VETOMAC 2010

Analysis of Torsional Vibration Characteristics of a

Jaw Coupling

D.Martin Sureshbabu1*, Dr.P.V.Mohanram1, Mr.M.Jeyaselvan2,

1
Dept. of Mechanical Engineering, PSG College of Technology, Coimbatore, India.
2
Manager, Technology Development, ELGI Equipments Ltd, Coimbatore, India.
*
martin_ceo@yahoo.co.in

Abstract

Comfort and convenience enjoyed in this technology driven world is made

possible by the numerous rotating machineries. Without which, everything from
transportation to manufacturing is a standstill. A repair or replacement of these are
very expensive and also time consuming. It is good to see them run for longer period.
To get an enhanced lifespan, it is necessary to prevent or predict imminent failures,
and its source(s). One such source was the torsional vibration experienced by the
system (rotating machinery). Extracting of torsional vibration from the system and
signal is considered as a stringent process. Misalignment in a jaw coupling (rotating
element) is the minor cause, when unnoticed leads to major breakdown and loss of
the system. Fatigue is built on the shafts, due to misalignment induced torsional
vibration. It is must to analyse the torsional vibration characteristics of the system
with and without misalignment condition. Accuracy, time and cost combined together
to adopt a methodology in detecting torsional vibration. Though, a jaw coupling is
capable of handling misalignment to some extent, efficiency in transmitting torque is
the desired one. The decrease in efficiency is due to the heat lost in the elastomer
(spider) during operation. Calculations show that the losses are negligible, but the
analysis is a useful one for larger machines and extensive research. Torsional
vibration experienced due to misalignment is identified by the instantaneous angular
speed of the rotating shaft. With the advent of Digital Signal Processing (DSP) and
applying Phase De-modulation, torsional (modulating) component is extracted
accurately. Processing the modulating component, the angular displacement, angular
velocity and angular acceleration are known to take necessary counter-action.
 VETOMAC 2010

1. Introduction

Due to current trends in the design of rotating machinery towards higher

speeds, manufacturers are tending to produce machines which operate closer to lateral
critical speeds. The need for a better understanding of torsional vibration is becoming
a necessity for practical engineers for the purpose of condition monitoring and
diagnosing. Most rotating machinery consists of a driver and driven machine element
coupled through a mechanical coupling, mainly to transmit torque/power.
Power transmission couplings are widely used for modification of stiffness
and damping in power transmission systems, both in torsion and in other directions
(misalignment compensation). To make a good coupling design is a very difficult
task, because couplings have severe limitations on size and rotational inertia, etc.
Torsionally flexible couplings are used for tuning dynamic characteristics (natural
frequencies and/or damping) of the drive/transmission by changing their stiffness and
damping.
Torsional vibration of a rotating element is the rapid fluctuations of angular
shaft velocity. As a machine changes speed, torque is applied to the shaft in one
direction or other. When the rotational speed of the machine fluctuates during one
rotation of shaft, it is considered torsional vibration. Basic units are either radians or
degrees.
The main sources of torsional vibration in rotating machines are;
i) Unbalance of rotors
ii) Misalignments of shafts
iii) Non-uniform driving torque
The above excitations result in dynamic force, which affects the
systems’ structural failure.

2. Literature Review
Though many books are available on the topic of torsional vibration, none is
described about the analysis of torsional vibration. Wachel & Fred R.Szenasi [1] in
their work clearly stated the analysis procedure, causes, effects and remedy. System
modeling, forced response analysis for different components, and improving the
system reliability were discussed in better way. According to Mark A.Corbo &
Stanley B. Malanoski [2] it is difficult to detect the incipient failures due to torsional
vibration. They stress the need of thorough torsional vibration analysis of an essential
component of the turbo-machinery design process. They provide a practical torsional
vibration procedure.
Various couplings and its roles in machine transmission and comparative
assessment are formulated by Rivin E.I [3]. At the end an optimized coupling design
index is formed. Boris M Klebanov et al [4] details the design and life of the
couplings. As the number of technical papers is less in this area, for selection and
 VETOMAC 2010

