Você está na página 1de 8

DISP '19, Oxford, United Kingdom

ISBN: 978-1-912532-09-4

Railways: Investigating the Second Order Differential


Equation of Motion for the case of the Suspended-Masses
Konstantinos Sp. Giannakos
Civil Engineer PhD, F. ASCE, M. TRB AR050-060Com.
108 Neoreion str., Piraeus 18534, Greece
tel: +302104133258; mob: +306973304141
kongiann@otenet.gr; kyannak@gmail.com
http://giannakoskonstantinos.com/wp

ABSTRACT a random dynamic load (equal to mass x acceleration) is applied on


the (concrete) sleeper/tie (Fig.1).
The circulation of a railway vehicle on a railway track can be
approached theoretically the second order differential equation of
motion specialized for this system “vehicle-track”. The motion of
a railway vehicle on a railway track is a forced oscillation with a
forcing excitation induced by the rail running table (rail running
surface) –which has the form of a “signal”–, expressed by a
random, non-periodic function. Railway-track, is simulated –with
the observer situated on the wheel– as an elastic means with springs
and dashpots. The railway vehicle has the Suspended-Masses (SM)
and the Non-Suspended Masses (NSM). The track defects/faults
are the random excitation for the rolling wheels of the vehicle. The
case of the Non-Suspended Masses has been analyzed in previous
scientific articles of the author. In the case of the Non-Suspended
Masses the forces resulting from the excitation, due to the track
defects of short wavelength, are large and have grate effect on the
rolling of the wheel. In the case of the Suspended-Masses of the
vehicles, the forces resulting from the excitation, due to the track
defects, when the wavelength of defects is short, are not large and
have small effect on the rolling of the wheel. The present paper
depicts that the longer the wavelength of defects is, the more
predominant the role of the Suspended-Masses is and they mainly Figure 1 (Upper-illustration) cross-section of Ballasted Track
influence the magnitude of the acting dynamic component of the as a multi-layered structure; (lower-illustration) characteristic
total Load. The differential equation is solved and been investigated Values of Static Stiffness Coefficients (spring-constants ρi) and
and these results are presented for the Suspended-Masses of the damping coefficients (ci) of each layer.
vehicle. A sensitivity is also performed and a comparison between
the theoretical results for the vertical accelerations (directly
proportional to the acting loads) due to the Suspended-Masses vs
the measured ones, in a track under operation. The magnitude of
the acting Loads is proportionally analogous to the annual
maintenance cost of a railway track according to the fourth rule of
the following Equation: m
 p1  cos t − 1
  =
 2
p cos t −2
Where pi= average stress on ballast, m=3 or 4.

Keywords
Track, Vehicle, Suspended-Masses, Fourier Analysis, Power
Spectral Density, Spectrum of Defects.

1. GENERAL
The circulation of a railway-vehicle on a track-panel is a random
dynamic phenomenon and, depending on the different frequencies
of the loads it imposes, there is a corresponding response of the Figure 2 Model of a railway-vehicle running on one (of two) rail
track panel and vice-versa the track defects/faults are the random running table, of a track-panel: n is the ordinate of the defect
excitation for the rolling wheels of the vehicle. When a vehicle’s of the rail running table, y is the deflection of the rail.
axle passes from the location of a sleeper/support-point of the rail, The railway vehicles consist of (a) the car-body, (b) the primary
at a track with defects, the track-panel deflects, a vertical and the secondary suspension with the bogie between them and (c)
acceleration (second derivative of the deflection) is generated and the axles with the wheels. In general, the mass of the (c) case under

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

the primary suspension is the Non-Suspended Mass of the vehicle, ρsubstructure ranging (i) 20-60 kN/mm for pebbly subgrade, (ii) 80-
since all the rest are the suspended masses of the vehicle (Fig.2). 100 kN/mm in the case of well compacted subgrade, (iii) 86-171
kN/mm for NBS of the German railways and (iv) 250 kN/mm for
The masses of the railway vehicle located under the primary the case of a ballast-bed of small thickness laid on the concrete base
suspension (axles, wheels and a percentage of the electric motor of a tunnel or on a bridge’s concrete deck.
weight in the case of locomotives) are the Non-Suspended Masses
(NSM) of the Vehicle, that act directly on the railway track without ρpad is calculated through the trial-and-error method from the Load-
any damping at all. Furthermore, a section of the track mass Deflection curve of the pad of the fastening as provided by the
(mTRACK) also participates in the motion of the vehicle’s Non- producer.
Suspended Masses, which also highly aggravates the stressing on
the railway track (and on the vehicle too). The reaction due to the total dynamic component of the load is
given according to the probability of occurrence (99.7% in
The remaining vehicle masses are called Suspended Masses (SM) railways) from:
or Sprung Masses: the car-body, the secondary suspension, the
R = Adynamic  3   2 ( QNSM ) +  2 ( QSM )  (1e)
frame of the bogie, a part of the electric motor’s weight and the dynamic
 
primary suspension.
1
3
 ρ dynamic
The case for the Non-Suspended Masses (NSM) has been presented A dynamic = 4 (1f)
in (Giannakos 2016a). The present paper presents the analysis for 2 2 EJ
the Suspended Masses (SM) of the vehicle: the car-body, the
secondary suspension, the frame of the bogie, a part of the electric ρtotal-(static)
total −dynamic = hTRACK = 2 2  4 E  J  (1g)
motor’s weight and the primary suspension.

