Modeling and experimental investigation on the vibration of

main drive chain in escalator

Xiaolong Ma
[email protected]
Xi Shi
[email protected]
Jiaohua Zhang
[email protected]
School of Mechanical Engineering, Shanghai Jiao Tong University,
800 Dong-chuan Road, Shanghai 200240, P. R. China

Abstract
In this paper, a coupled nonlinear string model was built to analyze the dynamic
behavior of the main drive chain in escalator. Following, Wavelet-Galerk in
method was applied to solve the coupled non linear equation. Meanwhile, a testing
rig was designed to experimentally investigate the vibration of drive chain in
escalator. The transverse vibration of drive chain was captured with high-speed
camera under different traction conditions. And then it was quantitative ly
obtained following an images processing which was put forward according to
binary and morphological algorithm. Finally, compared with experimental results,
the validity of the string model and the effects of pre-tension were discussed.

Keywords: Escalator; Drive chain; String vibrations

I-INCE Classification of Subject Number: 14

1. Introduction
Chain drives are widely used to transmit power between rotational machine elements,
where positive driving, high efficiency, and low maintenance cost are
required[1].Compared to belt drives, chain drives are characterized by the discrete
nature of the chain links and sprocket teeth, which lead to noise and vibration[2].
Therefore, vibration of chain is an important problem that needs to be resolved in design
and the operation of the chain drives[3]. There are two factors resulting in vibration of
the chain, which are internal and external stimuli. Periodic torsional loading and
imbalance in the drive are external sources. Polygonal action and
roller-tooth impacts are external sources[4].Because the chain, lying on the sprockets,
forms a polygon rather than a circle, the ends of the chain experience periodic
fluctuation of velocities, which is known as polygonal action [5].
Many experiments have been carried out to investigate the dynamic behavior of
roller chain drives. In 1995, James [6] used a strain gage mounted on a link side plate
to determine chain tension during normal operation over a wide range of linear chain
speeds and preloads. What’s more, Roberto and Yan [7] built a novel mechatronic test
rig to study the chain lubrication and its effect on the temperature, efficiency and
vibrations of the motorbike chain transmission. Lenkov et al. [8] have designed the
experimental installation of a chain drive for investigation of the influence of chain
tension on the amplitude of forced oscillations. Also there are many experiments to
measure the impact force of the chain rollers on the sprocket teeth [9-11].
Under the assumption that the chain span motion decouples from the dynamics
of sprockets and attached machinery, the studies of roller chain drive dynamics are
found to border the analysis of axially moving strings [12]. Early in 1957,
Mahalingam [13] treated the chain as a uniform string to determine the natural
frequencies. In 1987, Ariartnam and Asokanthan [14] studied the periodic fluctuation
of power transmitting chains brought about by factors such as polygonal action and
eccentricity of sprockets by treating the chain as a traveling uniform heavy string.
Nowadays, Destyl et al. [15] derived and studied a novel model of the coupled

-symmetric discrete nonlinear Schrödinger equations which described

parametrically driven chains of the coupled pendula pairs connected to the nearest
neighbors in the longitudinal and transverse directions. In studies of string and roller
chain drive dynamics, it is often assumed that polygonal action leads to a parametric
excitation described by time harmonic variation of span tension or velocity [16].
Attempting to solve the coupled vibration problems of axially moving string,
many researchers have investigated the applications of various mathematical methods
on it. Generally, employing the Galerkin discretization, the equation of motion is
reduced to a set of nonlinear ordinary differential equations (ODE) with coupled
terms. [17-19]. For example, Chen [20] applied the method of multiple scales based
on the Galerkin discretization to solve the governing equation of transverse nonlinear
vibration in the parametric resonances. Nowadays, approximation techniques play an
indispensable role in complementing exact solutions. Wavelets technique represents a
newly developed powerful mathematical tool, which has been broadly applied to
solve differential equations. Pernot [21] introduced a wavelet Galerkin method in
order to obtain transient and periodic solutions of multi-degree-of-freedom dynamical
systems with time-periodic coefficients. Liu et al. [22] presented an efficient
wavelet-based algorithm for solving a class of fractional vibration, diffusion and wave
equations with strong nonlinearities. Likewise, a new wavelet approximation scheme
of bounded functions based on techniques of boundary extension and Coiflet-type
wavelet expansion has been developed by Wang [23].
In this study, considering transverse and longitudinal vibration, a coupled
nonlinear string model was built to analyze the dynamic behavior of the main drive
chain in escalator. The coupled dynamics vibration equation was derived by The
Hamilton’s principle and solved with the Wavelet-Galerkin method. In next section, a
new experimental method for detecting dynamic vibration of main drive chain in
escalator by utilizing image processing technology was developed. Finally, compared
with experimental results, the validity of the string model, the effects of pre-tension
and excitation of chain were discussed, which in principle are conductive to the study
and engineering design of the main drive chain system in escalator.

