Professional Documents
Culture Documents
Abstract
The dynamic behaviors of vehicle and track in long and downhill section of high-speed railway are studied under braking condition. A
vehicle–track dynamic interaction model is constructed based on two longitudinal vehicle–track interaction models. In the model, the
vehicle is considered as a 21-degree-of-freedom multi-rigid body system composed of a car body, two bogies, and four wheels; using
the finite-element method, the track is modeled as a Euler beam; the ‘‘circular track’’ method is proposed to reduce the degree of
freedom of the model for the simulation of long-distance track; two longitudinal wheel–rail interaction models are considered: the
Polach creep theory (suitable for simulating condition of large creep resulting from heavy braking) and the longitudinal rigid-contact
theory. The dynamic responses of the substructures during vehicle braking calculated by the models based on the Polach creep model
and longitudinal contact model show little difference, but the Polach creep model can fully consider the motion of the wheels and
large wheel–rail creep in the braking process and can accurately analyze the wheel–rail interface damage. The results of dynamic inter-
action between wheel and rail under different conditions suggest that large braking torque will cause some or all of the wheels to slide
and then cause the damage of wheel–rail interface. The grade will lead to the increase in braking distance and time and also extend
the sliding time of locked wheels, increasing the risk of damage of the wheel–rail interface. The braking torque should be kept below a
reasonable value, so that the braking distance and braking time can be as short as possible without the occurrence of wheel sliding
along the track. A reasonable braking torque under the dry track and wet track conditions should be 7 and 4 kN m respectively,
according to the calculation in this article.
Keywords
braking torque, damage of wheel–rail interface, dynamic response, finite-element method, long-steep grade
1. Longitudinal impulse action of trains. The longi- law of the longitudinal force in welded rails on
tudinal impulse effects caused by train accelera- bridge using a nonlinear track–bridge interac-
tion and deceleration have been investigated tion model.
mainly through multi-mass kinematic models. 3. Longitudinal dynamic problems associated with
For example, Geike (2007) constructed a longi- vehicle–track interaction. In these studies, a
tudinal dynamic model of train with a linear vehicle–track coupling dynamic model is gener-
coupler and draft gear system and studied the ally constructed to analyze the longitudinal
reasons why the longitudinal hook force is too interaction between vehicle and track under
large during the operation of a subway vehicle. braking condition for a given longitudinal rela-
Cole and Sun (2006) compared and analyzed tionship between wheel and rail. For example,
the effects of three different types of hook Zhang et al. (2011) and Cheng et al. (2013) ana-
devices on the longitudinal dynamic perfor- lyzed a coupled vehicle–structure system under
mance of the train to examine factors such as braking condition based on a model of longitu-
the gap between hooks and the impedance hys- dinal rigid wheel–rail contact. Zhu and Yan
teresis of the buffer. Based on multi-mass (2017) and Huang et al. (2018) adopted a simi-
model, Ansari et al. (2009) comprehensively lar method to construct a coupled vehicle–
studied the longitudinal dynamic responses of a bridge dynamic model to investigate the
train and considered the influence of the stiff- dynamic response of bridge with high piers and
ness and damping of the buffer, the running curved box girders under braking condition.
speed of the train, and the distribution of empty Based on a model of longitudinal elastic
and heavy vehicles. Nasr and Mohammadi vehicle–bridge contact, Ju and Lin (2007) pro-
(2010) analyzed the effect of braking-system posed a finite-element model to investigate the
delay on the longitudinal dynamic performance vehicle–bridge interaction (VBI) under the con-
of trains, and the results showed that the com- ditions of braking and vehicle acceleration.
pressive hook force increases with the increase Based on the longitudinal creep contact theory,
in delay, while the tensive hook force decreases Tran et al. (2016) studied the response of
with the increase in delay. multiple-railcar high-speed train under braking
2. Longitudinal dynamic responses of track and condition and the results revealed significant
bridge structures under the action of braking interaction between neighboring railcars when
force. In these studies, fine structure models of the applied braking torque is between the opti-
track and bridge are often constructed, the mal and critical torques. Yang and Wu (2001)
dynamic action of train is simplified as dynamic studied the dynamic characteristics of bridges
load, and the transfer law is focused on the during train deceleration using the VBI element
longitudinal force between track and bridge method.