2. ACTIONS/REACTIONS ON TRACK The influence of the track defects on the standard deviation
On each support point of the rail (sleeper/tie) a reaction/action is σ2(ΔQSM) of the SM will be investigated in the following
applied due to the distribution of the acting load to the adjacent paragraphs based on the second order differential equation of
sleepers (support-points of the rail) which results from the total motion.
elasticity/total static stiffness coefficient of the track. The sum of
the static, semi-static and dynamic components of the acting load 3. THE 2ND ORDER DIFFERENTIAL
is given from the Eqns (1), as derived from the differential equation EQUATION OF MOTION
of motion (cf. Giannakos 2016a; 2016b): A forced oscillation of the Non-Suspended and the Suspended
Qwheel 4 3  total Rstat 1 4 3  total (1a) Masses of the vehicle is caused by the irregularities of the rail
Rstat =   = A = Astat = 
2 2 EJ Qwheel 2 2 EJ running table (like an input random signal) -as a system’s
excitation- which creates forces influencing the wheel’s trajectory.
2    hCG Qa
Q =  Qwheel  Rsemi − stat = = Astat (1b) In the case of the very short wavelength track defects, the Non-
e2 Qwheel Suspended Masses’ response is predominant since the response of
 ( Qdynamic ) =  2 ( QNSM ) +  2 ( QSM ) (1c) the Suspended Masses is negligible (Giannakos 2016a: for the
modelling of the motion of the Non-Suspended Masses of a railway
Where: Qwheel=the static load of the wheel, E is the modulus of vehicle). On the contrary in the case of long wavelength defects, a
elasticity of the rail (steel), J is the moment of inertia of the rail, forced oscillation is caused where the Non-Suspended Masses’
Rstatic is the reaction/action on each support point of the rail response is negligible since the response of the Suspended Masses
(sleeper); Āstatic is the static reaction coefficient of the sleeper/ is predominant as it will be depicted below.
support point of the rail, α is the cant deficiency; hCG is the height
If the random excitation (track irregularities) is given, it is difficult
of the center of gravity of the railway vehicle from the rail head; e
to derive the response, unless the system is linear and invariable. In
is the track gauge, σ(ΔQNSM) and σ(ΔQSM) the standard deviation
this case the input signal can be defined by its spectral density and
of the dynamic components of the load due to the NSM and the SM,
from this we can calculate the spectral density of the response. The
ℓ the distance among sleepers.
theoretical results confirm and explain the experimental
The total static stiffness coefficient of the track ρtotal is a quasi- verifications performed in the former British railway network.
coefficient of elasticity (spring constant) of the track, given by:

The motion of a railway vehicle on the rail running table/surface or
1 1 1 1 1 1 1
= = + + + + (1d) the motion of a road vehicle on the road, the response of the
total i =1 i rail  pad  sleeper  ballast  substructure structures to earthquakes, etc., is a forced oscillation with a forcing
excitation (force), and damping expressed by a random, non-
where i are the layers that constitute the multilayered structure periodic function. The motion is described by equations and, in
“Track”, and ρtotal the total static stiffness coefficient of track, railway engineering, it is illustrated through the simplified form of
which must be calculated for each case, for rail, pad, sleeper, ballast a spring-mass-damper system as depicted in Fig.3-Upper, with a
and substructure. The values used in the analyses, have been railway vehicle running on a track with longitudinal defects and the
derived from measurements of the former German Railways/DB forces exerted on the vehicle’s “car-body”. In this simplified model,
(Alias 1984; Prud’homme 1969): with the wheel rolling over a surface with defects, but undeflected
itself, the acting forces are:
ρrail ranging 50.000-100.000 kN/mm.
a. the weight of the vehicle m∙g;
ρsleeper ranging 12.000-15.000 kN/mm for concrete sleepers.
b. the dynamic component of the Load P dyn;
ρballast 380-500 kN/mm for ballast older than 2 years old ("fouled").

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

c. the motive force P(t) which is equal to the difference between the This equation describes only the dynamic part of the load. The
tractive force of the locomotive minus the friction and it is positive phenomena of the wheel-rail contact and of the wheel hunting,
in the case of increase of speed (accelerated motion) or negative in particularly the equivalent conicity of the wheel and the forces of
the case of braking (decrease of speed) since it is equal to zero if pseudo-glide, are non-linear. In any case the use of the linear
the motive force is equal to friction; system’s approach is valid for speeds lower than the V critical≈500
km/h. The integration for the non-linear model (wheel-rail contact,
d.the reaction R1 provided by the system “vehicle-track” equal to a wheel-hunting and pseudoglide forces) is performed through the
spring constant or coefficient of elasticity ρi (given in kN/mm) Runge Kutta method (for bibliographic references see Giannakos
multiplied by the subsidence u(x) of the center of gravity of the 2016a). Consequently, for all operational speeds in High Speed
vehicle; Railway Lines up to now the linear model, which we use (in this
e. the reaction R2 provided by the system “vehicle-track” equal to paper also), is quite reliable, since the maximum operational speeds
a damping constant ci multi-plied by the first derivative of the are much less (300 or 300+ km/h). As it is referred above, the
subsidence u(x) of the center of gravity of the vehicle. defects of the rail running table are categorized in short wavelength
defects and long wavelength defects. The analysis and investigation
of the Eqn (2), for consecutive short wavelength defects (e.g. the
rail surfaces corrugation, of wavelength of some centimeters), was
investigated in (Giannakos 2016a). The long wavelength defects
are difficult to be measured, since sometimes, their wavelength
overpass the measurements’ base, which is determined by the
distance between the measuring vehicle’s axles (Fig.3-lower). In
this article the long wavelength defects are modelled, analyzed and
investigated; in this analysis conclusions should be drawn about the
Suspended Masses also, and the region of wavelengths of their
influence. We hypothecate a cosine form defect on the rail running
table of the form (Giannakos 2018):
 V t 
 = a  cos t = a  cos  2 
 
