See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/322827279

A Simple Approach for Simulating the Road Surface Roughness Involved in

Vehicle-Bridge Interaction Systems

Article in International Journal of Structural Stability and Dynamics · January 2018

DOI: 10.1142/S0219455418710098

CITATIONS READS

15 316

3 authors, including:

Yao Zhang Haisheng Zhao

Nanyang Technological University Dalian University of Technology
27 PUBLICATIONS 740 CITATIONS 49 PUBLICATIONS 508 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Haisheng Zhao on 12 July 2021.

The user has requested enhancement of the downloaded file.

 International Journal of Structural Stability and Dynamics
Vol. 18, No. 7 (2018) 1871009 (10 pages)
#.c World Scienti¯c Publishing Company
DOI: 10.1142/S0219455418710098
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

A Simple Approach for Simulating the Road Surface

Roughness Involved in Vehicle-Bridge
Interaction Systems

Zhang Yao*, Zhao Hai Sheng and Lie Seng Tjhen

Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

School of Civil & Environmental Engineering

Nanyang Technological University
639798, Singapore
*zhangyao@ntu.edu.sg

Received 24 July 2017

Accepted 22 December 2017
Published 31 January 2018

This paper presents an idea for modeling the road surface roughness in a vehicle–bridge
interaction (VBI) system, by simulating it equivalently as two external forces each acting on the
two subsystems of vehicle and bridge. Such an idea can be easily included in general-purpose
commercial ¯nite element (FE) software like ABAQUS and ANSYS. Compared with frequently
used coupled and uncoupled FE models, the present approach is more convenient, since it does not
require any self-developed FE codes. The other advantage is that it does not require very small
elements in the FE modeling, as is the case with conventional approaches for simulating the
irregularity in the pro¯le of road surface roughness, which may be computationally ine±cient.

Keywords: Finite element model; commercial software; road surface roughness; vehicle;
vehicle–bridge interaction (VBI).

1. Introduction
The dynamic responses of vehicle–bridge interaction (VBI) system have been
extensively studied to identify modal parameters, detect damages, and evaluate
the dynamic ampli¯cation factor. It is a di±cult problem because the vehicle and
bridge are coupled. In theoretical studies, the two subsystems of vehicle and bridge
are usually considered independently1,2 so that the governing equations can
be de-coupled. But it is found that this approximate solution is suitable only if the
vehicle mass is negligible compared with the bridge mass.3 Extensive ¯nite element
(FE) models can be found in the literature, and they can be generally classi¯ed into
three categories4: the ¯rst two categories require self-developed FE codes and the
third one can be implemented by using general-purpose commercial FE software.
* Corresponding author.

1871009-1
 Y. Zhang, H. S. Zhao & S. T. Lie

The models in the ¯rst category consider the vehicle and the bridge as a coupled
system.5–10 The sti®ness, mass, and damping matrices are time dependent, and they
change with respect to the location of the vehicle. Road roughness is considered as a
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

pair of internal forces. These models usually use a step-by-step integration scheme
such as fourth-order Runge–Kutta method and Newmark method. Some researchers
used modal superposition method11–14 to simplify the mass, sti®ness, and damping
matrices in which the model parameters must be determined ¯rst by using
general-purpose commercial FE software. The models in this category generally
require the self-developed FE codes, which is quite di®erent in practice, and therefore
limits the wider application.
The models in the second category consider the interactive system uncoupled
and treat the vehicle and the bridge separately.15–17 Neither the bridge nor the
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