2. Equation of motion for drive chain

It is proposed here the chain can be modeled as a uniform traveling string for the
purpose of this analysis under the assumption that the chain span motion decouples
from the dynamics of sprockets and attached machinery.
2.1 Governing equation of drive chain
The schematic of main drive chain system of escalator is shown in Fig.1. The
main drive chain transmits the power at speed v between driving and driven sprocket.

Fig. 1 schematic of main drive chain system of escalator

It is proposed here the chain can be modeled as an axially moving string with
both ends hinged, which is shown in Fig.2. ρ c T E A are, respectively, density per unit
length, damping coefficient, tension, Young’s modulus and the cross-section area.
w(x,t) and u(x,t) are longitudinal displacement and transverse displacement at time t
and axial coordinate x related to coordinates translating at speed v. h(x,t) is the
excitation. Besides, the origin of the coordinate is set at one end of the string and the
instantaneous length of the string is l.

Fig. 2 A theoretical model of the string with the longitudinal and transverse vibration
The kinetic energy of the axially moving string is given by

1 l
K    v vdx (1)
2 0
where

 w w  u u
v  v   v i  (  v ) j
 t x  t x

The potential energy is obtained as

l 1 
U    EA 2 +T  dx (2)

0 2

The displacement strain relation ε is expressed as
w 1 u 2
  ( ) (3)
x 2 x
Using Hamilton’s principle, the governing equations of motion for the axially
moving string system with longitudinal and transverse motions are derived as
2 w 2 w 2  w
2
2 w  1 u
( +2V  V )  ( EA  T )  EA [ ( ) 2 ]  0
t 2
xt x 2
x 2
x 2 x
(4)
 2u  2u  2u  w u 1  u   2u
3

(  2V  V 2 2 )  EA [    ]  T 2  h ( x, t )
t 2
xt x x x x 2  x  x
The initial conditions and boundary conditions are given
w(0,t)=w(l,t)=0, w(x,0) =0
u(0,t)=u(l,t)=0, uxx(0,t)=uxx(l,t)=0, u(x,0) =0 (5)
 Equation (4) is a strong nonlinear differential equation. Following, we should to
truncate the infinite dimensional partial differential equation (PDE) into a
finite-dimensional ordinary differential equation.
2.2 Discretization of the governing equation in Wavelet Galerkin method
Firstly, some dimensionless variables are introduced as follows:
x u w
x ，u  ，w  (6)
l l l
𝜕𝑢 𝜕𝑤
We introduce 𝑢1 = 𝜕𝑥 and 𝑤1 = , and according to the wavelet theory [22],
𝜕𝑥

the dimensionless functions can be approximated through the scaling function series
as, respectively,
2 j 1 2 j 1
u  u ( x , t )
k 1
k j , k ( x) ， w  w( x , t )
k 1
k j ,k ( x) ，