analysis the procedure is adopted using manufacturers data in-line with standards
specified by Shigley [5].
The concept of compensating stiffness is derived by Rivin E.I [3]. The force
developed due to the misalignment is calculated using analytical concepts. Different
types of compensating members are analyzed and compared. Parallel misalignment
produces an extra deformation besides the deformation produced by transferring the
torque as analysed by Diangui Huang [6]. It equals one time rotating frequency. The
1x rotating frequency sometime accompanied by smaller torsional vibrations at 2x,
3x, 4x, 5x, 6x and natural torsional frequency. Force due to angular misalignment is
calculated analytically and also using a FEM model. Dimensionless stability criteria
of the non-linear system of differential equations of two misaligned rigid rotors are
derived using Liapunov’s direct method. The system kinetic, potential, and
dissipation energies are derived by Al-Hussain [7]. The results show that an increase
in angular misalignment leads to an increase in the model stability region.
A S Sekhar et al. [8], studies the effects of improper aligning of shafts, which
leads to severe vibration. They modelled the misalignment using FEM analysis. They
conclude that, the 2x vibration response is cause of misaligned shafts. The
applications of flexible couplings for turbo-machinery were discussed by Mancuso
[9]. This reference includes reasons for using flexible couplings, difference between
gear and flexible element coupling, and the selection of couplings for new
applications. Lorenzen et al. [10] introduced a comparison of critical speeds of a
high-speed compressor train alternatively equipped with different types of couplings.
The unbalance response using different types of couplings was calculated, which led
to the conclusion that solid-type couplings can make the system more stable
compared to other types of couplings.
Yuhua Li et al. [11] discusses several methods getting the instantaneous
angular speed. Jiri Tuma et al. [12][13], describes the several methods possible to
measure the angular displacement and in their work gives the details of analog
method of measurement. P S Meirelles at al. [14][15] in their work employs the
measurement techniques for practical applications. They highlight the modeling and
analysis methods and calculations. R Archambanlt [16] explains the sequence of
extraction information. The formation of analytic signal using Hilbert transform, and
the method to get the modulating components was written nicely.

3. Methodology

3.1 Torsional vibration analysis methodology

The analysis is divided into three sections:

1. Theoretical or analytical torsional vibration analysis, and
2. Experimental torsional vibration analysis
3. Coupling selection and analysis
 VETOMAC 2010

Theoretical analysis uses the fundamental concepts, equations and

calculations. This helps to avoid misled interpretation of experiment results. The
basic design concept built upon the theoretical analysis. If the theoretical outcome is
unsatisfactory, the analysis is carried by changing the system parameters. It saves
valuable time, costly fabrication of the system.

3.2 Experimental methodolgy

As seen in the literature studies, there exist many measurement techniques to

measure angular displacement. The accuracy, time and cost is to be considered in the
decision of selection. The two signal acquisition methods:
1. Analog method, and
2. Timer method (digital)
are considered for the extraction. These methods have some advantages as
well as disadvantages. Exactness of experiment can be verified, by performing both
the methods. The analog method considers the transducer signal as ordinary analog
signal, at the same time the timer method treats the transducer signal as train of
digital pulses. A separate methodology formulated for the signal acquisition. Digital
signal processing using Hilbert transform and phase demodulation for the analog
method. And counter based method for the timer method.

4. Analysis

4.1 Theoretical torsional vibration analysis

The equation of motion for a system can be found by many methods. Applying
the Newton’s law for a torsional (rotational) vibratory system gives
(5.1)
 VETOMAC 2010

Before developing a solution of the above equation, it is useful to consider first

some important simplified cases. The major case is considering damping, = 0 and
external force = 0. This gives the following equation, known as equation of free
vibration.
+ (5.2)
Where, I and k are mass moment of inertia and torsional stiffness respectively.
The system (Figure 5.1) parameters required for the analysis are listed as below:

System Parameters:
Mass Moment of Inertia J in kgm^2
Motor Inertia + Fan Inertia J1=0.07 + 0.1117
Coupling hub – motor side J2= 0.018395
Coupling hub – compressor side J3= 0.018395
Crank shaft + connecting rod assembly J4= 0.0425+0.005853

