1 s2.0 S0020768308004344 Main PDF

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

International Journal of Solids and Structures 46 (2009) 1085–1094

Contents lists available at ScienceDirect

International Journal of Solids and Structures


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

Absolute versus relative formulations of the moving oscillator problem


Giuseppe Muscolino a, Alessandro Palmeri b,*, Alba Sofi c
a
Dipartimento di Ingegneria Civile, Università di Messina, Villaggio S. Agata, 98166 Messina, Italy
b
School of Engineering, Design & Technology, University of Bradford, Richmond Road, Bradford BD7 1DP, United Kingdom
c
Dipartimento di Arte, Scienza e Tecnica del Costruire, Università ‘‘Mediterranea” di Reggio Calabria, Via Melissari, Feo di Vito, 89124 Reggio Calabria, Italy

a r t i c l e i n f o a b s t r a c t

Article history: In the general framework of the bridge-vehicle dynamic interaction, the so-called ‘‘moving oscillator”
Received 30 June 2008 problem is revisited in order to provide a deeper insight into some analytical and physical aspects not
Received in revised form 20 September specifically analyzed in previous investigations. Without lack of generality, the case of a stream of moving
2008
oscillators crossing a simply supported beam with arbitrary time law is considered. The formulations in
Available online 26 October 2008
terms of both absolute and relative displacements of the moving oscillators are critically reviewed and
compared, and alternative sets of differential equations with time-dependent coefficients are derived.
Keywords:
The study enlightens, both theoretically and numerically, that impulsive contributions to the dynamic
Moving loads
Moving oscillators
response appear in the relative displacement formulation at the time instants in which each vehicle
Bridge-vehicle dynamic interaction enters or exits the bridge. It is demonstrated that such contributions, somehow ‘‘hidden” in the absolute
Dirac’s delta functions displacement formulation, may have a significant influence on the vibration of the moving oscillators,
Bending vibration of flexible beams and thus cannot be a priori neglected in the analysis. It is also shown that the analytical and computa-
Modal equations of motion tional difficulties associated with these additional impulses make preferable the use of the absolute dis-
placement formulation. Far from being restricted to the case of simply supported beams, these findings
are valid for any type of bridge structure which induces a discontinuity in the slope of the road profile
experienced by the vehicles.
Ó 2008 Elsevier Ltd. All rights reserved.

1. Introduction without resorting to more sophisticated models, with many de-


grees of freedom.
A vast literature has been devoted to the dynamic analysis of This paper revisits the moving oscillator problem, extensively
distributed-parameter systems, such as beams and cables, crossed studied in the literature, with the intention of highlighting some
by one or more subsystems (see, e.g. Tzou and Bergman, 1998; Frý- theoretical and practical features which, to the authors’ best
ba, 1999). Indeed, this topic is of great interest in many engineering knowledge, have not been addressed in previous investigations.
applications, such as the design of bridges, railway tracks, cable- First, in the vast majority of contributions the problem is formu-
ways, etc. It has long been recognized that the passage of a vehicle lated in terms of absolute displacements of the moving oscillator
may induce significant impact (or dynamic amplification) effects in (Chatterjee et al., 1994; Pesterev and Bergman, 1997a,b, 1998;
the supporting structure. For this reason, in the last decades sev- Muscolino et al., 2001; Chen et al., 2002; Pesterev et al., 2003;
eral studies have focused on the problem of bridge-vehicle dy- Chang et al., 2006; Stăncioiu et al., 2008), while comparatively
namic interaction (see, e.g. Yang et al., 2004). The simplest few authors make use of relative displacements (Yang et al.,
vehicle model adopted in the literature with the aim of investigat- 2000; Biondi and Muscolino, 2005; Muscolino and Palmeri, 2007;
ing this phenomenon is the so-called ‘‘moving oscillator” model. Muscolino et al., 2007a,b). However, advantages and disadvantages
This type of moving subsystem is characterized by finite coupling related to the use of the two formulations seem to have never been
stiffness and damping between a single lumped mass and the sup- discussed in the literature through a comprehensive comparison.
porting structure, and allows to take into account the main vibra- Furthermore, the attention is usually focused on the time interval
tional properties of the vehicle; thus, it turns out to be more in which a single vehicle crosses the bridge, while just few studies
realistic than the well-known ‘‘moving force” and ‘‘moving mass” concerning multiple moving oscillators are available in the litera-
models. Furthermore, multiple moving oscillators at fixed relative ture (see, e.g. Cheng et al., 1999; Pesterev et al., 2001; Biondi and
distance can be used to simulate the behaviour of complex vehicles Muscolino, 2005). In particular, as far as the authors know, the dy-
namic effects arising when a vehicle crosses the supports of the
bridge have not been specifically investigated by other researchers.
* Corresponding author. Tel.: +44 (0) 1274 23 6206; fax: +44 (0) 1274 23 4111. In the present study, with the intent of providing a deep insight
E-mail address: a.palmeri@bradford.ac.uk (A. Palmeri). into the aforementioned aspects, the bridge-vehicle dynamic inter-

0020-7683/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijsolstr.2008.10.019
1086 G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094

action is scrutinized by considering a simply supported Bernoulli– 2. Simply supported bridge carrying multiple moving oscillators
Euler beam vibrating under a stream of moving oscillators. Both
the formulations in terms of absolute and relative displacements Without lack of generality, let us consider a single-span simply
are reviewed and compared. In particular, it is analytically demon- supported bridge, initially at rest, crossed by nv vehicles moving
strated and numerically confirmed that impulsive terms arise in from left to right (Fig. 1). The bridge structure is treated as a homo-
the relative displacement formulation at the time instants in which geneous Bernoulli–Euler beam of length lb , mass density qb , mod-
each vehicle enters or exits the bridge (Muscolino et al., 2007b); ulus of elasticity Eb , and cross section with area Ab and second
that is, Dirac’s delta functions ideally centred at bridge supports moment J b . The approaches are regarded as rigid soil (but a flexible
appear in the governing equations. It is worth emphasizing that model could be also used). The ith vehicle is modeled as a moving
such terms, which stem from the so-called convective acceleration, oscillator with mass mv;i , elastic stiffness kv;i and viscous damping
do not arise when absolute displacements of the oscillators are cv;i . The moving subsystems are supposed to be in permanent con-
considered, being in some sense ‘‘hidden” as the Dirac’s delta func- tact with the road surface along bridge and approaches.
tions enlightened by Makris (1997) in a completely different con- Assuming that the only dynamic loads are those due to the nv
text. As a matter of fact, numerical results prove that in the moving vehicles, the bridge vibration is governed by the partial dif-
relative displacement formulation, especially when many oscilla- ferential equation:
tors are dealt with, these impulsive terms may have a significant
influence on the vibration of the moving oscillators, and therefore o2 ub ðz; tÞ o4 ub ðz; tÞ
qb Ab 2
þ Eb J b þ Db ðz; tÞ
cannot be a priori neglected in the dynamic analysis. Classical ot oz4
mode superposition technique is applied to get an approximate ðsÞ
Xnv