under braking force. For example, Fryba
(1974a, 1974b) proposed a quasi-static method These previous studies considered the dynamic
to analyze the braking force and driving force responses of vehicle or bridge under braking condition
acting on rail and bridge and constructed the only in flat railway sections. However, according to
first model for integrated analysis of bridge–rail the actual operation, trains are more likely to brake on
interaction. Toth and Ruge (2001) analyzed the long grades, which also bring more safety risks. In
longitudinal dynamic response of long railway such cases, the creep characteristics of wheels, the
bridge under the action of train braking. Bose mechanism of the damage of wheel–rail interface, and
et al. (2018) studied the response of track under the effect of grade on longitudinal vehicle–track inter-
the action of train braking force using analyti- action are still unclear.
cal methods and pointed out that the range of In this article, based on the longitudinal and vertical
influence of the longitudinal response of track interaction between wheel and rail, a vehicle–track
is larger than that of the vertical response. Dai dynamic model is constructed on a grade. The effects
and Liu (2013) studied the effect of small- of braking torque, wheel–rail friction coefficient, line
resistance fasteners on the longitudinal force gradient, rail-surface irregularity, and the effect of flex-
acting on the structures through a finite- ibility of track structure on wheel–rail interaction in
element model of an integrated system com- the braking process are investigated, in the hope of
posed of spatial track bridge and pier founda- providing reference for the design and maintenance of
tion. Ruge and Birk (2007) studied the transfer high-speed railway on long grades.
Lu and Shi 525
mc€zc + 2ksz zc ksz zt1 ksz zt2 + 2csz z_ c csz z_ t1 csz z_ t2 = mc g mt €xtj + ksx h1 uc ksx xc + (ksx h2 2kpx h3 )utj
ð1Þ
+ (2kpx + ksx )xtj kpx xwk kpx xwk + csx h1 u_ c
€ wk + Rw Fk T = 0
jw u ð8Þ
Figure 1. The vehicle model.
526 Advances in Structural Engineering 23(3)
i
G
(a) (b)
Figure 2. The track model: (a) traditional track (b) circular track.
Lu and Shi 527
1. Polach creep contact theory velocity, and B is the coefficient of the exponential
attenuation of friction; constants A and B are acquired
The creep characteristics of wheel–rail contact directly by experiment.
determine the braking performance and affect the run-
ning safety of a vehicle. Many scholars (Kalker, 1990; 2. Longitudinal rigid-contact theory
Spiryagin et al., 2013; Vermeulen and Johnson, 1964)
have studied this issue, and the simplified algorithm The longitudinal rigid-contact theory assumes that the
based on the Kalker theory (Kalker, 1982; the corre- longitudinal displacements of wheel and rail are equal
sponding numerical program is the Fastsim) is now at the contact point, that is, the longitudinal DOF of
widely used in the simulation of vehicle dynamics. wheel is not independent; in addition, the DOFs associ-
However, according to Vollebregt (2014), in this algo- ated with the pitching movement of wheels can also be
rithm, the Coulomb friction model with a constant ignored because the pitching movement and the longi-
friction coefficient is used, so the creep force–creepage tudinal movement of wheels are coupled to each other.
characteristic calculated by the Kalker theory does not According to the assumption of longitudinal rigid con-
agree well with the measured results. Moreover, the tact, the longitudinal wheel–rail force acting on the
Kalker theory cannot describe the negative slope char- track can be calculated as follows
acteristic of the adhesion curve, so its applicability is
poor for the simulation of large wheel–rail creep result- Fk = kpx (xtj xwk h3 utj ) + cpx (_xtj x_ wk h3 u_ tj ) mw€xwk
ing from vehicle braking. ð27Þ
The Polach theory (Polach, 2005) introduced reduc-
tion factors together with a slip-velocity-dependent where Fk denotes the kth longitudinal wheel–rail force
friction coefficient, which decreases with the increase acting on the track; xwk , x_ wk , and €xwk represent the
in longitudinal creepage, so the results are more accu- longitudinal position, velocity, and acceleration of the
rate and the calculation is more efficient than the kth wheel, respectively; xtj and utj represent the longi-
Fastsim. Hence, it is more reasonable to simulate vehi- tudinal position and pitching displacement of the jth
cle braking using the Polach theory. Based on the bogie, respectively; x_ tj and u_ tj represent the longitudi-
Polach creep contact theory, the longitudinal wheel– nal velocity and pitching velocity of the jth bogie,
rail force under large creep condition can be calculated respectively.