2j 2j
u1   u1( kx , t) j ,k x ) w1   w1 ( xk , t ) j ,k ( x ) ，
(， (7)
k 0 k 0

To transform Equation (4) into a matrix equation,, by applying the wavelet Galekin
method, Equation (4) is expressed as:
2j
 2 w( xk , t ) 2 1
w( xk , t ) T  V 2
j
2j
l  alk  V  blk   w( x , t )c
t t
2 k lk
k 1 k 0 l k 1
j
12
  f [u1 ( xk , t ), w1 ( xk , t )]mlk  0
l k 1
2 j 1
 2 u ( xk , t ) 2 1
u ( xk , t ) ( V 2  T ) 2 1
j j

l  d  V  e   u ( xk , t ) f lk (8)
t 2 t
lk lk
k 1 k 0 l k 1
j j
12 2
 
l k 0
g[u1 ( xk , t ), w1 ( xk , t )] plk   h( x, t )qlk  0
k 0

where
1
alk    j ,k ( x ) j ,l ( x )dx
1 1
blk    ' j , k ( x) j ,l ( x)dx clk   j , k ( x) j ,l ( x)dx
0 0 0

1
d lk    j ,k ( x ) j ,l ( x )dx
1 1
mlk    ' j , k ( x) j ,l ( x)dx elk    ' j , k ( x) j ,l ( x)dx
0 0 0

1 1 1
flk    '' j , k ( x) j ,l ( x)dx plk    j , k ( x) j ,l ( x)dx qlk    j , k ( x) j ,l ( x)dx
0 0 0

1
f [u1 ( xk , t ), w1 ( xk , t )]  EAu12 ( xk , t )
2
1
g[u1 ( xk , t ), w1 ( xk , t )]  EA[ w1 ( xk , t ) u1 ( xk , t )  u13 ( xk , t )]
2
As above, by applying the Wavelet-Galerkin method, without inversion of matrix
and special solution technique to deal with the nonlinear spatial and temporal
operators, we can obtain the discretized nonlinear algebraic equations:
dU
 Ut
dt
dW
 Wt
dt
dUt V (T  V 2 ) 1 1 1
(9)
  D1EUt  D FU  2 D1PG t  D1QH
dt l l l l
dWt V T  V 2 1 1
  A 1BWt  A CW  2 A 1MFt
dt l l l

Where the matrices and vectors are A={alk}, B={bk}, C={clk}, D={dlk}, E={elk},
M={mlk}, F={flk},P={plk}, Q={qlk}, U=[u(x1,t), u(x2,t), …, u(x2m-1,t),]T, H=[h(x0,t),
h(x1,t), …, h(x2m-1,t), h(x2m,t)]T.
So far, the discretization in space of the governing equation in Equation (4) has
been fulfilled. It can be seen the wavelet-Galerkin procedure uncoupled the
complicated spatial and temporal dependences in the original PDEs, results in a set of
optimally minimized ODEs. Then based on the parameters given by Table 1,
numerical solutions can be computed by applying the wavelet Galekin method.
Table 1. The values of the parameters
l ρ V E A T
0.66m 5.4kg/m 0.449m/s 2.3Gpa 2 × 10−4 𝑚2 -
3. Experimental design and configuration
The experiment was carried out by the camera measurement method for the
transverse vibration of chain. Figure 3 depicts the chain vibration measurement
system, which consists of a LED light source, a high speed industrial camera, the
marked object and a computer for image processing.

1.Marks 2. LED 3. High speed camera 4. PC control and record

Fig.3 Structure of chain vibration measuring equipment
In the experiment, a series of marks were purposely made at each junction center
of chain links. During the traction, the positions of those marks were recorded with
camera and the raw images were processed first with five detailed processes: image
reading, threshold calculation and binarization, dilation, erosion and opening
operation. Vibrations of chain could be calculated after image processing. Besides, the
experiment for camera calibration was performed and the results showed that the
measurement relative error of displacement was roughly in the range of -2.7%~2.8%.

4. Results and discussion

4.1Analysis of the validity of the string model
There are two traction conditions for escalator: running down and running up.
Notably, it will need larger traction force for running up than running down under the
same load. In this section, experiment and numerical simulation have been done to
investigate the vibration of descending and ascending escalator chain under empty
load.
Firstly, the dynamic analysis of ascending escalator chain under empty load is
presented. Figure 4 depicts the comparison of transverse vibrations for one-third
position of the ascending chain between experiment and numerical simulation.
Notably, the tension of the chain is about 3300N. It can be seen that both of behaviors
including amplitude and waveform for experiment and numerical simulation are
comparable.

