Computer Modeling for the Complex Response Analysis

of Nonstandard Structural Dynamics Problems

Abhimanyu Kumar1 and Atef F. Saleeb2

Abstract: Over the past several decades, two intriguing classes of problems, having a wide range of applications in engineering, have
been of interest to many researchers: 共1兲 coupled dynamics of a distributed parameter system traversed by one or more moving oscillators;
Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

and 共2兲 transient dynamic analysis of axially moving media 共and associated phenomena of parametric resonances兲. Bridge vehicle
interaction falls into the first class of problems, and the analysis of flexible appendages deployed from a satellite or a spacecraft is typical
of the second class. Mathematically, these two problems are dual to each other, and they often are highly nonlinear in nature and typically
involve large overall motion in space with complex effects of convective inertia terms in their governing equations of dynamic equilib-
rium. The “nonstandard” analytical nature of these problems stems from the fact that we are dealing with one or more of the following
peculiarities: 共1兲 variable problem domain; 共2兲 varying spatial distribution of forces over the time duration of the analysis; and/or 共3兲
changing location and type of constraints. Many researchers are trying to formulate the response of these problems, each having a
different approach, but applicable only to certain specific details. Moreover, few researchers have concluded that these problems are
beyond the scope of the present commercial finite-element 共FE兲 software packages. However, we believe that if the nature and details of
these problems are studied properly and carefully, it is immediately possible to simulate these problems in available commercial FE
programs. An added advantage would also be the avoidance of many unrealistic simplifying assumptions that are often introduced to
reduce the mathematical complexity; e.g., neglecting possible separation 共after periods of prior contacts兲 in beam-moving vehicle prob-
lems, assuming linear behavior of suspension systems, and restriction to beam configuration only, among many others. For demonstration,
we use ABAQUS in a large number of test cases to be presented. The results are compared with those presented in literatures.
DOI: 10.1061/共ASCE兲0893-1321共2009兲22:3共324兲
CE Database subject headings: Bridges; Vehicles; Parameters; Computer models; Structural dynamics; Aerospace engineering.

Introduction lems involve forces that are changing their spatial location with
time, leading to singular terms in mass, stiffness, and damping
Motivation matrices of standard finite-element analysis 共FEA兲 approach, and
highly coupled nonlinear system of differential equations due to
This work is concerned with two spatial classes of dynamics; i.e., interaction between system and moving subsystem. Despite
共1兲 a primary 共continuous or distributed-parameters兲 system tra- physical differences, the essence of these problems is the same,
versed by a set of moving discrete sub- 共secondary-兲 systems, as and hence, dual to each other.
typical in bridge vehicle interaction 共BVI兲; and 共2兲 the problem of The available extensive literature on this subject brings out
axially moving media 共AMM兲. Each of these problems has a long many solution approaches that have been developed over the de-
standing history in mechanics and has become especially impor- cades. Most of the formulations are problem specific and often
tant in recent years for many applications, ranging from transpor- consider idealized cases only, such as a simply supported beam,
tation systems in civil engineering to advanced processes of continuous contact between system and subsystem, and linear
deployment and retrieval in space exploration industries. The dy- geometric deformations. A glimpse of different approaches for
namic response quantities, such as deflections, stresses, etc., are BVI problems can be found in Yu and Chan 共2007兲. Moreover,
always higher than those in the static case and thus require accu- each formulation has a special numerical scheme, and thus re-
rate analysis of forces and stresses for reliable design and service quire its independent implementation which may not be appli-
life prediction. Unlike standard dynamic problems, where the cable to other similar problems. However, there is notable paucity
forces 共e.g., applied load, inertia, and elastic interaction force兲 are in the available literature in the use of commercial, large-scale,
fixed in their spatial locations, these nonstandard dynamic prob- FEA codes. Our main motivation here is to fill such a gap. For a
greater benefit to the practicing engineers, we do not intend to
1
Graduate Student, Dept. of Civil Engineering, Univ. of Akron, show “bias” toward the use of either explicit or implicit dynam-
Akron, OH 44325-3905. ics, therefore we selectively report a number of results for each of
2
Professor, Dept. of Civil Engineering, Univ. of Akron, Akron, the options 共ABAQUS/Explicit versus ABAQUS/Standard兲 for
OH 44325-3905 共corresponding author兲. E-mail: [email protected] demonstration purposes only.
Note. This manuscript was submitted on September 12, 2008; ap-
proved on January 22, 2009; published online on June 15, 2009. Discus-
sion period open until December 1, 2009; separate discussions must be Scope and Formulation
submitted for individual papers. This paper is part of the Journal of
Aerospace Engineering, Vol. 22, No. 3, July 1, 2009. ©ASCE, ISSN We provide a solution using the ABAQUS program for a number
0893-1321/2009/3-324–330/$25.00. of BVI and AMM problems to compare with the extensive avail-