solution of the problem, formulated either in terms of absolute


¼ fb ðzÞ þ vb ðzv;i ðtÞÞfv;i ðtÞdðz  zv;i ðtÞÞ; ð1Þ
i¼1
or relative displacements. In both cases, a set of coupled ordinary
differential equations with time-dependent coefficients is ob- where t and z denote the time and the spatial coordinate measured
tained. More precisely, when the absolute displacements of the along the axis of the beam, respectively; ub ðz; tÞ is the field of the
ðsÞ
moving oscillators are used, the mass matrix in the modal space beam’s transverse displacements, positive if downward; fb ðzÞ is
is a constant diagonal matrix, while stiffness and damping matrices the field of the transverse static forces distributed along the beam
are time-dependent ones; furthermore, the stiffness matrix turns (Fig. 2 top), which in practice may be the effects of gravitational
out to be non symmetric. In the relative displacement formulation, dead and live loads, and of prestressing systems as well; Db ðz; tÞ is
on the contrary, all the matrices are time-dependent ones; more- the internal damping force per unit length of the beam; fv;i ðtÞ is
over, only the mass matrix is symmetric, while impulsive terms the point force transmitted by the ith traveling oscillator (Fig. 2 cen-
arise in the stiffness matrix at the entrance and exit times of each tre), which is located at the instantaneous position zv;i ðtÞ; vb ðzÞ de-
moving oscillator. Interestingly, simple transformations of coordi- notes the so-called window function for the supporting beam (Fig. 2
nates provided in Appendix allow to pass from the matrices in bottom), given by
the space of absolute displacements to the corresponding ones in vb ðzÞ ¼ UðzÞ  Uðz  lb Þ; ð2Þ
the space of relative displacements, and vice versa.
For validation purposes, a simply supported beam carrying a UðzÞ being the unit-step function, so defined: UðzÞ ¼ 0 when z < 0;
pair of moving oscillators is analyzed by applying the two formu- UðzÞ ¼ 1=2 at z ¼ 0; UðzÞ ¼ 1 when z > 0; and dðzÞ ¼ oUðzÞ=oz is the
lations derived in the paper. Numerical results in terms of beam’s Dirac’s delta function, symmetric with respect to z ¼ 0. It is recalled
transverse displacements and beam-oscillator interaction forces that the attendant boundary conditions for the simply supported
are presented and discussed, focusing the attention on the effects beam modeling the bridge read: ub ðz; tÞjz¼0 ¼ ub ðz; tÞjz¼lb ¼ 0 and
of the impulses in the relative displacement formulation. It has o2 ub ðz; tÞ=oz2 jz¼0 ¼ o2 ub ðz; tÞ=oz2 jz¼lb ¼ 0.
to be emphasized that these results are not restricted to the case Clearly, when all the oscillators are outside the bridge, the last
of simply supported beams only, being valid for any type of bridge term in Eq. (1) vanishes. Conversely, when the oscillators move
structure where a discontinuity appears in the slope of the road along the bridge, additional excitations arise as a consequence of
profile experienced by the vehicles. In all these situations the ana- bridge-vehicle dynamic interaction, which is realized by the nv
lytical and computational burden associated with the additional concentrated forces fv;i ðtÞ appearing in the right-hand side of Eq.
impulses arising in the relative displacement formulation make (1). Such forces depend on the response of both the supporting
preferable the one in terms of absolute displacements of the mov- beam and the moving oscillators, thus making the governing equa-
ing subsystems. tions coupled. It is worth emphasizing that the interaction force

Fig. 1. Simply supported bridge crossed by a stream of moving oscillators.


G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094 1087

Fig. 2. Schematic view of the forces acting on the simply supported bridge crossed by the ith moving oscillator.

fv;i ðtÞ takes different forms depending on whether the motion of where the over-dot denotes total derivative with respect to time t
the oscillators is described in terms of absolute or relative displace- and
ments, as will be outlined next in detail.  
Xb ¼ Diag xb;1 ; xb;2 ; . . . ; xb;nb ;
An approximate solution of the problem under consideration
can be derived by applying the classical modal analysis. Accord- Nb ¼ 2fb Xb ;
Z ð6a-cÞ
lb
ingly, the field of transverse displacements ub ðz; tÞ of the simply ðsÞ ðsÞ
pb ¼ /b ðzÞfb ðzÞdz:
supported beam can be approximated as 0
nb
X
ub ðz; tÞ ¼ /b;j ðzÞqb;j ðtÞ ¼ /Tb ðzÞqb ðtÞ; ð3Þ In the previous equations, Diag½ means diagonal matrix,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
j¼1 xb;j ¼ ðjp=lb Þ2 Eb Jb =ðqb Ab Þ denotes the jth undamped circular fre-
where quency of the beam, and fb is the modal damping ratio, herein as-
 T sumed constant in all vibration modes.
/b ðzÞ ¼ /b;1 ðzÞ; /b;2 ðzÞ; . . . ; /b;nb ðzÞ ; In the next sections, the expressions for the interaction forces
 T ð4a; bÞ
qb ðtÞ ¼ qb;1 ðtÞ; qb;2 ðtÞ; . . . ; qb;nb ðtÞ ; either in terms of absolute or relative displacements of the moving
oscillators will be deduced. Then, in both cases the equations gov-
are the vectors collecting thepfirst nb modal shapes of the simply
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi erning bridge-vehicle dynamics in the modal space will be derived.
supported beam, /b;j ðzÞ ¼ 2=ðqb Ab lb Þ sinðjpz=lb Þ, orthonormal
with respect to the mass per unit length qb Ab , and the associated
nb modal coordinates, qb;j ðtÞ, respectively. 3. Equations of motion in terms of absolute displacements
Substituting Eq. (3) into Eq. (1), pre-multiplying both sides by
/b ðzÞ, and integrating with respect to z between 0 and lb , the fol- The equation of motion of the ith traveling oscillator, in terms of
lowing set of nb ordinary differential equations with time-indepen- absolute displacements, can be written as
dent coefficients is obtained: mv;i u
ðaÞ ðaÞ ðaÞ
€ v;i ðtÞ ¼ cv;i ½u_ v;i ðtÞ  u_ w;i ðtÞ  kv;i ½uv;i ðtÞ  uw;i ðtÞ; ð7Þ
X
nv
ðaÞ
€ b ðtÞ þ Nb q_ b ðtÞ þ X2b qb ðtÞ ¼ pðsÞ
q b þ vb ðzv;i ðtÞÞfv;i ðtÞ/b ðzv;i ðtÞÞ; where uv;i ðtÞ and uw;i ðtÞ are the absolute displacements of the
i¼1 lumped mass mv;i and of the ideal point wheel, respectively (see
ð5Þ Fig. 3). The interaction force, fv;i ðtÞ, which depends on both the reac-

ðsÞ
Fig. 3. Sketch of the generic moving oscillator (a); static displacement uv;i of the mass mv;i (b); relative displacement uv;i ðtÞ of the mass mv;i , including the static contribution
ðaÞ ðaÞ
(c); displacement uw;i ðtÞ of the ideal point wheel, and absolute displacements, u  v;i ðtÞ and uv;i ðtÞ, of the mass mv;i , with or without the static contribution (d).
1088 G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094

tion of the spring-dashpot system and the weight of the ith oscilla- with
tor, can be expressed as (see Fig. 3):
Xv ðtÞ ¼ Diag½ vb ðzv;1 ðtÞÞ; vb ðzv;2 ðtÞÞ; . . . ; vb ðzv;nv ðtÞÞ ;
ðaÞ ðaÞ
fv;i ðtÞ ¼ mv;i g þ kv;i ½uv;i ðtÞ  uw;i ðtÞ þ cv;i ½u_ v;i ðtÞ  u_ w;i ðtÞ  
Z_ v ðtÞ ¼ Diag z_ v;1 ðtÞ; z_ v;2 ðtÞ; . . . ; z_ v;n ðtÞ ;
v
ðaÞ
€ v;i ðtÞ;
¼ mv;i ½g  u ð8Þ Ubv ðtÞ ¼ ½ /b ðzv;1 ðtÞÞ; /b ðzv;2 ðtÞÞ; . . . ; /b ðzv;nv ðtÞÞ ;
 