rapidly as follows According to the longitudinal rigid-contact theory,
the longitudinal response of the wheels can be directly
imported into the vehicle and track subsystem as
2mk Pk k A ek 1
Fk = + tan ðkS ek Þ ð23Þ known data.
p 1 + (kA ek )2
Gpak bk c11
ek = clk ð24Þ Solving method
4Pk mk
Based on the wheel–rail relationships mentioned
jx_ k Rw u_ wk j above, the kinetic equations of the vehicle subsystem
clk = ð25Þ
x_ k and the track subsystem are sorted out, and then with
where Fk represents the kth longitudinal wheel–rail the Newmark-b numerical integration method, the
force; Pk represents the kth vertical wheel–rail contact dynamic responses of the vehicle and track are calcu-
force; ek is the gradient of tangential stress in the longi- lated through iteration between the vehicle and track
tudinal direction; kA and kS are the reduction factors subsystem at each time step. The convergence of itera-
tion requires that the differences between the vertical
in the adhesion and slip areas, respectively; G is the
wheel–rail contact forces at adjacent steps are less than
shear modulus of rigidity; ak and bk are the semi-axes
100 N. Moreover, a and b are 0.25 and 0.5, respec-
of the contact ellipse at the kth wheel with the constant
tively; the integration step is 0.0002 s. The procedure
value of 6 3 1023 m; c11 is the Kalker coefficient with
of computation is shown in the following chart
the value of 4.12; clk is the longitudinal creepage of the
(Figure 3).
kth wheel; and mk is the frictional coefficient between
the kth wheel and rail and can be described by
Comparison of different longitudinal
mk = m0 (1 A)eBjx_ Rw u_ wk j + A ð26Þ wheel–rail interaction models
In the above equation, A is the ratio of the limit fric- According to the survey data, the actual speed of a
tion coefficient (mN) at infinite slip velocity to the China Railway High-speed (CRH) vehicle running in
maximum frictional coefficient (m0) at zero slip graded areas is in the range of 180–200 km/h. The case
Lu and Shi 529
(a) (b)
Figure 4. Comparison of the two models: (a) the time history of longitudinal wheel–rail force and (b) the time history of
longitudinal displacement of rail.
530 Advances in Structural Engineering 23(3)
(a) (b)
(c) (d)
(e)
Figure 5. Time history of wheel angular speed and vehicle speed under various braking torques: (a) 7 kN m, (b) 8 kN m, (c)
9 kN m, (d) 10.5 kN m, and (e) 11 kN m.
torque, line gradient, and rail-surface irregularity are of the vehicle is set to be 185 km/h (51.3 m/s), and the
investigated on the dynamic responses of vehicle and gradient of grade is set to be 5&.
track. The parameters of the simulated vehicle and Figure 5 lists the time history of the motion of each
track are shown in Appendix 1. wheel when various braking torques are applied.
Based on the difference between the angular speed of
wheels and the speed of vehicle, the braking torques
Braking torque can be roughly divided into five intervals: 0–8 kN m,
Under some unexpected situations, vehicles often need 8–9 kN m, 9–10.5 kN m, 10.5–11 kN m, and 11–
to decelerate and halt when running on a grade. 13 kN m). When the braking torque increases gradu-
Usually, depending on the actual situation, various ally from 0 to 8 kN m (as shown in Figure 5(a), the
braking levels may be applied; more specifically, vari- angular speed of wheel and the speed of vehicle
ous braking torques may be applied to the wheels to decrease proportionally; in this stage, the wheels
halt the vehicle. In order to simulate various braking remain at the rolling state, and the wheel–rail rolling
levels, the braking torque is set to be 3, 5, 7, 8, 9, 10.5, friction forces are used to resist the traveling of vehicle.