324 / JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009

J. Aerosp. Eng. 2009.22:324-330.

Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 1. Overall strategy for modeling of BVI and AMM problems in standard FEA packages

able literature. To this end, the primary systems can have any bility and complexity involved in their mathematical model. All
configuration 共i.e., cable, beam, plate, and shell configurations兲 finite-element simulations are performed in ABAQUS, treating
and we account for intermittent impact conditions 共contact and each problem as a nonlinear dynamic problem, involving small or
separation, as well as the possibility of having a series of moving large deformation mechanics 共depending on the targeted compari-
secondary oscillators兲. Furthermore, we account for the effects of sons with literature and/or the investigated phenomena of inter-
geometric nonlinearities when appropriate 共e.g., pretensioned est兲, frictionless tangential behavior between system and
cable, parametric resonance phenomena, etc.兲. Finally, we provide subsystem, and “hard” pressure overclosure relationship for con-
dynamic response quantities related to both primary and/or sec- tact definition, unless specifically mentioned. Furthermore, the re-
ondary oscillators. sults are presented for final converged mesh after performing
For a background on the mathematical formulations, we refer mesh convergence analysis.
to the typical works of Akin and Mofid 共1989兲, Yang and Yau
共1997兲, Biondi and Muscolino 共2005兲, and Sofi and Muscolino
Dynamic Analysis of Suspended Cable Carrying Two
共2007兲 for BVI; and Vu-Quoc and Li 共1995兲, and Behdinan and
Oscillating Subsystems with Varying Speed
Tabarrok 共1997a,b兲 for AMM problems. Note that the complexity
here stems from the appearance of singular time-dependent terms Although the dynamic analysis of continuum cable, carrying
for the interaction forces 共gravitational, inertia, elastic兲 that are moving subsystem, has been considered as a difficult problem due
spatially varying due to the motion of secondary oscillators. Com- to nonlinear geometry and coupling between cable and sub-
putationally, this leads to the need for varying topology in form- system, it has the same features and complexity involved as other
ing the “finite-element” operators 共inertia, stiffness, load vectors, BVI problems discussed in the literature. The schematic of the
etc.兲 rendering the data structure markedly different from conven- problem is shown in Fig. 2. A circular cross section for the cable
tional FEA approaches. was chosen in order to satisfy cable definition.
The key of our proposed methodology lies in the modeling of The cable was modeled with 50 Euler–Bernoulli beams 共B23兲,
the actual physics of the problem through contact interaction ca- whereas the oscillators were modeled as a spring—point mass—
pabilities of large scale FE software packages and large overall dashpot. The analysis was performed in two steps, accounting for
sliding. In our modeling, the system and subsystems are two in- fully nonlinear geometry and deformations; a static step to get an
dependent sets of objects and the contact/interaction provides a initial catenary shape by applying a uniform line load of magni-
mode of communication for different response quantities 共Fig. 1兲. tude equaling weight per unit length of the cable, followed by a
In our treatments using ABAQUS, we completely bypass the dif- dynamic step to analyze the response of cable due to the motion
ficulty by directly accounting for the spatial variations through of two moving oscillating subsystems. The midspan vertical de-
ABAQUS interaction-effects capabilities. In this approach, the flection of cable obtained at the end of the static step was
use of large sliding is essential, and all the problems are thus 11.7275 m, after accounting for fully nonlinear deformation. In
treated nonlinearly 共for detailed information, see Saleeb and the dynamic analysis step, the spring mass subsystems were
Kumar 2009; Kumar and Saleeb 2009兲. moved with constant velocity, v = 5 m s−1 for the first 90 s and
then a constant deceleration, a = −0.5 m s−2, so that the sub-
systems were brought to rest at t = 100 s. The motion of oscillators
Numerical Examples was defined by displacement amplitude function.
The time-histories of midspan vertical displacement, the verti-
To illustrate the validity of our methodology, six numerical simu- cal displacement at the contact point between the cable and the
lation examples are presented in this paper. The examples are second oscillator, and the slope at midspan, are shown in Fig. 3.
selected from literature on the basis of their engineering applica- The results obtained from ABAQUS/Standard simulation are in

JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009 / 325

J. Aerosp. Eng. 2009.22:324-330.

Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 2. Schematic of suspended cable carrying two oscillating subsystem with varying speed

good agreement with those presented in the reference 共Fig. 7 in Dynamic Response of Plate under Moving Distributed
Sofi and Muscolino 2007兲. As opposed to continuous contact as- Mass
sumption in reference, we considered the more realistic case of
contact where the oscillators were allowed to separate from the To demonstrate the generality and capabilities of present day
contact surface of the cable. Note that ABAQUS/Standard was finite-element software packages, a rectangular plate, simply
able to capture the small amplitude local vibration of cable result- supported at two edges, subjected to a moving distributed mass
ing from impact due to contact between oscillators and cable. The 共Fig. 5兲 is selected from the specialized work done by Wu 共2006兲.
time histories of vertical acceleration of two oscillators are shown Similar to the above reference, the plate was meshed with 60
in Fig. 4. There is a sudden jump in the vertical acceleration of shell elements of Type S4 共10⫻ 6 divisions along A and B, re-
first oscillator when the second oscillator starts moving on cable spectively, in Fig. 5兲. The distributed mass 共labeled M in Fig. 5兲
due instantaneous impact force exerted on the cable by a second was modeled with 64 shell element of 0.025⫻ 0.025 element size
oscillator, since at the time of its entrance the cable was oscillat- giving a total area of 0.04 m2 for distributed mass. Note that
ing. Note that the vertical acceleration of oscillators depicts the ABAQUS recommends using refined mesh for slave surfaces.
comfort of the rider. The thickness of distributed mass was taken as 0.001 m and elas-

Fig. 3. Time history of: 共a兲 midspan vertical deflection of cable; 共b兲 vertical deflection of contact point between oscillator two and cable; and 共c兲
slope at midspan. See figure copy for vertical deflection is plotted with respect to initial static deflection, i.e., 11.7275 m.

326 / JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009

J. Aerosp. Eng. 2009.22:324-330.

Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 6. Time history of midpoint vertical deflection of plate subjected

Fig. 4. Time history of vertical acceleration of two oscillators to moving distributed mass

Dynamic Analysis of Curved Panel Subjected

ticity modulus taken as ten times higher than that of the plate. The to Point Mass
total mass of the distributed mass was 6.5 kg. The Raleigh damp- The most general case that needs to be analyzed in BVI problems
ing coefficients for the plate were calculated using first two natu- has to be motion of traversing secondary subsystem in two-
ral frequencies of vibration of the plate giving an equivalent dimensional space. With this point of view, the dynamic response
damping ratio, ␰, equal to 0.005. of a curved panel subjected to a concentrated mass moving in a
Unlike the previous example in this paper, the linear pressure circumferential direction 共Fig. 8兲, selected from a journal paper
overclosure relationship for surface to surface interaction was de- by de Faria 共2004兲, is being considered here 共all data are given in
fied with contact stiffness 1015, such that it would give exactly the a global coordinate system兲. To our knowledge, only the works
same amount of midpoint static deflection if the distributed mass reported by de Faria 共2004兲 and de Faria and Oguamanam 共2005兲
was placed at midpoint of the plate and “hard” pressure overclo- involve dynamic analysis of curved shells subjected to moving
sure relationship was defined. An equivalent instantaneous pres- mass/load.
sure load was defined over the distributed mass surface to account The curved panel was modeled with shell elements 共30 ele-
for the weight of the distributed mass. The center of distributed ments along an axial direction, and 60 elements along a circum-
mass was moved linearly from x = 0.1 to 0.9 in 0.4 s along the ferential direction兲. In Fig. 8, dark edges represent fixed boundary
center line of plate at a constant speed of 2 m s−1. The time- conditions; i.e., edges parallel to the z axis are fixed, whereas
history of midpoint deflection of the plate is shown in Fig. 6, circumferential edges are free. The moving mass was modeled as
which is in good agreement with that reported in Wu 共2006兲 a point mass with predefined displacement in a global coordinate
共Fig. 5 for the case of dx = 0.1 m, dy = 0.4 m兲 thus, validating our along the circumferential center line of the panel. The motion of
methodology. Also, the contour of maximum in-plane principal mass was obtained by transforming the motion of mass in the
stress distribution at time t = 0.2 s, when the distributed mass was local coordinate system into a global coordinate system as per the
at the midspan of the plate, is shown in Fig. 7. The maximum following expression:
value of stress 共1.613 MPa兲 in the contour plot is appearing at the
midspan of the plate. For a larger value of distributed mass, maxi-
mum stress can reach the yield stress of material. The formulation
reported by Wu 共2006兲, which is limited to linear isotropic elastic
再冉 冊
ux = R sin ⍀t −
␲
12
+ sin
␲
12
冎 共1兲