(3)

where: η the ordinate of the defect along the track (abscissa x, cf.
Fig.3-lower), V the speed of the vehicle, t the time and λ the
wavelength of the defect; so in the case of a defect of wavelength
λ:
2 2 2 Vt
T=  t = t  t = (4)
 T 
If we set a new variable z, as:
mSM + mNSM dy dz d 2 y d 2 z
y=z+ g =  2 = 2 (5), where:
hTRACK dt dt dt dt

mNSM + mSM
g (6)
hTRACK

represents the subsidence due to the static loads only, and z is the
random subsidence due to the dynamic loads. Eqn (2) becomes:
Figure 3. (Upper) a spring-mass-damper-system: (left) a vehicle
d2z dz d 2n
on a railway track with longitudinal defects and (right) the mNSM 2 +   + hTRACK  z = − mNSM  2  (7a)
forces exerted on the “car-body”; (lower) measurement of the dt dt dt
ordinate d of a longitudinal vertical defect at the position x on
a chord of 2a length: 2a is much smaller than the defect’s  d 2 z d 2n  dz
mNSM  2 + 2  +   + hTRACK  z = 0 (7b)
wavelength 2a1, consequently the measured value d is smaller  dt dt  dt
than the real defect’s depth (sagitta) d1. Since, in this case, we are examining the dynamic loads only
(derived from the actions of the SM and the NSM), in order to
The equation for the dynamic interaction between the vehicle’s axle
approach their effect, we could narrow the study of equation (7b),
and the track-panel becomes finally (Giannakos 2016a; 2017):
by changing again with a new variable u:
d2y dy d 2 d 2u d 2 n d 2 z
( mNSM + mTRACK )  2
+   + hTRACK  y = −mNSM  2 + ( mNSM + mSM )  g (2) u =n+z = +
dt dt dt
where: mNSM the Non-Suspended (Unsprung) Masses of the dt 2 dt 2 dt 2
vehicle, mTRACK the mass of the track that participates in the motion Equation (7b) becomes:
(theoretical results vs measurements, in Giannakos 2010a; 2012; d 2u dz
2016c), mSM the Suspended (Sprung) Masses of the vehicle, Γ mNSM +   + hTRACK  z = 0  (8a)
dt 2 dt
damping coefficient of the track (for its calculation see Giannakos
d 2u d (u − n )
2010a, 2012), hTRACK=ρdynamic the total dynamic stiffness  mNSM + + hTRACK  ( u − n ) = 0 (8b)
coefficient of the track, η the fault ordinate of the rail running table dt 2 dt
where, u is the trajectory of the wheel over the vertical fault (of
and y the total deflection/subsidence of the track. ordinate n) in the longitudinal profile of the rail. If we apply the
Fourier transform (see Annex) to Eqn (7a):

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

  ( i ) hTRACK The increase of the vertical load on the track due to the NSM,
( i )  Z ( ) +  Z ( ) +  Z ( ) = − ( i )  N ( ) 
2 2

mNSM mNSM according to the principle force = mass x acceleration, is given by:
Z ( ) 2
mNSM  4 d 2u d 2 (n + Z ) (13)
 H ( ) = , H ( ) = Q = mNSM  = mNSM 
2