U0bv ðtÞ ¼ /0b ðzv;1 ðtÞÞ; /0b ðzv;2 ðtÞÞ; . . . ; /0b ðzv;nv ðtÞÞ :
where g is the acceleration of gravity, and Eq. (7) has been taken
into account. ð17a-dÞ
Assuming that there is no loss of contact between vehicles and Upon substitution of Eq. (11a,b) into Eq. (8), moreover, the
road surface, the compatibility between bridge and vehicle vibra- expression of the interaction force fv;i ðtÞ, in terms of beam’s modal
tions is ensured if the relationship: coordinates is obtained:
uw;i ðtÞ ¼ ½vb ðzÞub ðz; tÞjz¼zv;i ðtÞ ¼ vb ðzv;i ðtÞÞub ðzv;i ðtÞ; tÞ; ð9Þ h i
ðaÞ
fv;i ðtÞ ¼ mv;i g þ kv;i uv;i ðtÞ  aTvb;i ðtÞqb ðtÞ
holds at the instantaneous position zv;i ðtÞ of each oscillator. Taking h i
ðaÞ T
into account Eq. (9), the total time derivative of the absolute dis- þ cv;i u_ v;i ðtÞ  aTvb;i ðtÞq_ b ðtÞ  bvb;i ðtÞqb ðtÞ : ð18Þ
placement of the wheel uw;i ðtÞ, appearing in Eq. (8), is given by
  Finally, substituting Eq. (18) into Eq. (5), the equation of motion of
d the bridge in the modal space becomes
u_ w;i ðtÞ ¼ ðvb ðzÞub ðz; tÞÞ
dt z¼zv;i ðtÞ
  € b ðtÞ þ ½Nb þ DCb ðtÞq_ b ðtÞ  CTvb ðtÞu_ ðaÞ
q v ðtÞ

o½vb ðzÞub ðz; tÞ o vb ðzÞub ðz; tÞ
þ ½X2b þ DKb ðtÞqb ðtÞ  KTvb ðtÞuðaÞ
¼ þ z_ v;i ðtÞ v ðtÞ ¼ pb ðtÞ; ð19Þ
ot z¼zv;i ðtÞ oz z¼zv;i ðtÞ
" # where
oub ðz; tÞ oub ðz; tÞ
¼ vb ðzv;i ðtÞÞ _
þ zv;i ðtÞ ; ð10Þ
ot z¼zv;i ðtÞ oz z¼zv;i ðtÞ DCb ðtÞ ¼ Ubv ðtÞCv Xv ðtÞUTbv ðtÞ ¼ Ubv ðtÞCvb ðtÞ;
h i
where the geometric boundary conditions for the simply supported DKb ðtÞ ¼ Ubv ðtÞ Kv Xv ðtÞUTbv ðtÞ þ Cv Xv ðtÞZ_ v ðtÞU0T
bv ðtÞ
ð20a-dÞ
beam have been taken into account. Upon substitution of Eq. (3) ¼ Ubv ðtÞ½Kvb ðtÞ þ Lvb ðtÞ;
into Eqs. (9) and (10), we get the kinematical relationships: ðsÞ ðvÞ ðvÞ
pb ðtÞ ¼ pb þ pb ðtÞ; pb ðtÞ ¼ gUbv ðtÞMv Xv ðtÞsv :
uw;i ðtÞ ¼ vb ðzv;i ðtÞÞ/Tb ðzv;i ðtÞÞqb ðtÞ
¼ aTvb;i ðtÞqb ðtÞ; Interestingly, DCb ðtÞ and DKb ðtÞ in Eq. (19) may be viewed, respec-
  tively, as modifications of the damping and stiffness matrices in the
u_ w;i ðtÞ ¼ vb ðzv;i ðtÞÞ /Tb ðzv;i ðtÞÞq_ b ðtÞ þ z_ v;i ðtÞ/0T
b ðzv;i ðtÞÞqb ðtÞ modal space due to the passage of the moving oscillators, while
T
¼ aTvb;i ðtÞq_ b ðtÞ þ bvb;i ðtÞqb ðtÞ; sv ¼ f 1; 1; . . . ; 1 gT in the last of Eq. (20) is the nv -dimensional
ð11a; bÞ incidence vector of the moving oscillators. Eqs. (15) and (19) can be
rewritten in compact form as
where the prime means total derivative with respect to z, that is
Ma u _
€ ðtÞ þ Ca ðtÞuðtÞ þ Ka ðtÞuðtÞ ¼ pðtÞ; ð21Þ
d/b ðzÞ  0  T
¼ /b;1 ðzÞ; /b;2 ðzÞ; ...; /b;nb ðzÞ z¼z ðtÞ ;
0 0
/0b ðzv;i ðtÞÞ ¼
dz z¼zv;i ðtÞ v;i where the subscript ‘‘a” stands for ‘‘absolute displacements” and
where:
ð12Þ
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
with /0b;j ðzÞ
3
¼ jp 2=ðqb Ab lb Þ cosðjpz=lb Þ, and where the time- Mv 0nv 1
Ma ¼ ;
dependent vectors avb;i ðtÞ and bvb;i ðtÞ are so defined: 0nv 1 Inb
" #
avb;i ðtÞ ¼ vb ðzv;i ðtÞÞ/b ðzv;i ðtÞÞ; Cv Cvb ðtÞ
ð13a; bÞ Ca ðtÞ ¼ ;
bvb;i ðtÞ ¼ vb ðzv;i ðtÞÞz_ v;i ðtÞ/0b ðzv;i ðtÞÞ: CTvb ðtÞ Nb þ DCb ðtÞ
  ð22a-eÞ
Kv Kvb ðtÞ  Lvb ðtÞ
Substituting Eq. (11a,b) into Eq. (7), the equation of motion of Ka ðtÞ ¼ ;
the ith moving oscillator takes the form: KTvb ðtÞ X2b þ DKb ðtÞ
( )

ðaÞ
ðaÞ ðaÞ ðaÞ uv ðtÞ 0nv 1
€ v;i ðtÞ þ cv;i u_ v;i ðtÞ þ kv;i uv;i ðtÞ
mv;i u uðtÞ ¼ ; pðtÞ ¼ ;
h i qb ðtÞ pb ðtÞ
T
¼ cv;i bvb;i ðtÞ þ kv;i aTvb;i ðtÞ qb ðtÞ þ cv;i aTvb;i ðtÞq_ b ðtÞ: ð14Þ
the symbols In and 0nm being the identity matrix of size n and the
Then, the equations governing the motion of the nv oscillators zero matrix of dimensions ðn  mÞ, respectively. It is worth noting
can be posed in a compact matrix form as that, while the mass matrix Ma is diagonal and constant, the matri-
€ ðaÞ
Mv u _ ðaÞ ðaÞ
v ðtÞ þ Cv uv ðtÞ þ Kv uv ðtÞ ces Ca ðtÞ and Ka ðtÞ are sparse and time-dependent ones. Further-
¼ ½Kvb ðtÞ þ Lvb ðtÞqb ðtÞ þ Cvb ðtÞq_ b ðtÞ; ð15Þ more, the stiffness matrix Ka ðtÞ turns out to be non symmetric
because of the submatrix Lvb ðtÞ, whose elements are proportional
where the array ofn the absolute displacements ofothe T
moving oscil- to the horizontal speed of the moving oscillators.
ðaÞ
lators is uv ðtÞ ¼ uðaÞ
v;1 ðtÞ;
ðaÞ ðaÞ
uv;2 ðtÞ; . . . ; uv;nv ðtÞ , and where From a numerical point of view, it is advantageous to make all
Mv ¼ Diag½ mv;1 ; mv;2 ; . . . ; mv;nv ; the entries of the matrices appearing in Eq. (21) have the same
ðaÞ
physical dimensions. To this aim, the position uv ðtÞ ¼ M1=2
Cv ¼ Diag½ cv;1 ; cv;2 ; . . . ; cv;nv ; Cvb ðtÞ ¼ Cv Xv ðtÞUTbv ðtÞ; ðaÞ
v
qv ðtÞ is introduced, so that Eq. (21) can be rewritten as
Kv ¼ Diag½ kv;1 ; kv;2 ; . . . ; kv;nv ; Kvb ðtÞ ¼ Kv Xv ðtÞUTbv ðtÞ;
€ ðtÞ þ cðtÞqðtÞ
q _ þ kðtÞqðtÞ ¼ pðtÞ; ð23Þ
Lvb ðtÞ ¼ Cv Xv ðtÞZ_ v ðtÞU0T ðtÞ; bv