vehicle is time dependent, which is di®erent from the coupled method. The compati-
bility conditions at the interface between vehicle and bridge must be satis¯ed by
an iterative procedure. There are two commonly used iterative procedures. For the
¯rst one, at a certain time step, the interaction force is calculated ¯rst by solving the
FE model of the vehicle where the road pro¯le at the contact point is considered as
boundary condition; then it is applied to the bridge and the displacement of the bridge
at the contact point can be obtained by solving the FE model of the bridge; subse-
quently, the summation of the bridge displacement and the road pro¯le at the contact
point performs as boundary condition to calculate the new interaction force by solving
the FE model of the vehicle, which is then applied to the bridge as external force at the
contact point and the new bridge displacement is obtained. The procedure is repeated
for several times until some convergence criteria are satis¯ed. Then the iterative pro-
cedure moves to the next time step. For the second type of iterative procedure, an initial
estimation of the interaction force for the entire time history is obtained by solving the
vehicle FE model, where the road roughness is the only external excitation; then it
performs as a time varying force load passing through the bridge. The bridge dis-
placement can be obtained by the FE model of the bridge, subsequently, the bridge
de°ection is added to the road pro¯le and an updated estimation of the interaction force
for the entire time history is obtained by solving the FE model of the vehicle, which
performs as an updated moving load and the corresponding bridge displacement can be
calculated. The procedure is repeated several times until some convergence criteria are
satis¯ed. Regardless of the di®erent iterative procedure, this uncoupled method usually
requires self-developed FE codes which limits the wider application in practice.
The last category18,19 is to generate the FE model by using general-purpose com-
mercial software like ABAQUS and ANSYS directly, where the irregularity is modeled
explicitly in the ¯nite element mesh. The number of elements must be very large
to obtain high spatial resolution, which requires too much computational resource.
In this study, an equivalent coupled FE model of VBI system based on ABAQUS
is proposed in which the road surface roughness is considered as two external forces
each acting on the two subsystems of vehicle and bridge. This model has the same
accuracy with other FE models while it is much easier to generate in general-purpose

1871009-2
 Simple Approach for simulating roughness in VBI systems

commercial FE software. Moreover, because this model is based on ABAQUS,

nonlinear e®ect and time varying vehicle velocity e®ect can be easily included as well.
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

2. FE Model Based on ABAQUS

Figure 1 shows the typical simpli¯ed model of a VBI system, where the vehicle is
simpli¯ed as a single Degree of Freedom (DoF) spring–mass system with sti®ness kv
and mass mv , and the bridge is considered as a simply supported Euler beam with
bending sti®ness EI, mass mb per unit length and length l. The vehicle is passing
through the bridge at a constant speed, v, and rðxÞ is the road surface pro¯le, xc is the
coordinate of the contact point. For simplicity, damping is not considered in this study.
The governing equations of the VBI system are given as20
@ 2 ub @ 4 ub
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

mb þ EI ¼ fðtÞ
ðx
vtÞ; ð1Þ

@t 2 @x 4
d 2q
mv 2v þ kv ðqv
ub jx¼vt
rðxÞjx¼vt Þ ¼ 0; ð2Þ
dt
where qv ðtÞ and ub ðx; tÞ are the vertical displacements of the vehicle and the bridge
measured from the static equilibrium position, respectively. fðtÞ is the interaction
force and
is the Dirac delta function
fðtÞ ¼ kv ½qv
ub jx¼vt
rðxÞjx¼vt 
mv g; ð3Þ
where g is the gravitational acceleration.
Equations (1)–(2) can be written in the matrix form as20

 !
½mb  f€
ub g
mv ðnþ1Þðnþ1Þ q€v ðnþ1Þ1

0 1
@ 2 fNg T

B ½kb  þ v 2m v fNg þ kv fNgfNg
kv fNg C
T
@x 2 fub g
þB
@
C
A qv ðnþ1Þ1

kv fNg T
kv ðnþ1Þðnþ1Þ
! !

mv gfNgvt
fNgvt
¼ þ kv rðxÞjx¼vt ;
0 ðnþ1Þ1
1 ðnþ1Þ1 ð4Þ

Fig. 1. Simpli¯ed model of VBI system.