11, and 13 kN m alternately. The initial braking speed When the torque continues to increase in the range of
Lu and Shi 531
(a) (b)
Figure 6. Time history of the longitudinal deceleration of vehicle under various braking torques: (a) 7 kN m and (b) 13 kN m.
8–9 kN m (as shown in Figure 5(b)), the angular speed deceleration of vehicle is basically unchanged. When
of the fourth wheel is reduced to zero quickly owing to the braking torque is 13 kN m (Figure 6(b)), the
the resistance provided by the torque; however, at this motion of wheels will change during the braking pro-
time, the speed of vehicle is not zero, so the fourth cess, leading to the variation in longitudinal wheel–rail
wheel no longer roll, and begin to be locked and slide force, which will then affect the longitudinal decelera-
until the halt of vehicle. This wheel sliding will cause tion of vehicle, so the longitudinal deceleration can be
damage to the wheel–rail interface, which will induce divided into three stages based on the motion of
abnormal wheel–rail impact force and exacerbate the wheels: the creep stage, sliding stage, and low-speed
fatigue failure of the vehicle–track structure. When the sliding stage. In the creep stage, with the progress of
braking torque is increased further to the range of 9– braking, the wheel–rail creepage increases gradually,
10.5 kN m, 10.5–11 kN m, and 11–13 kN m (as the longitudinal wheel–rail force decreases correspond-
shown in Figure 5(c) to (e), respectively), the second ingly, and the deceleration of vehicle decreases gradu-
pair of wheels, the third pair of wheels, and the first ally. In the sliding stage, the frictional coefficient is less
pair of wheels will successively change from the rolling affected by the speed of vehicle, and the sliding friction
state to the sliding state. force remains stable, so the deceleration of vehicle
Figure 6 presents the time history of longitudinal keeps basically stable. In the low-speed sliding stage,
deceleration of vehicle when braking torques with vari- the frictional coefficient increases gradually with the
ous levels are applied. When the braking torque is decrease in sliding speed, and the longitudinal wheel–
7 kN m (Figure 6(a)), the wheels have been rolling rail force increases accordingly, leading to gradual
during the braking process, and the longitudinal increase in the deceleration of vehicle.
wheel–rail forces keep basically stable; however, the Figure 7 presents the variations of braking distance
longitudinal wheel–rail force is directly related to the and braking time of the vehicle with the level of brak-
longitudinal response of vehicle, so the longitudinal ing torque. When the braking torque is small, all the
wheels creep, and the braking distance and braking
time decrease noticeably with the increase of the tor-
que. When the torque is greater than 8 kN m, wheel
sliding occurs; with the increase of torque, the decline
of braking distance and braking time is not salient and
it is noted that any further increase in the torque does
not result in any decrease in the braking distance and
braking time, but will cause the damage of wheel–rail
interface. To ensure the safety of vehicle operation, the
braking distance and braking time should be shortened
as much as possible under the premise that the wheels
will not be locked or slide along the surface of rail.
Based on the results calculated in this section, it is
Figure 7. The braking distance and time under various braking
quite reasonable to set the level of braking torque to
torques. be 7 kN m.
532 Advances in Structural Engineering 23(3)
(a) (b)
Figure 8. The responses of vehicle under various braking torque under dry/wet track condition: (a) the sliding time (b) the braking
distance.
