation. A frequency analysis is performed with Fast Fourier
Transform method, and the fundamental natural frequency is obtained
as 8.63 Hz. After sensitivity analysis on cut-off frequency, a low-pass
filter of 1 Hz is employed to extract the quasi-static component in the
displacement time series.
Displacement results measured at mid-span of girder T2 under the
Fig. 5. Truck pass-by load scenario. scenario of 20ton truck running at 5 m/s are presented here as an
example. Both the original and filtered results are plotted in Fig. 6. The

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

5. Full-scale implementation of the proposed method

In order to verify the efficiency and efficacy of the proposed method


in practice, a full-scale implementation was conducted. Both static load
test and truck pass-by test were carried out on a simply supported pre­
stressed concrete (PC) T-girder bridge.

5.1. Description of tested bridge

Full-scale implementation of the proposed method was conducted on


Wenxi Bridge. The bridge is a PC T-girder type in the suburb area of
Fuzhou City, China (Fig. 9(a)). It was completed in 2007 and consists of
nine 30 m simply supported spans. The bridge has a total width of 12.5
m with two traffic lanes in the middle and two 1.75 m pedestrian
walkways. The 15.7 cm thick concrete deck is supported on five PC T-
girders. Details on the prestressed concrete (PC) T-girder bridge is shown
in Fig. 9 (c), and dimension of the deck cross section is described in Fig. 9
(d).
Fig. 7. Comparison of displacements obtained from static and dynamic load
scenarios for various truck weights.
5.2. Description of static load testing

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, γ DW = 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

C − (γDC )(DC) − (γDW )(DW) 8456 − 1.25 × 2323 − 1.5 × 1022


RF1 = = = 0.98
(γ L )(LL + IM) 1.75 × (1959 + 1959 × 0.2)

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)

Total weight Front axle Two rear axles T1 6.66


T2 7.18
1 30.69 6.14 24.55
T3 7.17
2 30.17 6.03 24.14
T4 6.80
3 29.62 5.92 23.70
T5 6.64
4 29.55 5.91 23.64

C − (γDC )(DC) − (γDW )(DW) 8009 − 1.25 × 2212 − 1.5 × 824


RF2 = = = 1.22
(γ L )(LL + IM) 1.75 × (1562 + 1562 × 0.2)

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) Parameter Girder T1 Girder T2
Displacement Difference Displacement Difference
(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