material, would require a revision in order to deal with such situ- where ux = displacement of moving mass along the x axis.
ations, whereas the problem can still be analyzed in ABAQUS The time history response at the midpoint of the panel is
with their available material models without any modification in shown in Fig. 9 共displacements are plotted as a function of angle
the proposed methodology. calculated at respective times to compare them with the reference
solution兲. The result from ABAQUS/Standard analysis shows
consistency with the reference solution 关Fig. 6 in de Faria
共2004兲兴. Note that ABAQUS/Standard was able to capture small
vibration at the panel’s midpoint, which is not present in the ref-
erence solution. The small amplitude local vibration at the mid-
point of the panel is attributed to the impact due to the
instantaneous contact and separation between the moving mass
and the panel’s surface.

Reverse Spaghetti Problem with Transverse

Tip Loading
Deployment from a rigid channel of a highly flexible beam of
undeformed length, L = 10 and with shear deformation is con-
sidered here 共Vu-Quoc and Li 1995兲. The “cantilever” beam
Fig. 5. Schematic of plate subjected to moving distributed mass 共A = 20, Ixx = Iyy = 1, Ixy = 0, ␳ = 0.1, E = 500, ␯ = 0兲 starts from its

JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009 / 327

J. Aerosp. Eng. 2009.22:324-330.

Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

Fig. 7. Contour plot of max. in-plane principal stress 共N m−2兲 at time t = 0.2 s

zero initial length outside the channel, i.e., the whole beam is The time history trace of tip vertical deflection, obtained from
inside the channel and undergoes exponential sliding motion to its geometric nonlinear analysis, is shown in Fig. 10. This is in
undeformed length as follows: agreement with the reference solution 共Fig. 14 in Vu-Quoc and Li
1995兲. Note that ABAQUS was able to capture the vibration of
u = 10 − 10e−共t/5兲 共2兲
the beam tip due to the impulsive load for time period 关0, 0.1兴
A time varying transverse load, F, is acting on the free end for the along with the vibration of the whole beam. The slight differences
analysis period of 40 s as between ABAQUS/Standard and ABAQUS/Explicit are due to