1871009-3
 Y. Zhang, H. S. Zhao & S. T. Lie
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

(a)
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

(b)

Fig. 2. FE models of VBI system: (a) single DoF vehicle and (b) double DoFs vehicle.

where n is the number of Dofs of bridge elements, fub g is the displacement vector of
the bridge elements, [mb ] and [kb ] are general mass and sti®ness matrices of the
bridge, respectively; fNg is a vector containing the cubic Hermitian interpolation
functions and fNgvt is the value that fNg evaluated at the contact point.
It can be seen from Eq. (4) that the road surface roughness performs as two
external forces acting on the bridge and the vehicle, respectively. Therefore, the FE
model based on ABAQUS can be generated accordingly. Figures 2(a) and 2(b) show
the FE models in which the vehicles are modeled as single DoF and double DoFs
spring mass systems, respectively. For single DoF spring mass system, the two
external forces induced by road surface roughness are applied on the reference points
representing the vehicle body and contact point, respectively; and the external force
of mv g is applied to the reference point of vehicle body which contains mass of mv .
The two reference points are controlled to move at a constant speed, v. For the
double DoFs spring system, mv and mw are the mass of the vehicle body and the
wheel, kv and kw are the sti®ness of the suspension system and the wheel. The forces
due to the road roughness are applied on the reference points representing the
contact point and the wheel axis, respectively; the external forces of mv g and mw g are
applied to the reference points of vehicle body and wheel axis which contain mass of
mv and mass of mw . The three reference points are controlled to move at a constant
speed, v. It should be noted herein that this FE model can be easily extended to a 3D
model.
The bridge can be modeled by using general beam element and the interaction
between the contact point and the bridge is general node to surface contact. It is hard
contact without tangential friction in ABAQUS and no separation is allowed.

1871009-4
 Simple Approach for simulating roughness in VBI systems

3. Numerical Simulations
Numerical simulations are carried out to study the accuracy of the proposed FE
model by comparing with the uncoupled FE model requiring iterative procedure.17
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

Two cases are conducted, where the vehicles are modeled as single DoF and double
DoFs spring mass system, respectively. A simply supported concrete bridge is used
for the two cases. The bridge20 has length of 25 m, cross-sectional area of 2 m2, and
bending sti®ness of 0.12 m4. It has a total of 40 elements (B21 in ABAQUS). The
density, elastic modulus, and Poisson's ratio of the concrete are 2 400 kg/m3,
27.5 Gpa and 0.3, respectively. For the single DoF spring mass system, the following
properties are adopted: mv ¼ 5 000 kg, kv ¼ 1  10 6 N/m; for the double DoFs spring
mass system, the following properties are used: mv ¼ 4 500 kg, kv ¼ 5  10 5 N/m,
mw ¼ 500 kg, kw ¼ 1  10 6 N/m. The vehicle velocity is 10 m/s and the sampling
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

frequency is 1 000 Hz for the two cases.

The road pro¯le is generated by Power Spectral Density (PSD) method21 and the
one-sided power spectral densities for the reference spatial frequency (0.1 m
1 ) are
16  10
6 and 64  10
6 m3 for Class A and Class B, respectively. The lower and
upper spatial frequency limits are set as 0.01 m
1 and 10 m
1 . The road pro¯les of
Class A and Class B can be seen in Figs. 3 and 4. The Class A road pro¯le is used for
the ¯rst case, and the Class B road pro¯le is used for the second case. All calculations
are performed using a workstation with an Intel Xeon E5-1630 processor.
For the ¯rst case, the vertical displacements, velocities and accelerations of
the bridge midpoint and vehicle, obtained by both the proposed model and the
uncoupled model using iterative procedure,17 are shown in Figs. 5–7, respectively.
The interactive procedure used in the uncoupled FE model in Ref. 17 is the second
type shown in the introduction and its convergence criterion is that the maximum
di®erence between the bridge displacements of two successive iterations is less than
0.01% of the dynamic displacement at midspan. The results obtained by the two
models show very good agreement, which indicates the accuracy of the proposed

(a) (b)

Fig. 3. Class A road pro¯le: (a) Elevation and (b) One-sided PSD.

1871009-5
 Y. Zhang, H. S. Zhao & S. T. Lie
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

(a) (b)
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

Fig. 4. Class B road pro¯le: (a) Elevation and (b) One-sided PSD.