N ( )
(9)
(m NSM   − hTRACK ) +   
2
2 2 2 2 dt dt 2
If we apply the Fourier transform to Eqn. (13):
H(ω) is a complex transfer function, called frequency response ˆ Q = mNSM  p 2  U ( ) = mNSM  p 2  fˆZ + n ( )  (14a)
function, that makes it possible to pass from the fault n to the
ˆ Q = mNSM  C ( )  N ( ) = mNSM  p  B ( )  N ( )
2
subsidence Z. If we apply the Fourier transform to equation (8a):
( i )  U ( ) +   ( i )  Z ( ) + hTRACK  ( i )  Z ( ) = 0 
2 0
ˆ Q = mNSM   2  n2  B ( )  N ( ) (14b)
U ( ) h2 + 2   2
G ( ) = , G ( )
2
= TRACK2 (10) The excitation (rail irregularities), in reality, is random and neither
Z ( ) mNSM   4 periodic nor analytically defined, like the Eqn. (3). It can be defined
G(ω) is a complex transfer function, the frequency response by its autocorrelation function in space and its spectral density (Alias
function, that makes it possible to pass from Z to Z+n. 1984, p.58; Gent and Janin, p.700; Papoulis 1977). If f(x) is a signal
If we name U the Fourier transform of u, N the Fourier transform with determined total energy and F(ν) its Fourier transform, from
of n, p=2πiν=iω the variable of frequency and Δ̂Q the Fourier Parseval’s modulus theorem (Roddier 1971), the total energy is
transform of ΔQ and apply the Fourier transform at equation (8b): (Wylie and Barrett 1995):
2 2
+ +
d 2u
 f ( x)  F ( )
du dn  dx =  d
mNSM 2 +   + hTRACK  u =   + hTRACK  n  (15a)
dt dt dt − −
where, F(ν)=Α(ν)∙eiφ(ν) and power spectral density:
(m NSM  p 2 +   p + hTRACK )  U = (   p + hTRACK )  N 
S ( ) = F ( ) = A2 ( )
2
(16)
  p + hTRACK
U ( ) =  N ( )
mNSM  p 2 +   p + hTRACK (11a) Wylie and Barrett (1995) solve Eqn (15a) as:
2
+ +
1
B ( )
 f ( t )  dt =  F ( )  d 
2
where: (15b)
2
 2   2 + hTRACK
2 − −
B ( ) =
2
(11b)
(m
  2 − h ) + 2   2
The square of the modulus F(ω) is called the energy spectrum of the
2
NSM signal because F2(ω)∙Δ(ω) represents the amount of energy in any
B(ω) is a complex transfer function, the frequency response ΔΩ segment of the frequency spectrum, and the integral of F2(ω)
function, that makes it possible to pass from the fault n to the over (-∞, +∞) gives the total energy of the signal. An input signal
u=n+Z. Practically it is verified also by the following equation: -like the running rail table- creates through the vehicle an output
2
hTRACK + 2   2
B ( ) = H ( )  G ( ) = signal: the wheel trajectory. The output spectral density and the
2 2 2
[11c]
( mNSM   − hTRACK ) + 2   2
2
2 input spectral density of the excitation are related through equation
(Clough and Penzien 1993; Giannakos 2004):
passing from n to Z through H(ω) and afterwards from Z to n+Z
SOUTPUT ( ) = H ( i )  S INPUT ( )
2
through G(ω). This is a formula that characterizes the transfer (16a)
function between the wheel trajectory and the fault in the
longitudinal level and enables, thereafter, the calculation of the In order to relate the temporal spectrum with the spectrum in space
transfer function between the dynamic load and the track defect we use the following equation:
(fault). 2 Vt 2
 t =  = V   =  V (16b)
 
The transfer function B(ω) allows us to calculate the effect of a where λ is the wavelength of the defect; the circular frequency in
spectrum of short wavelength sinusoidal faults, like the undulatory space Ω is the wave number k of the equation of oscillation, and
wear. If we replace ω/ωn=ρ, where ωn= the circular eigenfrequency (Roddier 1971):
 
(or natural cyclic frequency) of the oscillation, and: 1  
h 2 V    S (  )  d  =  s ( )  d    f ( ax )  =  fˆ   
a a
n2 = TRACK ,  = , 2 n = , = (11d) 0 0
 n   1
mNSM mNSM
 S ( ) = S   =  S (  ) (17a)
where ζ=damping coefficient. From Eqn (11b): V  V
where is the symbol for the application of the Fourier transform
1 + 4 2   2 of f and f̂ the function after the application of the Fourier transform
B ( ) = Bn (  ) =
2 2
(11e)
(1 −  2 ) + 4 2   2
2
(see Annex). This is a property of the Fourier transform.
The transfer function C(ω) of the second derivative of (Z+n) in Eqn (17a) applied in the case that the power spectrum of the vertical
relation to time: [d2(Z+n)/dt2], that is the acceleration γ, will be equal defects along the track (for the NSM) in the space domain is S(Ω)
to ω∙Β(ω): then the power spectrum of the excitation of the wheel in the time
p 2  U (  ) = p 2 B ( )  p 2  N ( )  domain will result after a replacement of Ω by the ω/V.
 U ( ) = p 2 B ( )  N ( ) (12a) 1  
s ( ) = S   (17b)
C ( )
V V 
The Variance or mean square value σ (x) of the function is given
2
2
C ( ) =  2  B ( ) = n2 
 B ( )  by Papoulis (1991):
n2 1
+