ð16a-fÞ where
G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094 1089

" # 2
Nv Mv1=2 Cvb ðtÞ d ðÞ o2 ðÞ o2 ðÞ oðÞ o2 ðÞ
cðtÞ ¼ ; Av;i ½ ¼ 2
¼ 2
þ 2z_ v;i ðtÞ þ €zv;i ðtÞ þ z_ 2v;i ðtÞ :
CTvb ðtÞMv1=2 Nb þ DCb ðtÞ dt ot ozot oz oz2
" # ð31Þ
X2v M1=2
v ½Kvb ðtÞ  Lvb ðtÞ
kðtÞ ¼ ; ð24a-cÞ As a result, the transverse acceleration experienced by the ith mov-
KT ðtÞM1=2 X2b þ DKb ðtÞ
( vb ) v ing oscillator when in contact with the supporting beam depends
ðaÞ
qv ðtÞ also on its longitudinal position zv;i ðtÞ, velocity z_ v;i ðtÞ and accelera-
qðtÞ ¼ ;
qb ðtÞ tion €zv;i ðtÞ.
By taking into account the geometric boundary conditions for
in which the simply supported bridge, and recalling also the properties of
h i the derivatives of the Dirac’s delta function (Bracewell, 1999), the
k kv;2 kv;nv
X2v ¼ Diag mv;1
v;1
; mv;2
; ...; mv;nv first- and second-order derivatives of vb ðzÞub ðz; tÞ with respect to
  the abscissa z, required by the operator Av;i ½ (Eq. (31)), are given by
¼ Diag x2v;1 ; x2v;2 ; . . . ; x2v;nv ;  
hc i ð25a; bÞ o vb ðzÞub ðz; tÞ oub ðz; tÞ
c c v
; . . . ; mv;n ¼ vb ðzÞ ;
Nv ¼ Diag mv;1 ; mv;2 oz oz

v;1 v;2 v;nv
  
¼ 2Diag fv;1 xv;1 fv;2 xv;2 ; . . . ; fv;nv xv;nv ; o2 vb ðzÞub ðz; tÞ o2 ub ðz; tÞ oub ðz; tÞ
2
¼ vb ðzÞ þ ½dðzÞ  dðz  lb Þ:
oz oz2 oz
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
xv;i ¼ kv;i =mv;i and fv;i ¼ cv;i =ð2 mv;i kv;i Þ being undamped natural ð32a; bÞ
circular frequency and viscous damping ratio of the ith moving
Then, the first term in the right-hand side of Eq. (30) takes the form:
oscillator, respectively. ( "
Eq. (23) can be solved with the help of any suitable step-by-step o2 ub ðz;tÞ o2 ub ðz;tÞ
algorithm, e.g. the Newmark’s b-method, to get the response of Av;i ½vb ðzÞub ðz;tÞjz¼zv;i ðtÞ ¼ vb ðzÞ 2
þ 2z_ v;i ðtÞ
ot ozot
both the bridge and the vehicles. The initial conditions at the time #
2
instant t0 in which the first moving oscillator approaches the beam oub ðz;tÞ 2 o ub ðz;tÞ
þ€zv;i ðtÞ þ z_ v;i ðtÞ
can be assumed to be oz oz2
( ) )

oub ðz;tÞ
0nv 1 0nv 1
qðt0 Þ ¼ ; _ 0Þ ¼
qðt ; ð26a; bÞ þz_ 2v;i ðtÞ ½dðzÞ  dðz  lb Þ :ð33Þ
ðsÞ oz
qb 0nb 1 z¼zv;i ðtÞ

ðsÞ ðsÞ
where qb ¼ X2b pb is the array of the modal coordinates of the
What Eqs. (32a,b) reveal is that when zv;i ðtÞ ¼ 0 or zv;i ðtÞ ¼ lb , i.e.
ðsÞ
beam associated with the static distributed load fb ðzÞ. when the ith moving oscillator enters or exits the beam, the slope
of the road profile felt by its ideal point wheel (Eq. (32a)) undergoes
a sudden jump, whereas the curvature (Eq. (32b)) exhibits an im-
4. Equations of motion in terms of relative displacements pulse centered at either the left or right support of the beam (Fig. 2).
Introducing Eq. (30) into Eq. (27), the equation of motion of the
The equation of motion of the ith traveling oscillator in terms of ith oscillator reads
relative displacements can be written as (analogous to Eq. (7)):
ðaÞ
€ v;i ðtÞ þ cv;i u_ v;i ðtÞ þ kv;i uv;i ðtÞ
mv;i u
€ ðtÞ þ cv;i u_ v;i ðtÞ þ kv;i uv;i ðtÞ ¼ mv;i g;
mv;i u ð27Þ
v;i
¼ mv;i fg  Av;i ½vb ðzÞub ðz; tÞjz¼zv;i ðtÞ g: ð34Þ
ðaÞ
where u  v;i ðtÞ and uv;i ðtÞ are, respectively, the absolute and relative
displacements of the mass mv;i , both including the static contribu- In a similar way, substitution of Eq. (30) into Eq. (29) yields the fol-
ðsÞ
tion uv;i ¼ mv;i g=kv;i (see Fig. 3). lowing expression of the ith dynamic interaction force:
If there is no loss of contact between vehicle and road surface, fv;i ðtÞ ¼ mv;i fg  Av;i ½vb ðzÞub ðz; tÞjz¼zv;i ðtÞ  u
€ v;i ðtÞg: ð35Þ
the following relationship holds at the instantaneous position
zv;i ðtÞ of each oscillator (see Fig. 3) It can be seen that when the ith oscillator moves on the left (or
right) rigid approach, i.e. zv;i ðtÞ < 0 (or zv;i ðtÞ > lb Þ, Eqs. (1) and (34)
 ðaÞ
u
ðaÞ ðsÞ
v;i ðtÞ ¼ uv;i ðtÞ þ uv;i ¼ uw;i ðtÞ þ uv;i ðtÞ are not coupled since vb ðzÞub ðz; tÞ ¼ 0 and consequently the con-
  vective acceleration is zero. Conversely, when the ith vehicle
¼ vb ðzÞub ðz; tÞ z¼z ðtÞ þ uv;i ðtÞ; ð28Þ
v;i crosses the left (or right) support of the bridge, i.e. zv;i ðtÞ ¼ 0 (or
zv;i ðtÞ ¼ lb Þ and z_ v;i ðtÞ > 0, it is subject to an additional excitation
where use has been made of Eq. (9).
due to a sudden change in the slope of the road profile. More pre-
The interaction force fv;i ðtÞ, concentrated at the abscissa zv;i ðtÞ,
cisely, when zv;i ðtÞ ¼ 0 (or zv;i ðtÞ ¼ lb Þ, besides the convective accel-
coincides with the reaction of the spring-dashpot system. Thus, it
eration terms associated with the slope o½vb ðzÞub ðz; tÞ=oz (Eq.
can be expressed as (analogous to Eq. (8))
h i (32a)), an impulsive excitation (see Eq. (33)) relating to the curva-
€ ðaÞ ðtÞ :
fv;i ðtÞ ¼ kv;i uv;i ðtÞ þ cv;i u_ v;i ðtÞ ¼ mv;i g  u ð29Þ ture o2 ½vb ðzÞub ðz; tÞ=oz2 (Eq. (32b)) arises in both Eqs. (1) and (34).
v;i
To the authors’ best knowledge, such term is not taken into ac-
ðaÞ
 v;i ðtÞ (see Eq. (28)) depends on time t both
Taking into account that u count by classical formulations, which commonly focus just on
directly and through the instantaneous position zv;i ðtÞ, it can be the time interval in which the vehicle travels along the bridge, in
readily verified that the absolute acceleration u  ðaÞ
€ € ðaÞ so neglecting the effects due to the passage from the approach,
v;i ðtÞ ¼ uv;i ðtÞ of
modeled as rigid soil, to the beam (and vice versa).
the mass mv;i , appearing in Eq. (29), is given by
Substituting Eq. (3) into Eq. (34), the equation of motion of the
  ith moving oscillator takes the form (analogous to Eq. (14))
€ ðaÞ ðtÞ ¼ Av;i
u v;i vb ðzÞub ðz; tÞ z¼zv;i ðtÞ þ u€v;i ðtÞ; ð30Þ
~ Tvb;i ðtÞq
€ v;i ðtÞ þ m
mv;i u € b ðtÞ þ cv;i u_ v;i ðtÞ þ ~cTvb;i ðtÞq_ b ðtÞ þ kv;i uv;i ðtÞ
where Av;i ½ denotes the so-called convective acceleration operator ~ T ðtÞq ðtÞ ¼ m g;
þk vb;i b v;i ð36Þ
for the ith moving oscillator:
1090 G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094