The proposed method was implemented and verified in a full-scale test


of a PC girder bridge using both static load test and truck pass-by test. The
results show that displacement estimated by the truck pass-by test is
reasonably accurate compared to the conventional static load test with
the largest difference within 10%. For better estimation, an amplification
factor is suggested to consider the nonlinearity effects under heavier loads
on damaged girder bridges. The study has demonstrated that the proposed
method can serve as an efficient substitute to traditional static load test,
especially for expressway or rural bridges. Based on the load-carrying
capacity rating factor, bridge owner and roadway operator can make
Fig. 13. Design load for Wenxi Bridge.
better decisions for maintenance and management. Traffic regulation can
be imposed by limiting the heavy trucks when the estimated load-carrying
Finally, the load rating factor is modified with an adjustment factor capacity has reduced significantly. Efficacy of the proposed method has
based on the load test result as: been shown by reducing the required number of persons involved in the
RFT = κRF (11) test and the testing time. In the future more tests are necessary to quantify
the nonlinear amplification factor for other bridge types such as slab, box,
where κ is the adjustment factor obtained from comparison between the and truss girder bridges.
test and analytical model and its value is taken as 1.15 according to the It should be mentioned that although the current study is conducted
AASHTO manual [6]. Therefore, the modified load rating factors for the for evaluation of load-carrying capacity following the AASHTO manual,
two girders T1 and T2 are 1.13 and 1.40, respectively. Generally, the the similar approach can be adopted to other codes and guideline pro­
load rating is represented by the rating factor RF as illustrated in Eq. (1). vided that they are based on capacity evaluation and that the load tests
It provides an estimation of the bridge carrying capacity with regard to of existing bridge are required. Such requirements demand an efficient
the design load. In this example, results of analyses suggest that the two and effective method for estimation of load-carrying capacity of existing
girders T1 and T2 can respectively sustain 1.13 and 1.40 times the bridge by a simple and less expensive load test procedure. Therefore, it is
design load shown in Fig. 13. anticipated that the similar study on the load rating factor will be
The results demonstrate that the proposed method can efficiently implemented on other codes for assessment of the load carrying capacity
provide an estimate of load carrying capacity using the truck pass-by of existing bridges.
test. Note that while the mid-span deflection of the bridge in this case
study is only around 3 mm, the proposed method can be implemented Declaration of Competing Interest
effectively since the radar device used here has the measurement ac­
curacy in an order of 0.01 mm [11]. This is one of the advantages of The authors declare that they have no known competing financial
using radar device compared to traditional displacement measurement interests or personal relationships that could have appeared to influence
device such as LVDT sensor or digital level have a precision of 0.1 mm. the work reported in this paper.
Therefore, under the above level of accuracy of radar device, the pro­
posed method can serve the needs in engineering practices. References
Based on the load carrying capacity obtained with the proposed
[1] Brownjohn JMW, Xia PQ, Hao H, Xia Y. Civil structure condition assessment by FE
method, bridge owner and roadway operator can make better decisions
model updating: methodology and case studies. Finite Elem Anal Des 2001;37(10):
for maintenance and management. For example, traffic regulation can 761–75.
be imposed the original legal gross vehicle weight of 40 ton could be [2] Ding L, Hao H, Xia Y, Deeks AJ. Evaluation of Bridge Load-carrying capacity Using
Updated Finite Element Model and Nonlinear Analysis. Adv Struct Eng 2012;15
regulated to 36 ton if the actual load rating factor is 0.9. The results
(10):1739–50.
show that accuracy of the truck pass-by test is reasonably good [3] Casas JR, Gómez JD. Load rating of highway bridges by proof-loading. KSCE J Civil
compared to the conventional static load test with less testing time and Eng 2013;17:556–67.
fewer labor forces. This could be a promising application for typical [4] Olaszeka P, Lagoda M, Casas JR. Diagnostic load testing and assessment of existing
bridges: examples of application. Struct Infrastruct Eng 2014;10(6):834–42.
medium-scale bridges on the rural area. [5] Lantsoght EOL, van der Veen C, de Boer A, Hordijk DA. State-of-the-art on load
testing of concrete bridges. Eng Struct 2017;150:231–41.
6. Conclusions [6] AASHTO. Manual for bridge evaluation. American Association of State Highway
and Transportation Officials. 2nd ed. Farmington Hills, MI; 2011.
[7] The Institution of Civil Engineers-National Steering Committee for the Load Testing
Evaluation of load-carrying capacity of existing bridges is essential of Bridges. Guidelines for the Supplementary Load Testing of Bridges. London, UK;
for rational and economical maintenance of severely damaged bridges. 1998.
[8] Ministry of Transport, China. Specifications for Inspection and Evaluation of Load-
Conventional static load testing is often considered too costly and labor- bearing Capacity of Highway Bridges, JTG /T-J21–2011; 2011.
intensive thus preventing its wide application. The conventional static [9] Nishio M, Abe M. Operation of a damaged prestressed-concrete girder bridge with
load test of existing highway bridges is very difficult for highway repair in Japan. P I Civil Eng Foren Eng 2019;172(1):19–26.
[10] Faber MH, Val DV, Stewart MG. Proof load testing for bridge assessment and
operator because it requires relatively long traffic closure time. In rural
upgrading. Eng Struct 2000;22:1677–89.
bridges, they are also rarely conducted due to lack of budget. [11] Su D, Nagayama T, Sun Z, Fujino Y. An interferometric radar for displacement
In this paper, an efficient method for evaluation of load-carrying measurement and its application in civil engineering structures. Proceedings of
Sensors and Smart Structures Technologies for Civil, Mechanical and Aerospace
capacity of existing bridge using moving vehicle is proposed. The
Systems 2012; San Diego, CA, United states.
method comprises three main procedures, namely, radar-based dynamic [12] Sun Z, Nagayama T, Nishio M, Fujino Y. Investigation on a curvature-based
displacement measurement under truck pass-by test, influence line damage detection method using displacement under moving vehicle. Struct
extraction, and estimation of the static load. Based on these procedures, Control Hlth 2018;25(1):e2044.
[13] Moses F. Weigh-in-motion system using instrumented bridge. J Transp Eng-ASCE
evaluation of load-carrying capacity is conduced following the AASHTO 1979;105(3):33–249.
manual. The method was verified in numerical simulations using three- [14] McNulty P, O’Brien EJ. Testing of bridge weigh-in-motion system in sub-arctic
dimensional finite element model of typical girder bridge under vehicle climate. J Test Eval 2003;31(6):1–10.
[15] Wang NB, He LX, Ren WX, Huang TL. Extraction of influence line through a fitting
loading. Effects of vehicle weight, vehicle speed, and road roughness on method from bridge dynamic response induced by a passing vehicle. Eng Struct.
the accuracy of displacement estimation was investigated in the nu­ 2017;151:648–64.
merical simulations.