 C ( ) = n    B ( ) =  2  B ( )
2 2 (12b)  2 ( x) =   S ( )  d  = x 2 (18)
2 −

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

where σ(x) the standard deviation of the function. From Eqn (18) During the research program performed by the Greek State
we derive: Railways/OSE and the French State Railways/SNCF for the
1
+
1
+ cracking of sleepers/ties, the measured values for the line “Les
 2 ( n) =  Sn ( )  d  ,  ( z ) =
2
 S ( )  d  , Aubrais – Vierzon” in France were used; the parameters values,
  z
0 0
taken into account, are: B=0.36, A=2.1∙10-6 and S(Ω) is calculated
+
1 in m3 and σ(z)=1.57 mm. The eigenfrequency of the NSM of the
 2 ( Q ) =  S ( )  d 
Q (19) vehicles is approximately 30–40 Hz and even for speeds of 300
 0
where n is the random variable of the defect (input), z the subsidence km/h there are wavelengths less than 3 m (Prud’Homme 1970).
of the wheel (output) and ΔQ the dynamic component of the Load From equations (17) and (24):
1 1 A V 3 A V 2
Sn ( ) =  S (  ) = 
that is added to the Static Load of the wheel due to the NSM (output = 3 3 (25)
also). V V  3
n  
From these equations and the analytic form of the spectrum of the 4. LONG WAVELENGTH DEFECTS
defects/faults, we can calculate the mean square value of the An analysis of the behavior of the NSM was presented in
dynamic component of the Load due to the NSM that is added to Giannakos (2016a; 2013), where diagrams are cited for the relation
the relevant dynamic component of the SM and the total dynamic between the wavelength of the defect and the subsidence z(x) and
component of the load is added to the Static Load of the wheel. the vertical acceleration z´´(x) of the NSM (consequently, the
From the power spectral density and the variance functions and their dynamic component of the load since force=mass x acceleration).
definitions: It is valid:
 
Q = mNSM     2 ( Q ) = mNSM   2 ( )   ( mNSM + mTRACK ) z  ( t )   2  1 
2 
 cos ( 2 ) − ( n )  cos ( 2n )  
1
(20a)    = −   
2
(26)
 m    n  Tn  g  2
 
2  1 − ( n )  steady − state 
 2 ( Q )     
NSM
transient − part

  2 ( ) = (20b)
mNSM The result of this equation (26) is a dimensionless number given as
S Q ( ) = S n ( )  B ( ) (20c)
2
a percentage [%] of the gravitational acceleration g; where:
and using the Eqns. (20) and (11-12b): n=ωn/ω1=Τ1/Τn, Tn=2π/ωn=the period of the free oscillation of the
+
wheel circulating on the rail, T1=2π/ω1=the necessary time for the
1 1
 2 ( ) =   2 ( Q ) =  B ( )  S n ( )  d   wheel to run over a defect of wavelength λ, ω1=λ/V, V=speed, μ=
2

mNSM m 2
NSM  0 a percentage of the wavelength λ [0, 0.1, 0.2,...., 0.8, 0.9, 1], α=the
2 + amplitude of the defects’ depth [derived from track recording-cars].
mNSM
  2 ( ) =    4  n4  B ( )  Sn ( )  d 
2
In the case of the SM, the forces resulting from the excitation, due
mNSM   0
2
to the short wavelength track defects, are not large and have small
+ effect on the rolling of the wheel, as it has been presented in
1 1 + 4 2   2
  2 ( ) =  
4
 n4   S n ( )  d  (21) Giannakos 2016a.
 0 (1 −  ) 2 2
+ 4 2   2
From the above equations and the analytical form of the spectrum of
the longitudinal defects/faults of the track we could effectively
calculate the variance (mean square value) of the dynamic
component of the Loads on the track panel due to the NSM. All the
results of measurements on track in the French railways network
show that the spectrum of defects in the longitudinal level has the
form (SNCF 1981; Prud’Homme 1970):
A
Sn (  ) = (22),
( B + )
3

implying that the mean square value or variance of the defects is


given by the following Eqn (23):
+
1
+
A A
+
1 A  1 
 2 ( z) =   ( B + )  d =   ( x)  dx = −   =
  2  ( B +  )  0 Figure 4. Vertical acceleration of NSM vs defects’ wavelength.
3 3 2
0 0

+
A  1  A  1  1 A
=−  2 =− 0 − B 2    ( z ) = 2  B 2
2
2 
(23) Fig.4 depicts that: for wavelengths inferring cyclic-frequencies
2  B + 2 B +   0 2  
If we examine only the much more severe case, for the case of the higher than two times the eigenfrequency of the wheel, the vertical
NSM, of the defects of short wavelength, consequently large Ω acceleration of the NSM (consequently the correspondent dynamic
–like the undulatory wear– then we can omit the term B, and using component of the load) becomes almost negligible, compared to the
Eqn. (16b): results presented in Giannakos 2016a, for smaller values of ω n/ω1
A A A V 3 (0.5 till 2). In the cases of the long wavelength defects, which are
Sn (  ) = = = (24) of decisive importance in High-Speed Lines, the influence of the
 3
1
3 3
V3 SM must be examined. If we examine the trajectory of one wheel,
The term B characterizes the defects with long wavelengths, for then the conclusion that will be derived can be used for more
which the maintenance of track is effective, and when we examine complicated cases of rolling of vehicles, motion of car-bodies etc.
this kind of defects term B should be taken into account. SM should Furthermore, we consider the simplified model of Fig.3-Upper with
be examined for long wavelength defects. one-floor mass-spring-damper system rolling on a rail’s running
surface.