(a) (b)

Fig. 5. Displacement of (a) bridge midpoint and (b) vehicle for Case 1.

(a) (b)

Fig. 6. Velocity of (a) bridge midpoint and (b) vehicle for Case 1.

1871009-6
 Simple Approach for simulating roughness in VBI systems
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

(a) (b)
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

Fig. 7. Acceleration of (a) bridge midpoint and (b) vehicle for Case 1.

method is as good as the uncoupled method with iterative procedures. However, the
uncoupled method requires the iterative procedure to be repeated 6 times to con-
verge; this is because the frequencies of vehicle and bridge are 2.25 Hz and 2.08 Hz
which are close. It is noteworthy that one iterative procedure indicates that the
interaction force for entire time history calculated from the FE model of the vehicle is
applied to the FE model of the bridge as moving load and then the bridge
displacement and road surface roughness perform as boundary condition in the FE
model of the vehicle. More details about the uncoupled method can be found in
Ref. 17. The CPU time of the uncoupled method is 2 904 s while the CPU time of the
proposed method is only 467 s. It should be addressed herein that for this speci¯c
case, when the frequencies of vehicle and bridge are close, it is quite possible that the
process cannot converge.
For the second case, the vertical displacements, velocities, and accelerations of
bridge midpoint and vehicle, obtained by both the proposed FE model and the
uncoupled FE model, are shown in Figs. 8–10, respectively. The results also match

(a) (b)

Fig. 8. Displacement of (a) bridge midpoint and (b) vehicle for Case 2.

1871009-7
 Y. Zhang, H. S. Zhao & S. T. Lie
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

(a) (b)
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

Fig. 9. Velocity of (a) bridge midpoint and (b) vehicle for Case 2.

(a) (b)

Fig. 10. Acceleration of (a) bridge midpoint and (b) vehicle for Case 2.

quite well. For this case, the uncoupled method requires the iterative procedure to be
repeated for only 3 times to converge. The CPU time of the uncoupled model reduces
to 1 452 s while the CPU time of the proposed model reduces to 427 s. Moreover, it
should be noted herein that the uncoupled model usually needs operation in both
ABAQUS (or ANSYS) and MATLAB in which self-developed FE codes are required,
while the proposed model only requires general-purpose commercial FE software like
ANSYS and ABAQUS.

4. Conclusions
This paper proposes a method to model the road surface roughness in a VBI system,
by considering it equivalently as two external forces each applied on the two
subsystems of vehicle and bridge. This idea can be easily included in general-purpose
commercial FE software like ABAQUS and ANSYS. Previous studies have proved
that the coupled FE model is as accurate as the uncoupled Fe model; therefore,

1871009-8
 Simple Approach for simulating roughness in VBI systems

comparison between the proposed approach and uncoupled FE model indicates the
proposed model has the same accuracy. Moreover, the proposed method is more
convenient compared with commonly used coupled and uncoupled FE models
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

because it does not require self-developed FE codes. Compared with conventional FE

models generated by general-purpose commercial FE software, where the irregularity
is modeled explicitly in the FM mesh, this proposed method does not require ex-
tremely small elements and consequently saves computational resources. Moreover,
the proposed method can easily include nonlinear e®ect and time varying vehicle
velocity e®ect, and it can also be extended to 3D model conveniently.

Acknowledgment
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

The authors would like to thank the School of Civil and Environmental Engineering
at Nanyang Technological University, Singapore for kindly supporting this research
project.