where Interestingly, DM ~ b ðtÞ and DK


~ b ðtÞ, DC ~ b ðtÞ may be viewed, respec-
tively, as modifications of the mass, damping and stiffness matri-
m~ vb;i ðtÞ ¼ mv;i vb ðzv;i ðtÞÞ/b ðzv;i ðtÞÞ ¼ mv;i avb;i ðtÞ;
ces in the modal space due to the passage of the moving
c~vb;i ðtÞ ¼ 2mv;i z_ v;i ðtÞvb ðzv;i ðtÞÞ/0b ðzv;i ðtÞÞ ¼ 2mv;i bvb;i ðtÞ; oscillators.
h i
k~ ðtÞ ¼ m v ðz ðtÞÞ z_ 2 ðtÞ/00 ðz ðtÞÞ þ €z ðtÞ/0 ðz ðtÞÞ Analogously to Eq. 21, Eqs. 36 and 43 can be rewritten in com-
vb;i v;i b v;i v;i b v;i v;i b v;i
  pact form as
þ mv;i z_ 2v;i ðtÞ /0b ð0Þdðzv;i ðtÞÞ  /0b ðlb Þdðzv;i ðtÞ  lb Þ ;
M €~ ðtÞ þ C
~ r ðtÞu ~_ ðtÞ þ K
~ r ðtÞu ~ r ðtÞu
~ ðtÞ ¼ p
~ ðtÞ; ð45Þ
ð37a-cÞ
where the subscript ‘‘r” stands for ‘‘relative displacements” and
in which
where
d /b ðzÞ " #
2
 00  T ~ vb ðtÞ
¼ /b;1 ðzÞ; /b;2 ðzÞ; ...; /b;nb ðzÞ z¼z
00 00
/00b ðzv;i ðtÞÞ ¼ ; ~ r ðtÞ ¼ Mv M
dz2 v;i ðtÞ M ;
z¼zv;i ðtÞ ~ T ðtÞ In þ DM
M ~ b ðtÞ
ð38Þ " vb b
#
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Cv ~ vb ðtÞ
C
2 5
~ r ðtÞ ¼
C ;
with /00b;j ðzÞ
¼ j p 2=ðqb Ab lb Þ sinðjpz=lb Þ.
2
0n nv Nb þ DC ~ b ðtÞ
Then, the equations governing the motion of the nv oscillators " b # ð46a-eÞ
Kv K~ vb ðtÞ
can be written in a compact matrix form as (analogously to Eq. ~ r ðtÞ ¼
K ;
(15)):
2
0nb nv Xb þ DK ~ b ðtÞ



~ vb ðtÞq
€ v ðtÞ þ M ~ vb ðtÞq_ b ðtÞ þ Kv uv ðtÞ
€ b ðtÞ þ Cv u_ v ðtÞ þ C uv ðtÞ Mv sv g
Mv u ~ ðtÞ ¼
u ~ ðtÞ ¼
; p :
~ qb ðtÞ pb ðtÞ
þ Kvb ðtÞq ðtÞ ¼ Mv sv g;
b ð39Þ
As opposed to the formulation in terms of absolute displace-
where the array of the absolute displacements of the moving oscil- ments, the mass matrix M ~ r ðtÞ is not diagonal and depends on
T
lators is uv ðtÞ ¼ f uv;1 ðtÞ; uv;2 ðtÞ; . . . ; uv;nv ðtÞ g , and where time. Furthermore, both the damping and stiffness matrices,
~ vb ðtÞ ¼ Mv Xv ðtÞUT ðtÞ; ~ r ðtÞ and K
C ~ r ðtÞ, are not symmetric. Finally, impulsive terms arise
M bv
in the stiffness matrix K ~ r ðtÞ at the time instants in which each
~ vb ðtÞ ¼ 2Mv Xv ðtÞZ_ v ðtÞU0T ðtÞ;
C
h
bv
i oscillator enters and exits the simply supported bridge.
ð40a-cÞ
~ vb ðtÞ ¼ Mv Xv ðtÞ Z_ 2 ðtÞU00T ðtÞ þ Z
K € v ðtÞU0T ðtÞ As mentioned in the previous section, from a numerical point
v bv bv
  of view, it is preferable to homogenise the physical dimensions
þ Mv Z_ 2v ðtÞ Dv;0 ðtÞsv /0T 0T
b ð0Þ  Dv;1 ðtÞsv /b ðlb Þ ; of the elements of the matrices appearing in the coupled equa-
tions of motion. To do this, the position uv ðtÞ ¼ M1=2 qv ðtÞ is
with v

  made, so that Eq. (45) can be rewritten as (analogous to


U00bv ðtÞ ¼ /00b ðzv;1 ðtÞÞ; /00b ðzv;2 ðtÞÞ; . . . ; /00b ðzv;nv ðtÞÞ ; Eq. (23))
 