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

In order to calculate the power spectrum density of the excitation which converges for ω infinite and can be calculated in a computer.
sE(ω) from the excitation spectrum of the wheel sυ(ω), we apply the Consequently, the variance of the part of the dynamic component
Eqn. 16a with the Eqns. 17, Eqn. 22 (the parameter B is not omitted of the load due to the SM of the vehicle is given by (see Eqn. 20b):
for the case of SM) and Eqns. 10 (in order to pass from the defect
n to n+Z): 2
1 + 4 2 
n2 AV 2
sE ( ) = B ( )  s ( ) = sE ( ) =
2
 (27a)
( BV +  )
2 3
   2
2 
2

1 − 2  + 4  2
 n  n

where: sE(ω) is the power spectrum density of the excitation, ωn is


always the eigenfrequency of the NSM, sυ(ω) the spectrum of the
excitation of the wheel due to the track defects/faults, │B(ω)│the
modulus of the transfer function of the motion of the wheel and ζ
the damping coefficient of the track:
1 
 = 
2 mNSM  hTRACK
The sυ(ω) can be calculated from Eqn. 22. We have to take into
account Eqn. (17a) also:
1 1 A 1 A AV 2 (27b)
s ( ) =  S () =  =  =
V V ( B + ) V 
3
  ( BV +  )
3 3

B+ 
 V
The value of sυ(ω) derived from Eqn. 27b is used in Eqn. 27a.
From the Eqns. 11 and the Eqns. 12, 13 and 14, with the analysis Figure 5. Variance of the vertical acceleration in relation to the
cited above, we keep that, C(ω) is the transfer function of the second cyclic frequency of the defects of the rail running table (input).
derivative of (Z+n) in relation to time: [d2(Z+n)/dt2], that is the
acceleration γ and it is equal to ω∙B(ω). In the case of the SM,
│C(ω)│is the modulus of the transfer function of the accelerations QSM = mSM     2 ( QSM ) = mSM
2
  2 ( ) (30)
of the car-body. Consequently, for the spectrum of the accelerations
It is obvious that the variance sγ(ω) and the standard deviation σ(γ)
of the car-body we will use Eqn. 21, 20, 15a substituting the
are functions of the inverse of the cyclic eigenfrequency ωn of the
parameters ζ and ωn of the track with the relevant parameters ζ′
track, which (Eqns. 17, 27a) is a function of the static (ρTRACK)
(damping coefficient) and ω′n (eigenfrequency) of the car-body. For
and/or dynamic (hTRACK) stiffness coefficients of the track.
the railway passenger-vehicles the eigenfrequencies ω′n of the car-
body are in the area of 1 Hz, since with the development of high-
speeds it could arrive 10 Hz. For the damping coefficient of the car- 5. MEASUREMENTS vs THEORY
body of the railway vehicles two characteristic values of ζ′ could be In real conditions, according to measurements of the French State
used with reliability: 0.15 and 0.20 (see relevantly Prud’Homme Railways (SNCF), the standard deviation σ(γ) of the vertical
1970; Alias 1984). From Eqn. 26a (Giannakos 2016b) we derive the accelerations (as a percentage of the gravitational acceleration g)
below Eqn (27c): due to the SM of the vehicles, in relation to the running speed is
AV 2 1 + 4 2   2 1 + 4 2   2 depicted in Fig.5, as presented by Alias (1984; 1987) and
s ( ) = C ( )  s E ( ) =  4 
2
 
( BV +  ) (1 −  ) + 4   (1 −  2 ) + 4 2   2 Prud’Homme (1970). The measurements were performed in tracks
3 2 2 2 2 2

There are two (2) maximums in the curve of sγ(ω) as depicted in under operation/circulation in the French network (SNCF) and
Fig.6: The first maximum corresponds –in practice– to the obviously, the measured values of σ(γ) have been influenced by the
eigenfrequency of the car-body of the vehicle. It is approximately variation of the real values of ρsubgrade and ρtrack, as it existed along
(for ω=ω′n→β′=1): the measured tracks (consequently the values of ρ were not
2
constant), since the variability of the track stiffness e.g. due to
  imperfect sleeper support and inhomogeneities of the track structure
1 + 4 2   n 
s ( )max −1 = n4 
AV 2  1 
  2 + 1   n  is an inherent property. Fig.5 is the final product of the
(28a)
( BV + n ) 
3 2 2
 4       
2
measurements performed along the tracks of the French network,
1 −  n   + 4 2   n 
  n    n 
and it is given in the publications above, as of general validity. In
 
Fig.5, six curves are presented for eigenfrequencies of the car-body
the second maximum corresponds –in practice– to the
1 to 6 Hz. More analytically, the coupled system car-body-bogie-
eigenfrequency of the system “vehicle-track”, that is the
axles (a two-floor system) presents an eigen-frequency, that of the
eigenfrequency of the system of the NSM and the track, which is
axle on the track, approximately 30-40 Hz, in an attenuated form,
approximately 30-40 Hz (see relevantly Prud’Homme 1970). It is
corresponding to the track defects which will become much more
approximately (for ω=ωn→β=1):
important when the speed in consideration will provoke a frequency
 1   2 very close to the coupled frequency of the car-body with the axles.
s ( )max −2 = AV 2   2 + 1  4 2 n (28b)
 4  n
This coupled low frequency of the car-body-bogie-axles, which
The variance of the accelerations of the car-body of the railway mainly affects the car-body and it is specifically interesting to us in
vehicles is given by (Eqns. 18, 19): this analysis, is approximately 1 Hz for the passenger wagons and

1
 (  ) =   s ( )  d 
2
(29) higher for the freight wagons (see relevantly Alias 1987, p. 47). This
 0 implies that the curve for 1Hz eigenfrequency represents the