It is well known that the friction coefficient of the force, and longitudinal creepage in the braking pro-
wheel–rail interface has a great influence on the longi- cess. The longitudinal deceleration of the vehicle
tudinal creepage; in order to further study the influ- remains stable on the flat track; however, when the
ence of the braking torque on the vehicle response vehicle travels to the grade, the longitudinal decelera-
under different friction coefficients, the case of a train tion suddenly decreases owing to increased load along
traveling on the dry/wet track before experiencing sud- the line direction. The sudden change in the longitudi-
den deceleration due to the application of braking tor- nal deceleration of the vehicle on the grade leads to
ques is considered. Wheel–rail contact parameters are sudden change in the longitudinal velocity of the
shown in Appendix 2. Figure 8 presents the variations wheels, resulting in sudden change in the longitudinal
of sliding time and braking distance of the vehicle with creepage of the wheels at the point of gradient jump.
the level of braking torque under both wet track and The longitudinal wheel–rail force also changes corre-
dry track conditions; it can be seen that when the spondingly. It should be noted that the damage of
braking torque is less than 5 kN m, there is no occur- wheel–rail interface is more likely to happen with the
rence of wheel sliding under two kinds of wheel–rail increase in longitudinal creepage.
contact conditions; in this stage, the braking distance Next, a long enough grade with gradient of 0, 10&,
of the vehicle is also equal. When the torque increases 20&, 30&, and 40& is set up alternately to investigate
to 5 kN m, here are occurrences of wheel sliding under the effect of gradient on vehicle braking in more depth.
wet track condition; at this time, the braking distance The vehicle is decelerated at various braking levels on
is minimized. When the torque reaches 8 kN m, the the grade.
wheel begins to slide under dry track condition and the In view of the fact that the motions of the four
braking distance is minimized. On the whole, under wheels are similar during the process of vehicle braking
wet track condition, the friction coefficient of the (although the fourth wheel begins to slide first with the
wheel–rail interface is smaller and the maximum longi- increase in torque), Figure 10 only shows the sliding
tudinal wheel–rail force is smaller correspondingly; time of the fourth wheel under various gradients when
therefore, it is easier for the wheel to slide; however, braking torques with different levels are applied.
the sliding time and the braking distance are longer. Figure 10 also shows that under various gradient con-
ditions, the maximum braking torque without wheel
sliding is 7 kN m, which is consistent with the result
Gradient obtained in section ‘‘Comparison of different longitu-
The maximum gradients of high-speed railway lines dinal wheel–rail interaction model.’’ Hence, it can be
are generally 20&–40&. In order to analyze the effects concluded that gradient will not alter the reasonable
of different line conditions on the braking of vehicle, a level of braking torque. However, when the wheel
grade with gradient of 40& is set in the braking path begins to slide, the sliding time increases gradually
of the vehicle. The initial braking speed of the vehicle with the increase in gradient, suggesting that the risk
is set to be 185 km/h (51.3 m/s), and the braking tor- of wheel–rail interface damage is greater when the
que is set to be 7 kN m. vehicle brake is on a grade.
Figure 9 presents the time history of the longitudi- Figure 11 shows the longitudinal deceleration, brak-
nal deceleration of the vehicle, longitudinal wheel–rail ing distance, and braking time of the vehicle on a grade
Lu and Shi 533
(a) (b)
(c)
Figure 9. (a) The time history of longitudinal deceleration of vehicle, (b) the time history of longitudinal wheel–rail creepage, and
(c) the time history of longitudinal wheel–rail force.
Random irregularity
The rail surface cannot be smooth in an actual situa-
tion, and random irregularity is present owing to the
combined effects of many factors such as initial bend-
ing, wear, and damage of track. In order to study the
effect of random irregularity on the dynamic interac-
Figure 10. The wheel sliding time under various gradient and tion between wheel and track under braking condi-
torque conditions. tions, superimposition of short-wave irregularity over
medium-long-wave irregularity is adopted to simulate
the actual track irregularity. The initial braking speed
with various gradients, when the wheels are subject to of the vehicle is set to be 185 km/h (51.3 m/s), and the
the braking torque of 7 kN m. With the increase in gradient of the grade is set to be 5&.
gradient, the load along the line direction gradually Figure 12 shows the vertical wheel–rail force of the
becomes larger and the longitudinal deceleration of the first and fourth wheels at two states: on a smooth track
vehicle decreases gradually. The decrease in longitudi- and on a track with irregularity. When the braking tor-
nal deceleration leads to significantly increased braking que is applied, the body and bogies pitch downward
distance and braking time (541 m and 21.3 s on the flat and axle-load transfer occur, so the vertical wheel–rail
534 Advances in Structural Engineering 23(3)
(a) (b)
Figure 11. (a) The longitudinal deceleration of vehicle under various gradient and (b) the braking distance and time under various
gradient.