€ v ðtÞ ¼ Diag €zv;1 ðtÞ; €zv;2 ðtÞ; . . . ; €zv;n ðtÞ ;
Z €~ ðtÞ þ ~cðtÞq
v ~ q
mðtÞ ~_ ðtÞ þ kðtÞ
~ q ~ ðtÞ ¼ p
~ ðtÞ; ð47Þ
Dv;0 ðtÞ ¼ Diag½ dðzv;1 ðtÞÞ; dðzv;2 ðtÞÞ; . . . ; dðzv;nv ðtÞÞ ;
where
Dv;1 ðtÞ ¼ Diag½ dðzv;1 ðtÞ  lb Þ; dðzv;2 ðtÞ  lb Þ; . . . ; dðzv;nv ðtÞ  lb Þ : " #
Inv M1=2 M~ vb ðtÞ
ð41a-dÞ ~
mðtÞ ¼ v
;
M~ T ðtÞM1=2 In þ DM ~ b ðtÞ
By applying to Eq. (35) the coordinate transformation of Eq. (3), vb v b
" #
the expression of the interaction force in terms of beam’s modal Nv ~ vb ðtÞ
Mv1=2 C
~cðtÞ ¼ ;
coordinates is obtained (analogous to Eq. (18)): 0nb nv Nb þ DC ~ b ðtÞ
" # ð48a-dÞ
~ T ðtÞq ðtÞ  m u X2v M1=2 K~ vb ðtÞ
~ Tvb;i ðtÞq
fv;i ðtÞ ¼ mv;i g  m € b ðtÞ  ~cTvb;i ðtÞq_ b ðtÞ  k v;ib b v;i € v;i ðtÞ; ~ ¼
kðtÞ v
;
0nb nv X2b þ DK ~ b ðtÞ
ð42Þ

qv ðtÞ
where use has been made of Eq. (37a). Finally, substituting Eq. (42) ~ ðtÞ ¼
q :
qb ðtÞ
into Eq. (5), the set of nb ordinary differential equations with time-
dependent coefficients governing the beam’s modal coordinates The corresponding initial conditions at the time instant t0
(Eq. (5)) takes the form (analogous to Eq. (19)): immediately before the first moving oscillator reaches the beam
h i can be defined as (analogous to Eq. (26a,b)):
~ b ðtÞq
½Inb þ DM € b ðtÞ þ M~ T ðtÞu ~ b ðtÞ q_ b ðtÞ
€ v ðtÞ þ Nb þ DC ( 1=2
)

h i
vb
X2
v Mv s v g 0nv 1
~ ðt 0 Þ ¼
q ~_ ðt0 Þ ¼
;q : ð49a; bÞ
þ X2b þ DK ~ b ðtÞ q ðtÞ ¼ p ~ b ðtÞ; ð43Þ ðsÞ
qb 0nb 1
b

where The equations of motion (47) can be integrated by means of a


suitable step-by-step algorithm, paying special attention to the
~ b ðtÞ ¼ Ubv ðtÞMv Xv ðtÞUT ðtÞ ¼ Ubv ðtÞM
DM ~ vb ðtÞ; ~
bv impulsive terms appearing in the stiffness matrix kðtÞ at the en-
~ b ðtÞ ¼ 2Ubv ðtÞMv Xv ðtÞZ_ v ðtÞU0T ðtÞ ¼ Ubv ðtÞC
DC ~ vb ðtÞ; trance and exit times of each oscillator.
bv
h i
~ b ðtÞ ¼ Ubv ðtÞMv Xv ðtÞ Z_ 2 ðtÞU00T ðtÞ þ Z
DK € v ðtÞU0T ðtÞ
v bv bv
  5. Numerical applications
þUbv ðtÞMv Z_ 2v ðtÞ Dv;0 ðtÞsTv /0T T 0T
b ð0Þ  Dv;1 ðtÞsv /b ðlb Þ
~ vb ðtÞ;
¼ Ubv ðtÞK For validation and comparison purposes, the alternative formu-
ðsÞ ðvÞ lations presented in the previous sections have been applied to
~ b ðtÞ ¼ pb ðtÞ ¼
p pb þ pb ðtÞ:
study the transverse vibration experienced by an homogeneous
ð44a-dÞ simply supported beam subjected to a pair of moving oscillators
G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094 1091

 
(nv ¼ 2Þ. The following mechanical parameters have been selected qv;i ðtÞ
fv;i ðtÞ ¼ mv;i g þ kv;i pffiffiffiffiffiffiffiffi  aTb ðzv;i ðtÞÞqb ðtÞ
for the supporting beam: lb ¼ 27:5 m, qb ¼ 2500 kg=m3 , Eb ¼ mv;i
35:0  106 kN=m2 , Ab ¼ 0:954 m2 , J b ¼ 0:355 m4 and fb ¼ 0:020.  
q_ v;i ðtÞ T
The first moving oscillator, having lumped inertia mv;1 ¼ þ cv;i pffiffiffiffiffiffiffiffi  aTb ðzv;i ðtÞÞq_ b ðtÞ  bb ðzv;i ðtÞÞqb ðtÞ ; ð50Þ
mv;i
9; 840 kg, elastic stiffness kv;1 ¼ 7:07  103 kN=m and viscous
damping coefficient cv;1 ¼ 52:8 kN s=m, enters the beam at the time qv;i ðtÞ and q_ v;i ðtÞ being the ith elements of the vectors qv ðtÞ and q_ v ðtÞ,
instant t ¼ 0:000 s with constant speed z_ v;1 ðtÞ ¼ 26:1 m=s respectively. The time histories so obtained are depicted with solid
(¼ 94:0km=hÞ. The second moving oscillator, characterized by thick lines in Figs. 4 and 5.
mv;2 ¼ 3280 kg, kv;2 ¼ 6:54  103 kN=m and cv;2 ¼ 14:6 kN s=m, en- In a second stage, the formulation in terms of relative displace-
ters the beam at t ¼ 0:527 s with z_ v;2 ðtÞ ¼ 39:1 m=s (¼ 141 km=hÞ. ments (Section 4) has been applied. With respect to the first one,
It is assumed that the only static load acting on the supporting this second formulation proves to be more cumbersome, mainly
ðsÞ
beam is its self weight, fb ðtÞ ¼ qb Ab g ¼ 23:4 kN=m. The first five because of the impulses arising each time in which a moving oscil-
modal shapes are retained in the analyses (nb ¼ 5Þ. lator enters (zv;i ðtÞ ¼ 0Þ or exits (zv;i ðtÞ ¼ lb Þ the supporting beam.
If total mass of the beam, mb ¼ qb Ab lb ¼ 65; 600 kg, first period Moreover, the second-order derivatives of the modal shapes of
of vibration of the beam, T b;1 ¼ 2p=xb;1 ¼ 0:211 s, and acceleration the beam, /00b;j ðzÞ, as well as the horizontal accelerations of the mov-
of gravity, g ¼ 9:81 m=s2 , are assumed as independent dimensional ing oscillators, €zv;i ðtÞ, are required in the analysis.
parameters, beam and oscillators are fully defined by the following Since the Newmark’s b-method, like other standard numerical
dimensionless quantities: flexural stiffness Eb J b =ðmb g 3 T 4b;1 Þ ¼ schemes, does not allow to include directly the impulses, the Dir-
101; 000, length lb =ðgT 2b;1 Þ ¼ 63:0 and viscous damping ratio ac’s delta function has been approximated in the form of a Gauss-
fb ¼ 0:020 for the beam; mass ratio mv;1 =mb ¼ 0:150, frequency ian Probability Density Function (PDF), with zero mean and
ratio xv;1 T b;1 =ð2pÞ ¼ 0:900, viscous damping ratio fv;1 ¼ 0:100 standard deviation r  lb :
and speed z_ v;1 =ðgT b;1 Þ ¼ 12:6 for the first moving oscillator;  
1 1 z 2
mv;2 =mb ¼ 0:050, xv;2 T b;1 =ð2pÞ ¼ 1:500, fv;2 ¼ 0:050 and z_ v;2 = dr ðzÞ ¼ pffiffiffiffiffiffiffi exp  : ð51Þ
2pr 2 r
ðgT b;1 Þ ¼ 18:9 for the second moving oscillator (lighter and faster).
In a first stage, the formulation in terms of absolute displace- It is worth noting that, in order to get accurate results, the standard
ments (Section 3) has been applied. The unconditionally stable deviation r has to be properly selected. On the one hand, this param-
Newmark’s b-method, with coefficients b ¼ 1=4 and c ¼ 1=2 eter should be small enough, since the generalized function dðzÞ can
(constant average acceleration), has been used in order to solve be theoretically defined as the limit of the PDF dr ðzÞ when the stan-
the equations of motion in terms of generalized displacements dard deviation r goes to zero; on the other hand, the PDF dr ðzÞ can
qv ðtÞ for the moving oscillators and qb ðtÞ for the supporting beam. be accurately represented in the numerical solution only if the
The time step has been chosen as Dt ¼ p=ð10xb;nb Þ ¼ 0:000422 s, parameter r is greater than the distance covered in the time step Dt
which allows representing the contribution of the highest mode by the fastest moving oscillator entering or exiting the supporting
of vibration retained in the analysis. Once the numerical integra- beam. In Fig. 6, three approximate impulse functions dr ðzÞ, for three
tion has been performed in the time interval 0:1s 6 t 6 2:0s, all different values of the dimensionless ratio r=lb ¼ 0:04, 0:02 and
the statical and kinematical quantities of interest can be com- 0:008, are depicted in the same horizontal scale as the sketch of the
puted starting from the knowledge of qv ðtÞ and qb ðtÞ, e.g. the simply supported bridge. From a qualitative comparison it emerges
beam’s deflection at the midspan position is given by that for r=lb ¼ 0:008 the Gaussian PDF dr ðzÞ provides a good approx-
ub ðlb =2; tÞ ¼ /Tb ðlb =2Þqb ðtÞ, while the interaction force between imation of the Dirac’s delta function, while for r=lb ¼ 0:04 the Gauss-
the beam and the ith oscillator is given by ian PDF is spread over an excessively large interval.