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

passenger wagons and the higher frequencies the freight wagons, In more details, NL, the average of the brutal signal on a basis of
running at lower speeds. According to the analysis cited above approximately 300 m for the vertical and horizontal defects of the
–about the coupled system car-body-bogie-axles– it is clear that, two rails, is the convolution:
+  x − x0 
finally, in this coupled motion (of 1 Hz frequency) all the parts of 1  
N L ( x0 ) =  l ( x )e
300 
dx (31)
the vehicle participate, so we can approach the equations using the 300 −
total Qwheel instead of mSM. Consequently, an approximation could
be used for the calculation of the variance of this part of the dynamic where ηl(x) is the value of the primary signal; in practice a weighted
component of the load (Giannakos 2004; SNCF 1981): average index which “crashes” less the isolated defects than one
classic average and simulates roughly the “memory” of the vehicle
(see relevantly Janin 1982). In Fig.7 the curves of the
eigenfrequencies of the vehicles as measured by the SNCF are
depicted and, also, the Eqn (30) for four values of NL: 0.25, 0.5, 1
and 1.5. The approximation is close enough, since the dynamic
component of the load due to the SM represents a small percentage
(less than 10% usually) but a further theoretical research and
sensitivity analysis should be performed in the future.

Figure 6. Standard deviation σ(γ) of the measured vertical It should be underlined that -after measurements on track (in situ)
acceleration due to the SM of railway vehicles in relation to the and in laboratory- Eisenmann and Mattner (1984) refer: the
running speed as published [Alias (1984; 1987) and theoretical deflection of the track is the same as the measured ones
Prud’Homme (1970)]. The figure has been designed by the on track. Professor Eisenmann (1980, 1988; cf. Giannakos 2011)
author of the present paper. assures that the theoretical calculations for the dimensioning of the
track correspond to the average of the measured values on track.
These ascertainments imply that the theoretical approach is reliable
and dependent on the probability of occurrence, which in railway
engineering is 99.7%.
The analysis in the present paper leads to the conclusion that Eqn.
(31) gives results very close to the real values measured in the
system “vehicle-track”. Anyway, in the future a closer approach
should be applied for each vehicle’s eigenfrequency-curve, since
these eigenfrequencies characterize different types of railway
vehicles: 1Hz for passenger vehicles and the rest 2-6Hz for freight
vehicles. Furthermore, the measurements of long wavelength
defects -which are predominantly influential for the High-Speed
Lines- are referred to wavelengths over than 33m, that is longer
than the lengths of the measurement-bases of the track-defects’
recording cars (Fig.3-lower) and the measurements’ results depict
smaller sagitta than the real ones of the defects.

6. CONCLUSIONS
For defects of very long wavelength, of period T>2 and frequencies
1 to 3 Hz, the forced oscillations -due to these defects- of the
Suspended Masses become predominant since, at that case, the
Figure 7. Standard deviation σ(γ) of the measured vertical eigen-frequency of the SM is approached by the forced vibration.
acceleration, due to the SM of the railway vehicles, in relation
to the running speed as published [Alias (1984; 1987) and The standard deviation of the dynamic component of the load due
Prud’Homme (1970)] and Eqn. (30) for NL=0.25, 0.5, 1.0, 1.5. to the Suspended Masses as it is calculated by the theoretical
The figure has been designed by the author. calculations (Eqn 31) approaches sufficiently the loads which are
derived by the multiplication of the measured value of the vertical
acceleration by the value of the Suspended Masses of the vehicle.
V − 40
 ( QSM ) =  N L  Qwheel (31) The closest approach of the magnitude of the acting loads is very
1000
where: Qwheel is the static wheel load, V is the operational speed, useful, since the acting loads are proportionally analogous to the
and the coefficient NL is the mean standard deviation of the annual maintenance cost of a railway track according to the fourth
longitudinal level condition of the track, on a 300 m length rule of the following Equation (Giannakos 2011):
m
approximately, for both rails is the mean standard deviation of the  p1  cos t − 1
  = (32)
longitudinal level condition of the track, on a 300 m length  p2  cos t − 2
fluctuating between 0.7–1.5 mm or more (see SNCF 1981; Janin where pi= average stress on ballast, m=3 or 4.
1982, p.335–336); for the Greek network NL is estimated to This implies that 10% increase of the stress on the ballast-bed of a
fluctuate –mainly– between 1 and 1,5 (Giannakos 2004). railway track will infer 1.14=1.51 times higher annual maintenance
cost. The accurate approach for the calculation of the contribution
of the SM in the total dynamic component of the load leads to more