References
1. L. Fryba, Vibrations of Solids and Structures Under Moving Loads (Nordho®
International Publication Co., Groningen, 1972).
2. J. W. Smith, Vibration of Structures: Applications in Civil Engineering Design (Chapman
and Hall, 1988).
3. Y. B. Yang, M. C. Cheng and K. C. Chang, Frequency variation in vehicle bridge
interaction systems, Int. J. Struct. Stab. Dyn. 13(2) (2013) 1350019.
4. Q. L. Zou, L. Deng, T. D. Guo and X. F. Yin, Comparative study of di®erent numerical
models for Vehicle-Bridge Interaction analysis, Int. J. Struct. Stab. Dyn. 16(9) (2016)
1550057.
5. Y. B. Yang and J. D. Yau, Vehicle bridge interaction element for dynamic analysis,
J. Struct. Eng. ASCE 123(11) (1997) 1512–1518.
6. S. S. Law and X. Q. Zhu, Dynamic behavior of damaged concrete bridge structures under
moving vehicular loads, Eng. Struct. 26 (2004) 1279–1293.
7. A. Gonzalez, P. Rattigan, E. J. Obrien and C. Caprani, Determination of bridge lifetime
dynamic ampli¯cation factor using ¯nite element analysis of critical loading scenarios,
Eng. Struct. 30 (2008) 2330–2337.
8. S. G. M. Neves, A. F. M. Azevedo and R. Calcada, A direct method for analyzing the
vertical vehicle structure interaction, Eng. Struct. 34 (2012) 414–420.
9. S. G. M. Neves, A. F. M. Azevedo and R. Calcada, A direct method for analyzing
the nonlinear vertical vehicle structure interaction, Eng. Struct. 69 (2014) 83–89.
10. Y. L. Li, S. Y. Zhu, C. S. Cai, C. Yang and S. Z. Qiang, Dynamic response of railway
vehicles running on long-span cable-stayed bridge under uniform seismic excitations,
Int. J. Struct. Stab. Dyn. 16(5) (2016) 1550005.
11. C. W. Kim, M. Kawatani and K. B. Kim, Three-dimensional dynamic analysis for bridge
vehicle interaction with roadway roughness, Comput. Struct. 83 (2005) 1627–1645.
12. L. Deng and C. S. Cai, Identi¯cation of parameters of vehicles moving on bridges,
Eng. Struct. 31 (2009) 2474–2485.
13. X. F. Yin, Y. Liu, S. H. Guo, W. Zhang and C. S. Cai, Three-dimensional vibrations of a
suspension bridge under stochastic tra±c °ows and road roughness, Int. J. Struct. Stab.
Dyn. 16(7) (2016) 1550038.

1871009-9
 Y. Zhang, H. S. Zhao & S. T. Lie

14. L. Deng, R. Cao, W. Wang and X. F. Yin, A multi-point tire model for studying
Bridge–Vehicle coupled vibration, Int. J. Struct. Stab. Dyn. 16(8) (2016) 1550047.
15. J. C. Wyss, D. Su and Y. Fujino, Prediction of vehicle-induced local responses and
by NANYANG TECHNOLOGICAL UNIVERSITY on 03/01/19. Re-use and distribution is strictly not permitted, except for Open Access articles.

application to a skewed girder bridge, Eng. Struct. 33 (2011) 1088–1097.

16. A. Gonzalez, D. Cantero and E. J. Obrien, Dynamic increment for shear force due to
heavy vehicles crossing a highway bridge, Comput. Struct. 89 (2011) 2261–2272.
17. Z. Sun, T. Nagayama, D. Su and Y. Fujino, A damage detection algorithm utilizing
dynamic displacement of bridge under moving vehicle, Shock Vib. 2016 (2016) 8454567.
18. H. Moghimi and H. R. Ronagh, Impact factors for a composite steel bridge using
non-linear dynamic simulation, Int. J. Impact Eng. 35 (2008) 1228–1243.
19. J. Oliva, J. M. Goicolea, P. Antolin and M. A. Astiz, Relevance of a complete road surface
description in vehicle-bridge interaction dynamics, Eng. Struct. 56 (2013) 466–476.
20. Y. B. Yang, C. W. Lin and J. D. Yau, Extracting bridge frequencies from the dynamic
response of a passing vehicle, J. Sound Vib. 272 (2004) 471–493.
Int. J. Str. Stab. Dyn. 2018.18. Downloaded from www.worldscientific.com

21. ISO-8608, Mechanical vibration — road surface pro¯les — reporting of measured data.

1871009-10

View publication stats