F共t兲 = 再 30t for t 苸 关0,0.1兴

1 for t 苸 关0.1,40兴
冎 共3兲
the differences in the contact algorithms employed 共see “Interac-
tions,” ABAQUS Analysis Users Manual 2004兲. Also, note that
the number of points to plot the results from ABAQUS/Standard
Once the deployment operation is complete, the fully extended is 472, whereas that in results from ABAQUS/Explicit is 4,001.
flexible beam will eventually undergo some form of free flight in
space and the description “spaghetti” was “coined” for this class Reverse Spaghetti Problem: Repositional Motion
of problem.
Another reverse spaghetti problem with repositional motion
For discretization, 100 “2D Timoshenko beam” elements
共Behdinan and Tabarrok 1997a,b兲 is considered here. The total
共B21兲 were used to model the beam. To restrain all the degree of
length of beam L is 1.05 m 共A = 4.3434⫻ 10−5 m2, Ixx = Iyy
freedoms 共DOFs兲 of the beam inside the channel, the separation
= 1.059⫻ 10−11 m4, Ixy = 0, ␳ = 3,144.39 kg/ m3, E = 6.896
between the beam and rigid channel was made zero. The rigid
⫻ 10 N / m , ␯ = 0兲. With the initial length of 0.35 m outside the
10 2
channel was moved relative to the beam and an exponential lin-
channel, the beam is extruding from the channel in 1.2 s with x
early distributed axial force was applied in order to account for
displacement of the tip varying as

再 冉 冊冎
the inertia effect. The purpose of this was to make analysis easier
on ABAQUS, since element level calculations would not be per- 0.7 1.2 2␲t
formed for rigid parts. u共t兲 = t− sin 共4兲
1.2 2␲ 1.2
To simulate this problem in ABAQUS/Explicit, 100 two-
dimensional Timoshenko beam 共B21兲 elements were used to

Fig. 8. Schematic of curved panel subjected to moving mass Fig. 9. Time history of response of midpoint of curved panel

328 / JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009

J. Aerosp. Eng. 2009.22:324-330.

 Fig. 12. Schematic of model for parametric resonance

E = 1.6761⫻ 1012, ␯ = 0.4748兲 are modified in order to match the

ratios given in Vu-Quoc and Li 共1995兲, Eq. 共5.26兲兴.
Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

The problem was solved in ABAQUS/Standard and ABAQUS/

Explicit with two approaches: first, the beam was moved while
Fig. 10. Reverse spaghetti problem: transverse tip deflection of beam rigid channels were kept at rest and second, when channels were
from ABAQUS/Standard and ABAQUS/Explicit moved relative to the beam. In all the analysis cases, the beam
was discretized by 100 Timoshenko beam elements 共B21兲. Unlike
the reference, the initial geometric perturbation was applied while
model the beam. Unlike reference conditions, a separation of modeling the beam due to limitation of ABAQUS. The cantilever
10−3 m between the rigid channel surface and the beam surface beam member, in the initial configuration, was modeled as a
was invoked in order to allow some motion to the beam elements straight segment inside the channel 关coordinates 共0, 0兲–共0, 6兲兴 and
inside the channel. A transverse force of 0.75 N was applied stati- a slightly inclined segment outside the channel 关coordinates
cally at the tip of the beam for the first 1 s to match the initial 共6,0兲–共22, −0.001兲兴. The channel’s walls were at 0.001 units away
vertical deflection of approximately 0.02 m as presented in the from the beam axis, in order to model a more realistic case, and
reference solution 关Fig. 4 in Behdinan and Tabarrok 共1997b兲兴. For an initial transverse velocity of −0.1 was assigned to the tip of the