(a) (b)
(c) (d)
Figure 13. The time history of longitudinal response of vehicle and track: (a) longitudinal wheel–rail force, (b) the longitudinal
deceleration of vehicle, (c) the longitudinal acceleration of rail, and (d) the longitudinal displacement of rail.
(a) (b)
(c) (d)
Figure 14. (a) Time history of the vertical wheel–rail force, (b) frequency spectrum of the vertical wheel–rail force, (c) time history
of the longitudinal wheel–rail force, and (d) frequency spectrum of the longitudinal wheel–rail force.
536 Advances in Structural Engineering 23(3)
attenuation of the wheel–rail force by the elasticity/ torque under the dry track and wet track condi-
damping of the track structure. In the spectrum, it is tions should be 7 and 4 kN m, respectively.
not difficult to find that the vertical wheel–rail force 4. In the process of vehicle braking, the car body
calculated by flexible and rigid track is basically the and bogie pitch downward, causing increased
same when the frequency is lower than 30 Hz. In the load at the front wheel and reduced load at the
frequency range of 30–50 Hz, the vertical wheel–rail rear wheel. The irregularity of rail surface
force of the former is slightly higher than that of exacerbates high-frequency fluctuation of the
the latter. For the high-frequency range exceeding vertical and longitudinal wheel–rail contact
50 Hz, the vertical wheel–rail force when the track is forces, mostly likely leading to the fatigue dam-
flexible is significantly lower than that when the track age of wheel–rail and the corrugation of rail.
is rigid. 5. Since the flexible track structure has elasticity
Figure 14(c) and (d) shows the time history and fre- and damping effect, the vertical wheel–rail
quency spectrum of the longitudinal wheel–rail force. force, especially the high-frequency component,
Similarly, in the presence of the random irregularity, can be effectively buffered and attenuated, and
due to the attenuation effect of track structure on the longitudinal wheel–rail force will also be
vibration, the fluctuation of the longitudinal wheel–rail moderated. Thus, the elasticity of the track
force of flexible track is also lower than that of the structure can improve running stability of the
rigid track. The discrepancy of longitudinal wheel–rail vehicle under the action of braking torque.
force calculated by two models mainly reflected in the
high frequency. Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with
Conclusion respect to the research, authorship, and/or publication of this
article.
In this article, a vehicle–track interaction model on a
grade is constructed based on two longitudinal wheel– Funding
rail interaction models (the Polach creep model and
The author(s) disclosed receipt of the following financial sup-
the longitudinal rigid-contact model), and the dynamic
port for the research, authorship, and/or publication of this
responses of vehicle and track caused by vehicle brak-
article: The research is sponsored by Beijing Municipal
ing on the grade is studied. The following conclusions Natural Science Foundation (Grant No. 8182041) and the
are obtained: National Natural Science Foundation of China (Grant No.
51578054).
1. When the Polach creep model and the longitu-
dinal rigid-contact model are used as the longi- ORCID iD
tudinal wheel–rail interaction model, the
Jin Shi https://orcid.org/0000-0003-2202-0048
calculated dynamic responses of substructure
show little difference. However, the Polach
creep model can fully consider the motion of References
wheels and large wheel–rail creep phenomenon Ansari M, Esmailzadeh E and Younesian D (2009) Longitu-
in the braking process and can accurately ana- dinal dynamics of freight trains. International Journal of
lyze the damage of wheel–rail interface. Heavy Vehicle Systems 16(1/2): 102–131.
2. Large braking torque will cause some or all of Bose T, Levenberg E and Zania V (2018) Analysis of track
the wheels to slide and then trigger the damage responses to train braking. Proceedings of the Institution
of wheel–rail interface. The grade will lead to of Mechanical Engineers, Part F: Journal of Rail and Rapid
the increase in braking distance and time and Transit 232(7): 1984–1993.
extend the sliding time of locked wheel, increas- Cheng Q, Zhang N, Xia H, et al. (2013) Dynamic response
analysis of vehicle-track-bridge system considering brak-
ing the risk of damage at the wheel–rail
ing conditions for high-speed railway. China Railway Sci-
interface.
ence 34(1): 8–14.