Fig.4 Comparison of transverse vibrations for one-third position of the ascending chain between
experiment and numerical simulation
Meanwhile, amplitude-frequency characteristics for one-third position of the
chain between numerical and experimental data have been shown in Figure 5. It can
be seen the natural frequency is identified as 18.8Hz from experiment while it is
18.52Hz from numerical simulation. Obviously, the numerical result is in good line
with the experimental result and the relative deviation is about 1.5%. This confirms
that the proposed string model performs well in simulating the vibration of ascending
escalator under empty load. It should be noted that some inconsistencies between
experimental and numerical results also exist in other frequencies, which may be on
account of some practical factors, such as environment noise and inexact excitation in
simulation.

Fig.5 Comparison of frequency characteristics for one-third position of the ascending chain
between experiment and numerical simulation
Following, the dynamic analysis of descending escalator chain under empty load
is provided. Figure 6 shows the comparison for one-third position of the descending
chain between experiment and numerical simulation. Notably, the tension of the chain
is about 10N. Obviously, dynamic behaviors including amplitude, waveform and
frequency for experiment and numerical simulation are comparable. Namely, the
proposed string model is not valid for the chain of descending escalator under empty
load.

a) b)

Fig.6 Comparison for one-third position of the descending chain between experiment and
numerical simulation (a) transverse vibrations (b) frequency characteristics
 In particular, the chain is characterized by the nature of the chain links and
sprocket teeth. When the tension of the chain is so small, we can’t neglect the
different mechanical mechanism between the chain links and junctions. Namely,
continuous model can’t be applied to study chain. That is the reason the proposed
string model is valid for the ascending chain but failed for the descending chain.
4.2 The effects of pre-tension
In this section, the relationship between amplitude, frequency and tension has
been investigated. As referred above, the proposed string model is valid when the
tension is large enough, so we studied the transverse vibrations of the string model
when the tension is larger than 2000N. Figure 7 shows the relationship between
amplitude, frequency and tension. It can be seen that with the increase of tension, the
amplitude decreases while the frequency increases.

a) b)

Fig.7 The relationship between tension and (a) amplitude (b) frequency
5. Conclusion
In this paper, a Wavelet-Galerkin method is presented for solving the dynamic
behavior of the main drive chain in escalator. It can be found the Wavelet-Galerkin
method makes it possible to avoid numerically calculating integral of
multifold-products of scaling functions and their derivatives in the string vibration
equations. Furthermore, the experiment was carried out by the camera measurement
method for the transverse vibration of chain, whose results were compared with
numerical results. Based upon these results, the following conclusions are obtained:
(1) The Wavelet-Galerkin method presented in this paper can be applied to solve
the coupled nonlinear string vibration equations.
(2) The proposed string model is valid for the ascending chain but failed for the
descending chain under low load.
 (3) For the transverse vibrations of the string model in this paper when the
tension is large, with the increase of tension, the amplitude decreases while the
frequency increases.