the next 1.2 s, rigid channels were moved according to the above beam’s free end. A linear pressure overclosure relationship was
defined displacement function u共t兲 relative to the beam and a chosen for contact normal behavior with contact stiffness of
linearly distributed axial load of total magnitude equaling inertia 2.04984⫻ 108.
force due to axial motion, over the beam length, in the direction The time history traces of tip vertical deflection are shown in
of the beam’s translation was applied to account for the inertia Fig. 13. ABAQUS/Standard completed both runs successfully for
effect. Geometric nonlinearity was accounted for in this analysis. a specified analysis time of 20. Both approaches show consis-
The time history trace of vertical deflection at the beam tip is tency in the case of ABAQUS/Standard, though the run time for
shown in Fig. 11. The slight deviation from the reference solution moving the rigid channel approach was less. ABAQUS/Explicit
关Fig. 4 in Behdinan and Tabarrok 共1997b兲兴 is due to the fact that run was aborted at time equals to 13.84 in the case of a sliding
the initial deflection of the beam tip was not given in the refer- beam approach, whereas the sliding channel run failed after time
ence. Therefore, an approximate force was applied for 1 s in the equal to 18.06. Also, significant differences can be seen, which
initial step to achieve the desired approximate deflection. How- were expected due the differences in contact formulations in
ever, the results are in good agreement, showing similar orders of ABAQUS/Standard and ABAQUS/Explicit. From these results, it
magnitude and shapes. is clear that ABAQUS/Standard is capable of solving the sliding
beam problem using either approach. This problem needs to be
Parametric Resonance of Sliding Beam investigated further in ABAQUS/Explicit.
A cantilever beam problem 共Fig. 12兲 subjected to sinusoidal slid-
ing motion, 6 · sin共2.8␲t兲, is selected to analyze parametric reso-
Summary and Conclusions
nance 共Vu-Quoc and Li 1995兲. The material and cross-sectional
properties 共A = 0.001223, Ixx = Iyy = 1.631⫻ 10−7, Ixy = 0, ␳ = 1,000,
From the viewpoint of practical applicability, the current status of
the available literature on BVI and AMM problems is found to be
incomplete in several respects. First, the reported works remain
largely of a very specialized nature 共requiring high-level math-
ematics not typically within the conventional training of practic-
ing engineers; e.g., advanced perturbation techniques, modified
mode superposition schemes, dynamic greens function, etc., see
Behdinan et al. 1998a,b, 1997a,b兲, with different researchers con-
sidering one or the other of these two problems classes. Almost
always, there is the restriction to the case of beams or cables, and
“unrealistic” assumptions are sometimes made to simplify the
analyses 共e.g., continued/persistent contact, or only the at-rest
conditions for the primary system, irrespective of the number of
traversing secondary oscillators in BVI problems兲. Second, con-
tradicting conclusions were also reported in regard to the com-
monly utilized limiting cases of moving mass and moving load in
Fig. 11. Reverse spaghetti problem 共repositional motion兲: tip vertical BVI problems with different initial conditions for the primary and
deflection secondary dynamical systems 共Pesterev et al. 2003兲.

JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009 / 329

J. Aerosp. Eng. 2009.22:324-330.

 problems to clearly demonstrate the power of commercial FEA
codes 共using the ABAQUS program exclusively here兲. In this, we
bypass completely any complications resulting from the change in
data structure of standard FEA analysis; i.e., by realistically ac-
counting for all interaction forces 共be it of any type such as gravi-
tational, inertia, or elastic in BVI, or of the constraint/Lagrange
multipliers type in AMM problems兲 using the large sliding and
intermittent impulse/impact capabilities in ABAQUS. Conse-
quently, we have been able to consider a rather impressive range
of BVI and AMM problems, ranging from simpler beam/cable
type to plates and shells, accounting for large displacement effects
and many parametric resonance phenomena.
Downloaded from ascelibrary.org by CAMBRIDGE UNIVERSITY on 12/20/14. Copyright ASCE. For personal use only; all rights reserved.

References

ABAQUS. 共2004兲. ABAQUS analysis user’s manual, version 6.5, SIM-

ULIA, Providence, R.I.
Akin, J. E., and Mofid, M. 共1989兲. “Numerical solution for response of
beams with moving mass.” J. Struct. Eng., 115共1兲, 120–131.
Behdinan, K., and Tabarrok, B. 共1997a兲. “Dynamics of flexible sliding
beams—Non-linear analysis. Part I: Formulation.” J. Sound Vib.,
208共4兲, 517–539.
Behdinan, K., and Tabarrok, B. 共1997b兲. “Dynamics of flexible sliding
beams—Non-linear analysis. Part II: Transient response.” J. Sound
Vib., 208共4兲, 541–565.
Behdinan, K., and Tabarrok, B. 共1998a兲. “A finite-element formulation
for sliding beams. Part I.” Int. J. Numer. Methods Eng., 43共7兲, 1309–
1333.
Behdinan, K., and Tabarrok, B. 共1998b兲. “Sliding beams. Part II: Time
Fig. 13. Comparison of tip vertical deflection using two approaches: integration.” Int. J. Numer. Methods Eng., 43共7兲, 1335–1363.
共a兲 ABAQUS/Standard; 共b兲 ABAQUS/Explicit Biondi, B., and Muscolino, G. 共2005兲. “New improved series expansion
for solving the moving oscillator problem.” J. Sound Vib., 281共1–2兲,
99–117.
Third, there has been a notable absence of any applications Biondi, B., Muscolino, G., and Sidoti, A. 共2004兲. “Methods for calcu-
involving large-scale FEA codes. In fact, these were sometimes lating bending moment and shear force in the moving mass problem.”
dismissed 共either explicitly or implicitly兲 at the outset as inappli- J. Vibr. Acoust., 126共4兲, 542–552.
de Faria, A. R. 共2004兲. “Finite element analysis of the dynamic response
cable to the complex situations in BVI problems 共Biondi and
of cylindrical panels under traversing loads.” Eur. J. Mech. A/Solids,
Muscolino 2005; Wu 2006兲. For a case to the point, we summa- 23共4兲, 677–687.
rize here the explicit statement made by Biondi and Muscolino de Faria, A. R., and Oguamanam, D. C. D. 共2005兲. “Adaptive finite-
共2005兲 in conjunction with a solution for a BVI problem and element analysis of the dynamic response of spherical caps under
referring to a standard commercial FEA code ADINA 共Biondi et traversing loads.” Finite Elem. Anal. Design, 41共11–12兲, 1027–1042.
al. 2004兲, as a means for a solution instead of the newly devel- Kumar, A., and Saleeb, A. F. 共2009兲. “On the interaction dynamics of
oped techniques by these authors to solve the problem. These new primary continuous structural systems traversed by oscillating
and highly specialized methods were referred to as conventional subsystems—Part II: Axially-moving media problems.” Adv. Eng.
series expansion 共CSE兲, mode-acceleration method 共MAM兲, and Software, submitted.
dynamic correction method 共DCM兲 共differing in their treatment of Pesterev, A. V., Bergman, L. A., Tan, C. A., Tsao, T.-C., and Yang, B.
共2003兲. “On asymptotics of the solution of the moving oscillator prob-
the bending and shear forces calculations in the BVI analysis兲.
lem.” J. Sound Vib., 260共3兲, 519–536.
For the standard case of an oscillating force at a fixed location on
Saleeb, A. F., and Kumar, A. 共2009兲. “On the interaction dynamics of
a girder, accurate solution from ADINA provided a base for as- primary continuous structural systems traversed by oscillating
sessing the quality of CSE, MAM, and DCM methods. However, subsystems—Part I: Bridge–vehicle interaction problems.” Adv. Eng.
when the motion of the oscillator on the girder was considered Software, submitted.
next, no solution from any standard FEA code was provided for Sofi, A., and Muscolino, G. 共2007兲. “Dynamic analysis of suspended
comparison; the reason being the “poor accuracy” anticipated by cables carrying moving oscillators.” Int. J. Solids Struct., 44共21兲,
the authors from these codes. Similarly, Wu 共2006兲 had to avoid 6725–6743.
the use of any standard finite-element code, and instead created Vu-Quoc, L., and Li, S. 共1995兲. “Dynamics of sliding geometrically-exact
his own specialized program 共accounting for the freshly derived beams: Large angle maneuver and parametric resonance.” Comput.
stiffness, mass, damping, and load arrays兲 to analyze the rectan- Methods Appl. Mech. Eng., 120共1–2兲, 65–118.
Wu, J.-J. 共2006兲. “Use of moving distributed mass element for the dy-
gular plate under moving distributed mass.
namic analysis of a flat plate undergoing a moving distributed load.”
Finally, there is a lack of useful guidelines as to which 共if any兲 Int. J. Numer. Methods Eng., 71共3兲, 347–362.
of the described schemes is appropriate for treating other equiva- Yang, Y.-B., and Yau, J.-D. 共1997兲. “Vehicle-bridge interaction element
lent classes of problems. for dynamic analysis.” J. Struct. Eng., 123共11兲, 1512–1518.
In an attempt to improve the above situation, we have pre- Yu, L., and Chan, T. H. T. 共2007兲. “Recent research on identification of
sented the solutions of a rather extensive set of BVI and AMM moving loads on bridges.” J. Sound Vib., 305共1–2兲, 3–21.

330 / JOURNAL OF AEROSPACE ENGINEERING © ASCE / JULY 2009

J. Aerosp. Eng. 2009.22:324-330.