Fig. 4. Time history of beam deflection at mid-span: (a) solutions provided by the absolute and relative displacement formulations (with and without impulses); (b)
enlargement showing the comparison between the responses obtained modeling the impulses as Gaussian PDFs with different values of the ratio r=lb .
1092 G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094

Fig. 5. Time histories of the interaction forces obtained by applying the absolute and relative displacement formulations (with and without impulses): (a) force transmitted
by the first moving oscillator; (b) force transmitted by the second moving oscillator; (c and d) enlargements showing the comparison between the solutions obtained
modeling the impulses as Gaussian PDFs with different values of the ratio r=lb .

For each of the ratios r=lb specified above, the dynamic re-
sponse of the beam-oscillator coupled system has been numeri-
cally evaluated in terms of midspan deflection of the beam,
ub ðlb =2; htÞ ¼ /Tb ðlb =2Þqb ðtÞ, and beam-oscillator interaction forces,
i pffiffiffiffiffiffiffiffi
~_ v;i ðtÞ = mv;i , q
~v;i ðtÞ þ cv;i q
fv;i ðtÞ ¼ kv;i q ~_ v;i ðtÞ being the ith
~v;i ðtÞ and q
~ ~_
elements of the vectors qv ðtÞ and qv ðtÞ, respectively. The time his-
tories so obtained are compared in Figs. 4 and 5 with those pro-
vided by the formulation in terms of absolute displacements,
where different symbols denote different ratios r=lb . As expected,
the results obtained with the smallest value of the ratio r=lb are
in good agreement with those provided by the formulation in
terms of absolute displacements of the moving oscillators, while
the accuracy reduces as the ratio r=lb increases. This tendency is
specifically highlighted in the enlargements depicted on the
right-hand side of Figs. 4 and 5. For this specific example, more-
over, one can see that the peak response of the beam is slightly
underestimated by the formulation in terms of relative displace-
ments (Fig. 4b), while the response of the moving oscillators is
slightly anticipated (Fig. 5d). These differences, however, are abso-
lutely negligible from an engineering point of view. For the sake of
completeness, in Figs. 4 and 5 the time histories obtained by
neglecting the impulses in the formulation in terms of relative dis-
placements of the moving oscillators are also depicted (dashed
thick lines). In the latter case the bridge-vehicle interaction forces,
which are strictly related to the absolute acceleration experienced
by the moving oscillators (Eqs. (8) and (29)), and hence to the com-
fort of the passengers, are hugely inaccurate (Fig. 5), and this is
only due to the neglected impulses. Minor discrepancies also exist
Fig. 6. Impulse function approximated as a Gaussian PDF for three different values
in the midspan deflection of the beam (Fig. 4). These numerical re-
of the ratio r=lb . sults clearly demonstrate that (i) the impulses theoretically de-
G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094 1093

( )
duced in Section 4 really exist, and that (ii) their effects cannot be M1=2 2
v Xv sv g
neglected when the formulation in terms of relative displacements
~ ðtÞ ¼ CðtÞqðtÞ þ
q ; ðA:3Þ
0nb 1
is applied.
being
" #
6. Conclusions Inv M1=2 ~ vb ðtÞ
M
v
CðtÞ ¼ ;
0nb nv Inb
The ‘‘moving oscillator” problem, extensively adopted in the lit- " # ðA:4a-cÞ
  ðaÞ
erature to simulate bridge-vehicle dynamic interaction, has been qv ðtÞ qv ðtÞ
~ ðtÞ ¼
q ; qðtÞ ¼ :
reviewed in this paper, with the aim of providing a deeper insight qb ðtÞ qb ðtÞ
into some theoretical and physical issues not specifically addressed
by other investigators. Without lack of generality, the case of a ~ ðtÞ
Then, the first- and second-order time derivatives of the vector q
simply supported Bernoulli–Euler beam carrying multiple moving defined in Eq. (A.3) are given by
oscillators has been considered. For comparison purposes, the gov-
~_ ðtÞ ¼ CðtÞqðtÞ
q _ _
þ CðtÞqðtÞ;
erning equations have been derived both in terms of relative and ðA:5a; bÞ
absolute displacements of the oscillators. It has been shown that ~€ ðtÞ ¼ CðtÞq
q _ qðtÞ
€ ðtÞ þ 2CðtÞ _ €
þ CðtÞqðtÞ
impulsive terms, i.e. Dirac’s delta functions, appear in the relative
where
displacement formulation when a vehicle enters or exits the beam.
" #
To the authors’ knowledge, presence and relevance of such terms 0nv nv ~ vb ðtÞ
 12 Mv1=2 C
_
CðtÞ ¼ ;
have been not revealed in the past, probably because the attention
0nb nv 0nb nb
in most of the previous studies has been mainly focused on the " # ðA:6a; bÞ
time interval in which the vehicle travels on the bridge, rather than 0nv nv M1=2 ~ vb ðtÞ
K

CðtÞ ¼ v
:
on the dynamic effects arising when the vehicle crosses the sup- 0nb nv 0nb nb
ports of the bridge. Numerical results have demonstrated that in
the context of the relative displacement formulation these impul- In order to derive the equations of motion Eq. (23) in terms of
sive terms, in some sense ‘‘hidden” in the absolute displacement absolute displacements from those in terms of relative displace-
formulation, cannot be neglected, since they prove to have a signif- ments, substitute Eq. (A.3) into Eq. (46) and premultiply the result-
icant influence on the response of the moving oscillators. Further- ing equation by CT ðtÞ. Taking into account Eq. (A.5a,b), and
more, it has been shown that these impulses make the relative observing that the following relationships hold:
displacement formulation more cumbersome, both theoretically CT ðtÞmðtÞCðtÞ
~ ¼ Inb þnv ;
and practically, than the one in terms of absolute displacements. T
 