6. Acknowledgment
This research was supported by the National Key Research and
Development Program of China under Grant No. 2017YFC0805700.
REFERENCE
1. Wang, K. W., and S. P. Liu. "On the noise and vibration of chain drive systems."
The Shock and vibration digest 23.4 (1991): 8-13.
2. Wang, K. W. "On the stability of chain drive systems under periodic sprocket
oscillations." Journal of Vibration and Acoustics 114.1 (1992): 119-126.
3. Zheng, H., et al. "Efficient modelling and prediction of meshing noise from chain
drives." Journal of Sound and Vibration245.1 (2001): 133-150.
4. Choi, Woosuk, and Glen E. Johnson. "Vibration of roller chain drives at low,
medium and high operating speeds." (1992).
5. Yang, Zhou, et al. "Transversal vibration of chain ropeway system having support
boundary conditions with polygonal action." Shock and Vibration 2015 (2015).
6. Conwell, James C., and G. E. Johnson. "Experimental investigation of link tension
and roller-sprocket impact in roller chain drives." Mechanism and Machine
theory 31.4 (1996): 533-544.
7. Palazzetti, Roberto, and Xiu-Tian Yan. "Study on lubrication effect on motorbike
chain transmissions." Industrial Lubrication and Tribology 68.5 (2016): 561-568.
8. Lenkov, Stanislav N., Vil U. Yamaliev, and Sibagat G. Zubairov. "Vibrodiagnostic
investigations at a chain drive stand of a sucker rod well pump." Bulletin of the
Tomsk Polytechnic University-Geo Assets Engineering 329.5 (2018): 76-82.
9. Conwell, J. C., G. E. Johnson, and S. W. Peterson. "Design and Construction of a
Machine To Evaluate the Forces in Roller Chain Drives." ASME, Design Engineering
Division (Publication) DE43 Pt 2 (1992).
10. Conwell, James C., and G. E. Johnson. "Experimental investigation of link tension
and roller-sprocket impact in roller chain drives." Mechanism and Machine theory
31.4 (1996): 533-544.
11. Conwell, J. C., G. E. Johnson, and S. W. Paterson. "‘Experimental Investigation of
the Impact Force that Occurs when a Roller Seats on the Sprocket During Normal
Operation of a Roller Chain Drive." ASME International Power Transmission and
Gearing Conference. ASME New York, 1992.
12. Fuglede, Niels, and Jon Juel Thomsen. "Kinematic and dynamic modeling and
approximate analysis of a roller chain drive." Journal of Sound and Vibration 366
(2016): 447-470.
13. Mahalingam, S. "Transverse vibrations of power transmission chains." British
Journal of Applied Physics 8.4 (1957): 145.
14. Ariaratnam, S. T., and S. F. Asokanthan. "Dynamic stability of chain drives."
Journal of mechanisms, transmissions, and automation in design 109.3 (1987):
412-418.
15. Destyl, E., et al. "Coupled pendula chains under parametric PT-symmetric driving
force." Physics Letters A 381.46 (2017): 3884-3892.
16. Fuglede, Niels, and Jon Juel Thomsen. "Kinematics of roller chain drives—Exact
and approximate analysis." Mechanism and Machine Theory 100 (2016): 17-32.
17. Fung R F, Huang J S, Chen Y C. The transient amplitude of the viscoelastic
travelling string: an integral constitutive law[J]. Journal of Sound and Vibration, 1997,
201(2): 153-167.
18. Fung, R-F., J-S. Huang, and Y-C. Chen. "The transient amplitude of the
viscoelastic travelling string: an integral constitutive law." Journal of Sound and
Vibration 201.2 (1997): 153-167.
19. Chen, Li-Qun, Wei-Jia Zhao, and Jean W. Zu. "Transient responses of an axially
accelerating viscoelastic string constituted by a fractional differentiation law."
Journal of Sound and Vibration 278.4-5 (2004): 861-871.
20. Chen, Li-Qun, You-Qi Tang, and Jean W. Zu. "Nonlinear transverse vibration of
axially accelerating strings with exact internal resonances and longitudinally varying
tensions." Nonlinear Dynamics 76.2 (2014): 1443-1468.
21. Pernot, Stéphane, and C-H. Lamarque. "A wavelet-Galerkin procedure to
investigate time-periodic systems: transient vibration and stability analysis." Journal
of Sound and Vibration 245.5 (2001): 845-875.
22. Zhou, Y. H., et al. "A wavelet numerical method for solving nonlinear fractional
vibration, diffusion and wave equations." Computer Modeling in Engineering and
Sciences 77.2 (2011): 137.
23. Liu, Xiaojing, et al. "A wavelet method for solving a class of nonlinear boundary
value problems." Communications in Nonlinear Science and Numerical Simulation
18.8 (2013): 1939-1948.