DOI: http://dx.doi.org/10.17501........................................
DISP '19, Oxford, United Kingdom
ISBN: 978-1-912532-09-4

accurate estimation of their influence in the annual maintenance [17] Giannakos, K. 2018. Variance of the Vertical Accelerations
cost of the railway track (and of the vehicles: action-reaction). due to the Suspended Masses in the System ‘Railway
Vehicle-Track’: Simulation Technique, Differential
7. REFERENCES Equation, Fourier Transform and Approximants,
[1] Alias, J. 1984. La Voie Ferree, IIeme edition, Eyrolles, International Journal of Theoretical and Applied Mechanics,
Paris. vol.3, 10-16, http://www.iaras.org/iaras/journals/ijtam.
[2] Alias, J. 1987. Le Rail, Eyrolles, Paris. [18] Janin, G. 1982. La Maintenance de la Geometrie de la Voie,
[3] Clough, R.W., and Penzien, J. 1993. Dynamics of Structures, Revue Generale des Chemins de Fer (RGCF), Juin, 331–346.
second edition, McGraw-Hill Intl., Singapore. [19] Papoulis, A. 1977. Signal Analysis, McGraw-Hill Inc., USA.
[4] Eisenmann, J. 1980. Verjormungsverhalten der Schiene. [20] Papoulis, A. 1987. Reissued. The Fourier Integral and its
Auswirkungen auf die Oberbaubeanspruchung, Lagestabilitat Applications, McGraw-Hill Classic Textbook Reissue, 1962.
und Storungsfunktion, ZEV- Glasers Annalen, Berlin, 37–41. [21] Papoulis, A. 1991. Probability, Random Variables and
[5] Eisenmann, J. 1988. Schotteroberbau: Moglichkeiten und Stochastic Processes, third edition, McGraw-Hill Intl.,
Perspektiven fur die Moderne Bahn, Giesel-Verlag, Singapore.
Isernhagen, Germany. [22] Prud’homme, A. 1969. Reprint. Sollicitations Statiques et
Dynamiques de la Voie, SNCF/Direction des Installations
[6] Eisenmann, J., and Mattner, L. 1984. Auswirkung der Fixes, Paris, 1966.
Oberbaukonstruction auf die Schotter und [23] Prud’Homme, A. 1970. La Voie, Revue Generale des
Untergrundbeanspruchung, Eisenbahningenieur, Vol. 35, Chemins de Fer (RGCF), Janvier, extrait de RGCF.
No 3, 99–107. [24] Roddier, F. 1971. Distributions et Transformation de Fourier,
[7] Gent, I., and Janin, G. La Qualite de la Voie Ferree, SNCF’s Edisciense, Paris.
reprint. [25] SNCF. 1981. Mecanique de la voie, Paris.
[8] Giannakos, K. 2004. Actions on the Railway Track, Papazissis [26] Wylie, C.R., and Barrett, L.C. 1995. Advanced Engineering
publ., www.papazisi.gr, Athens. Mathematics, sixth edition, McGraw-Hill, Inc., USA.
[9] Giannakos, K. 2010a. Theoretical calculation of the track-
mass in the motion of unsprung masses in relation to track
dynamic stiffness and damping, IJPE-Special Rail Issue:
“High-Speed Railway Infrastructure: Recent Developments ANNEX – Fourier Transform
and Performance”, Vol. 11, number 4, 319-330. Since this article is mainly addressed to Civil Engineers,
specialized in Railway Engineering, some general reminders about
[10] Giannakos, K. 2011. Track Maintenance of Heavy Haul the Fourier Transform are cited in this Annex (for more details and
Railways with Resilient Versus Stiff Fastenings, Transport properties of the Fourier Transform cf. Papoulis 1987; 1977, 56-79;
Research Record (TRR), 155-162. Wylie and Barret 1995, 543-597):
[11] Giannakos, K. 2012. Influence of the Track's Damping on the The Fourier Transform F(ω) of a function f(t) is given by:
Track Mass Participating in the Motion of the Non-Suspended +

Masses of Railway Vehicles-Theoretical Calculation and F(  ) = ˆf =  f ( t )  e− jt dt (33)


−
Comparison to Measurements, volume published in honor of Under the assumption that the above integral exists as a Cauchy
professor G. Giannopoulos, Aristotle University of principal value for every real ω. Where:
Thessaloniki.
[12] Giannakos, K. 2013. Track Defects and the Dynamic Loads
e jt = cos t + j  sin t (34)
due to Non-Suspended Masses of Railway Vehicles, NAUN The Cauchy principal value of an integral from -∞ to +∞ is the
International Journal of Mechanics, Vol.7, Issue 3, 180-191. limit:
[13] Giannakos, K. 2016a. Modeling the Influence of Short + _+ A

Wavelength Defects in a Railway Track on the Dynamic  x( t )dt = lim 


−
A→
−A
x( t )dt (35)
Behavior of the Non-Suspended Masses, jmssp, Elsevier, 68- The inversion formula or the inverse Fourier Transform is given by:
+
83.
 F(  )  e
jt
f (t ) = d (36).
[14] Giannakos, K. 2016b. Deflection of a railway reinforced −
concrete slab track: Comparing the theoretical results with
experimental measurements, Engineering Structures Journal,
Elsevier, (122), 296-309.
[15] Giannakos, K. 2016c. Deflection of a Railway Reinforced
Concrete Slab Track: Comparing the Theoretical Results
with Experimental Measurements, Engineering Structures
Journal, Elsevier, 122, 296-309.
[16] Giannakos, K. 2017. Track Defects and the Dynamic Loads
due to Suspended (Sprung) Masses of Railway Vehicles, in
Loizos A, Al-Qadi I., Scarpas T., (eds), Intl. Conference
“Bearing Capacity of Roads, Rails, Airfields”, Athens,
proceedings, Taylor & Francis group, 1911-1919.

DOI: http://dx.doi.org/10.17501........................................