~
C ðtÞ cðtÞCðtÞ þ 2mðtÞ~ CðtÞ _ ¼ cðtÞ;
On the other hand, the formulation of the problem in terms of rel- h i
ative displacements of the moving oscillators might be properly T ~ ~ _ ~
C ðtÞ kðtÞCðtÞ þ cðtÞCðtÞ þ mðtÞCðtÞ € ¼ kðtÞ; ðA:7a-dÞ
exploited to cope with more complicated models of vehicles. Spe- " ( )#
2 1=2
cifically, in this case it appears more advantageous to define first Xv Mv sv g
CT ðtÞ p~ ðtÞ  kðtÞ ¼ pðtÞ;
the mass, damping and stiffness matrices in terms of relative dis- 0nb 1
placements resorting to the substructure approach. Then, in order
to perform the dynamic analysis, the corresponding matrices in the Eq. (23) can be easily derived.
space of absolute displacements could be derived by using an In a similar way, Eq. (46) can be readily deduced from the abso-
appropriate extension of the transformations of coordinates pro- lute displacement formulation by introducing the following coor-
vided in Appendix for a stream of moving oscillators. dinate transformation into Eq. (23)
( )
~ q M1=2 2
v Xv sv g
qðtÞ ¼ CðtÞ ~ ðtÞ  ðA:8Þ
Appendix 0nb 1

where
In this Appendix, the coordinate transformations useful to ob-
" #
tain the equations of motion of the combined system (supporting Inv M1=2 ~ vb ðtÞ
M
bridge-moving oscillators) in terms of absolute displacements
~
CðtÞ ¼ C1 ðtÞ ¼ v
ðA:9Þ
0nb nv Inb ;
starting from the formulation in terms of relative displacements,
and vice versa, are derived. To this aim, it is recalled that (see is a time-dependent transformation matrix analogous to CðtÞ.
Fig. 3 and Eq. (28)):
References
uv ðtÞ ¼ uðaÞ ðsÞ
v ðtÞ  uw ðtÞ þ uv ðtÞ; ðA:1Þ
ðaÞ Biondi, B., Muscolino, G., 2005. New improved series expansion for solving the
where uv ðtÞ and uv ðtÞ are the nv -dimensional vectors collecting the
ðaÞ moving oscillator problem. Journal of Sound and Vibration 281, 99–117.
relative displacements uv;i ðtÞ and the absolute displacements uv;i ðtÞ Bracewell, R., 1999. The Fourier Transform and Its Applications, Third ed. McGraw-
of the moving oscillators, respectively with and without the static Hill, New York.
ðsÞ ðsÞ Chang, T.-P., Lin, G.-L., Chang, E., 2006. Vibration analysis of a beam with an internal
contributions uv;i ¼ Mv;i g=K v;i ; while uw ðtÞ and uv ðtÞ are the vectors
hinge subjected to a random moving oscillator. Journal of Solids and Structures
listing the displacements of the ideal point wheels and the static 43, 6398–6412.
displacements of the suspended masses, respectively. These vectors Chatterjee, P.K., Datta, T.K., Surana, C.S., 1994. Vibration of continuous bridges under
can be written as follows: moving vehicles. Journal of Sound and Vibration 169 (5), 619–632.
Chen, Y., Tan, C.A., Bergman, L.A., 2002. Effects of boundary flexibility on the
~ vibration of a continuum with a moving oscillator. Journal of Vibration and
uw ðtÞ ¼ M1
v Mvb ðtÞqb ðtÞ;
ðA:2a; bÞ Acoustics, Transactions of the ASME 124 (4), 552–560.
ðsÞ
uv ðtÞ ¼ K1
v Mv sv g:
Cheng, Y.S., Au, F.T.K., Cheung, Y.K., Zheng, D.Y., 1999. On the separation between
moving vehicles and bridges. Journal of Sound and Vibration 222 (5), 781–801.
ðaÞ ðaÞ
Taking into account that uv ðtÞ ¼ Mv1=2 qv ðtÞ and Frýba, L., 1999. Vibration of Solids and Structures Under Moving Loads. Thomas
1=2 Telford, London.
uv ðtÞ ¼ Mv qv ðtÞ, after very simple algebra, the following coordi- Makris, N., 1997. Stiffness, flexibility, impedance, mobility, and hidden delta
nate transformation is obtained: function. Journal of Engineering Mechanics ASCE 123 (11), 1202–1208.
1094 G. Muscolino et al. / International Journal of Solids and Structures 46 (2009) 1085–1094

Muscolino, G., Palmeri, A., 2007. Response of beams resting on viscoelastically Pesterev, A.V., Bergman, L.A., 1998. Response of a nonconservative continuous
damped foundation to moving oscillators. International Journal of Solids and system to a moving concentrated load. Journal of Applied Mechanics ASME 65,
Structures 44, 1317–1336. 436–444.
Muscolino, G., Benfratello, S., Sidoti, A., 2001. Dynamics analysis of distributed Pesterev, A.V., Yang, B., Bergman, L.A., Tan, C.-A., 2001. Response of elastic
parameter system subjected to a moving oscillator with random mass, velocity continuum carrying multiple moving oscillators. Journal of Engineering
and acceleration. Probabilistic Engineering Mechanics 17 (1), 63–72. Mechanics 127 (3), 260–265.
Muscolino, G., Palmeri, A., Sofi, A., 2007a. Random vibrations of linear oscillators Pesterev, A.V., Bergman, L.A., Tan, C.A., Tsao, T.-C., Yang, B., 2003. On asymptotics of
moving on rough beams. In: Proceedings of 10th International Conference on the solution of the moving oscillator problem. Journal of Sound and Vibration
Applications of Statistics and Probability in Civil Engineering (ICASP10), Tokyo, 260, 519–536.
Japan. Stăncioiu, D., Ouyang, H., Mottershead, J.E., 2008. Vibration of a beam excited by a
Muscolino, G., Palmeri, A., Sofi, A., 2007b. Stochastic analysis of linear oscillators moving oscillator considering separation and reattachment. Journal of Sound
crossing beams with random surface roughness. In: Proceedings of and Vibration 310, 1128–1140.
International Symposium on Recent Advances in Mechanics, Dynamical Tzou, H.S., Bergman, L.A., 1998. Dynamics and Control of Distributed Systems.
Systems and Probability Theory (MDP2007) Palermo, Italy. Cambridge University Press, New York.
Pesterev, A.V., Bergman, L.A., 1997a. Response of elastic continuum carrying moving Yang, B., Tan, C.A., Bergman, L.A., 2000. Direct numerical procedure for solution of
linear oscillator. Journal of Engineering Mechanics ASCE 123 (8), 878–884. moving oscillator problems. Journal of Engineering Mechanics 126 (5), 462–
Pesterev, A.V., Bergman, L.A., 1997b. Vibration of elastic continuum carrying 469.
accelerating oscillator. Journal of Engineering Mechanics ASCE 123 (8), 886– Yang, B.Y., Yau, J.D., Wu, Y.S., 2004. Vehicle – Bridge Interaction Dynamics. World
889. Scientific Publishing Co., Singapore.

You might also like