3. In order to ensure the safety of running, the Cole C and Sun YQ (2006) Simulated comparisons of wagon
braking torque should be kept below a reason- coupler systems in heavy haul trains. Proceedings of the
able value, so the braking distance and time Institution of Mechanical Engineers, Part F: Journal of
could be shortened as far as possible without Rail and Rapid Transit 220(3): 247–256.
the occurrence of sliding wheel. According to Dai GL and Liu WS (2013) Applicability of small resistance
the calculation in this article, the braking fastener on long-span continuous bridges of high-speed
Lu and Shi 537
railway. Journal of Central South University 20(5): bridge interaction. Computers & Structures 85(7–8):
1426–1433. 458–475.
Fryba L (1974a) Response of a beam to a rolling mass in the Spiryagin M, Polach O and Cole C (2013) Creep force mod-
presence of adhesion. Acta Technica CSAV 19(6): elling for rail traction vehicles based on the Fastsim algo-
673–687. rithm. Vehicle System Dynamics 51(11): 1765–1783.
Fryba L (1974b) Quasi-static distribution of braking and Toth J and Ruge P (2001) Spectral assessment of mesh adap-
starting forces in rails and bridge. Rail International 5(11): tations for the analysis of the dynamical longitudinal
698–716. behavior of railway bridges. Archive of Applied Mechanics
Geike T (2007) Understanding high coupler forces at metro 71(6–7): 453–462.
vehicles. Vehicle System Dynamics 45(4): 389–396. Tran MT, Ang KK and Luong VH (2016) Multiple-railcar
Huang XY, Fu CC, Zhuo WD, et al. (2018) Concrete curved high-speed train subject to braking. International Journal
box girders interacted with vehicles in braking or accel- of Structural Stability & Dynamics 17(7): 1750071.
eration. International Journal of Structural Stability & Vermeulen PJ and Johnson KL (1964) Contact of nonspheri-
Dynamics 18(4): 1850053. cal elastic bodies transmitting tangential Forces. Journal
Ju SH and Lin HT (2007) A finite element model of vehicle– of Applied Mechanics 31(2): 338–340.
bridge interaction considering braking and acceleration. Vollebregt EAH (2014) Numerical modeling of measured
Journal of Sound & Vibration 303(1–2): 46–57. railway creep versus creep-force curves with CONTACT.
Kalker JJ (1982) A fast algorithm for the simplified theory of Wear 314(1–2): 87–95.
rolling contact. Vehicle System Dynamics 11(1): 1–13. Yang YB and Wu YS (2001) A versatile element for analyz-
Kalker JJ (1990) Three-Dimensional Elastic Bodies in Rolling ing vehicle–bridge interaction response. Engineering Struc-
Contact. Dordrecht: Kluwer Academic Publishers. tures 23(5): 452–469.
Nasr A and Mohammadi S (2010) The effects of train brake Zhai WM and Sun X (1994) A detailed model for investigat-
delay time on in-train forces. Proceedings of the Institution ing vertical interaction between railway vehicle and track.
of Mechanical Engineers, Part F: Journal of Rail and Rapid Vehicle System Dynamics 23(Suppl. 1): 603–615.
Transit 224(6): 523–534. Zhang N, Xia H, Cheng Q, et al. (2011) Analysis method for
Polach O (2005) Creep forces in simulations of traction vehi- a vehicle structure coupled system under braking force.
cles running on adhesion limit. Wear 258(7–8): 992–1000. Journal of Vibration & Shock 30(2): 138–143.
Ruge P and Birk C (2007) Longitudinal forces in continu-
ously welded rails on bridgedecks due to nonlinear track–
Appendix 1
Parameters of vehicle and track.
Appendix 2
Parameters for wheel–rail contact conditions.
Notation Value
Dry Wet
kA 1.00 0.3
kS 0.40 0.1
A 0.40 0.4
B 0.60 0.2
m0 0.5 0.25