11
Z. Sun et al. Engineering Structures 229 (2021) 111645

[16] Sun SW, Sun LM, Chen L. Damage detection based on structural response induced [23] Wang H, Nagayama T, Nakasuka J, Zhao B, Su D. Extraction of bridge fundamental
by traffic load: methodology and application. Int J Struct Stab Dy 2015;16(4): frequency from estimated vehicle excitation through a particle filter approach.
1640026. J Sound Vib 2018;428:44–58.
[17] Zheng X, Yang DH, Yi TH, Li HN. Development of bridge influence line [24] Siringoringo DM, Fujino Y. Estimating Bridge Fundamental Frequency from
identification methods based on direct measurement data: a comprehensive review Vibration Response of Instrumented Passing Vehicle: Analytical and Experimental
and comparison. Eng Struct 2019;198:109539. Study. Adv Struct Eng 2012;15(3):417–33.
[18] Frýba L. Vibration of solids and structures under moving loads, Vol. 1. Springer [25] O’ Brien EJ, Fitzgerald PC, Malekjafarian A, Sevillano E. Bridge damage detection
Science & Business Media; 2013. using vehicle axle-force information. Eng Struct 2017, 153(15): 71–80.
[19] Weaver JR, Timoshenko SP, Young DH. Vibration problem in engineering. New [26] Sun Z, Zou ZL. Towards an efficient method of predicting vehicle-induced response
York: John Wiley and Sons; 1990. of bridge. Eng Computation 2016;33(7):2067–89.
[20] Yang YB, Lin CW, Yau JD. Extracting bridge frequencies from the dynamic [27] Yang YB, Cheng MC, Chang KC. Frequency variation in vehicle-bridge interaction
response of a passing vehicle. J. Sound & Vibr. 2004;272(3–5):471–93. systems. Int. J. of Struct. Stability & Dyn. 2013;13(2):1350019.
[21] Lin CW, Yang YB. Use of a passing vehicle to scan the bridge frequencies - an [28] Abaqus 6.10 [Computer software]. Dassault Systemes Simulia Corp.,Providence,
experimental verification. Eng. Struct. 2005;27(13):1865–78. RI.
[22] Sun Z, Nagayama T, Su D, Fujino Y. A damage detection algorithm utilizing [29] MIDAS.IT Corporation 2006.
dynamic displacement of bridge under moving vehicle. Shock Vib 2016;8454567. [30] Budette EG, Goodpasture DW. Test to failure of a prestressed concrete bridge. PCI J
https://doi.org/10.1155/2016/8454567. 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.

12

You might also like