Stiffness in Nm/rad
Motor input shaft stiffness Kt1= 344090
Coupling stiffness Kt2= 3610
Compressor/Top block input stiffness Kt3= 276195

Compressor speed n= 980 rpm

Power P= 14 HP
 VETOMAC 2010

For any torsional vibration analysis, it is mandatory to form a physical model

of the system. The model must have all the required for further calculations. The
physical model (Figure 5.2) of the system is illustrated as below:

Kt1 Kt2 Kt3

J2 J3
J4
J1

Fig. 5.2 Physical model of the system

J1 – Motor Kt1 – Motor shaft torsional stiffness

J2 – Coupling motor side Kt2 – Coupling torsional stiffness
J3 – Coupling compressor side Kt3 – Compressor torsional stiffness
J4 – Compressor
Firstly the equation of motion for the system has to be formed, to proceed further
to carry free vibration. The number of rotating masses or mass inertias in the system
decides the number of equations of motion. As seen from the mathematical model, for
the four rotating masses, four equations of motion formed as below:
(5.3)
(5.4)
(5.5)
(5.6)

The above equations are solved by standard Eigen value approach. Solving the
characteristic equation, one gets the Eigen values and Eigen vectors. Those are
nothing but the torsional natural frequencies and mode shapes. Determination of
natural frequencies and mode shapes are important to avoid resonance is essential in
machine elements. The effects of damping on natural frequencies have been
neglected. Rotating machines can develop excessive stresses in torsion, because of
low torsional natural frequencies of the system involving flexible couplings.
The natural frequencies (Table. 5.1) and mode shapes (Table. 5.2) are as follows.
 VETOMAC 2010

Table 5.1 Torsional Natural Frequencies

ωn1 0

ωn2 266.5 rad/sec

ωn3 4545.1 rad/sec

ωn4 4581.5 rad/sec

Table 5.2 Mode Shapes

Mode ω, J1 J2 J3 J4
No. rad/sec
0 1.9359 1.9359 1.9359 1.9359
1
266.5 -1.1219 -1.0798 3.3214 3.3632
2
4545.1 -0.5662 5.6104 3.7747 -1.4426
3
4581.5 -0.4204 4.2393 -5.0334 1.8819
4

An interference diagram (Figure 5.5) also known as Campbell diagram is

prepared. It includes the details regarding operating speed(s), safety margin(s),
torsional natural frequency (ies) and the exciting orders. Making note of interference
points and safety margin (20%) is essential.
 VETOMAC 2010

Vibration which occurs under the influence of external force is called forced
vibration. The external force known as excitation keeps the system vibrating.
Rotating machines during operations have this kind of vibration. The excitation may
be periodic, impulsive or random in nature. In this analysis periodic excitation is
assumed. The excitation may be external or internal. The external being easily
detected and is due to motion produced by one dynamic system to another. The
torsional vibration of the crank shaft is of particular importance, because this form of
vibration not captured by normal vibration monitoring methods. Many crank shafts
have been broken on account of it. Such failures are unnecessary.
Since the torsional vibrations in the crank shaft are excited by the non-
uniformities either in the driving torque or driven torque. As known, it is made up of
two parts, one due to air pressure and the other due to inertia. The analysis can be
carried out based on the, Torque – Θ curve, which is a system characteristic. It is a
periodic curve. When the machine operated at half load or no load conditions, the
curve gets it shape changed. Or, otherwise whenever there is a change in torque, there
exist change in the curve shape. The fact that the torque is so irregular constitutes one
of the inherent disadvantages of the reciprocating compressor as compared with
dynamic compressor where the torque curve is fairly constant.
It is possible to break up the torque – Θ curve into its harmonic components. The
break up is known as decomposition. Decomposition can be done by wavelet or
fourier series method. The superposition of split harmonic periodic motions gives the
earlier non-harmonic periodic motion. The mathematic theory shows that any
periodic curve f(t), of frequency ω can be split up into a series of sine curves of
frequencies ω, 2ω, 3ω, etc.
(5.9)
Where A0, A1, A2 are fourier coefficients, ω is operating frequency and φ is the
phase component. The magnitude and phase of each order ‘n’ can be found as below:

then

Decomposed data are then added to get the restructured torque –

theta curve and then the restructured curve superimposed on the original torque-theta
curve to check the accuracy of the approximation. The excitation torque (magnitude)
arrived from the decomposition is used for the formation of force vector. From
characteristic equation the modal matrix {φ} is formed. Using the modal matrix,
modal inertia and modal stiffness are calculated.
 VETOMAC 2010

Since, the excitation force (torque) is acting at the compressor end; the exciting
torque function for nth order is formed as;

Next the modal displacement has to found out. To get modal displacement, the
equations of motion have to be uncoupled first, to get the local displacement {q}. The
uncoupling of the equations of motion result in;

The effects and differences between local and global displacement function is
illustrated (Figure 5.9) below.
 VETOMAC 2010

4.2 Experimental torsional vibration analysis

The success of experiment starts with correct selection of transducer. The end
depends on the data processing technique. It is a very difficult task to select a suitable
transducer among many. It lot depends on the application and accuracy of
measurement concerned. Signal conditioning and processing further enhances the
result. The measurement torsional vibration, which is the transformation of
instantaneous time data, the processing should be very effective. The method should
able to process real time data. Torsional vibration data are obtained by the following
procedure:
1. Selection of transducer - It is not a simple parameter to analyze torsional
vibration, because transducer requirements are stringent and shaft access may be
limited. The most commonly used parameter for expressing torsional vibration is
angular displacement, whose units are degrees, peak – to – peak value. Torsional
vibration is the rapid fluctuation of angular velocity of shaft during one rotation.
Details about the speed at each specified instant has to be known, to calculate
torsional vibration. Considering instantaneous sensing, this requires a signal of
many pulses per revolution (PPR), an incremental rotary encoder (1000 PPR) is
used.
2. Signal conditioning
3. Selection of data acquisition method - Based on the mechanism of data
acquisition, the measurement of angular speed can be categorized into two broad
groups:
• Timer/counter-based methods and
• ADC-based methods
The timer/counter-based methods treat the signal from an angular transducer
as pulse train. The ADC-based methods treat an angular speed as an ordinary
analog signal. The angular speed is then derived from the sampled data using an
appropriate angular speed extraction technique.
4. Performance parameters and analysis - A number of parameters like
minimum measurable speed, maximum measurable speed, resolution and number
of channels characterize the performance of torsional vibration measurement
methods.
5. Method signal processing - Signal acquisition is process of converting
physical phenomena into data the computer can use. The transducer (incremental
rotary encoder) converts the electrical signals into digital pulse train.
The output of the encoder in a direct ADC method provides the time-domain
data. Time-domain plots must be used for all linear and reciprocating motion
 VETOMAC 2010

machinery. They are useful in the overall analysis of machine-trains to study changes
in operating conditions. It is then important to extract the phase from the signal, to
know about the modulating component. Only an analytic signal can express the
details about the phase. A method called Hilbert Transform is performed to get the
so called analytic signal. This possesses instantaneous amplitude and instantaneous
phase. The time derivative of the unwrapped instantaneous phase gives the
instantaneous frequency. Removal of operating frequency from it yields the
modulating component in time-domain. This time-domain waveform can be Fourier
analyzed in order to obtain the frequency demodulated spectrum.
Using the two counter large-range method the measurement principles are clearly
stated as the conversion of physical data into angular displacement data. The
measurement of instantaneous angular speed is realized in the following difference
form:

where is angular displacement and is the corresponding time duration.

4.2.1 Experimental set-up

4.2.1.1 Steady-state analysis

Steady state means, the operating conditions remain the same throughout the
measurement period. The machine speed should not be time dependent. The acquiring
of signal should begin only after the driving element (motor) attains its constant
operating speed. The motor is a constant speed AC induction motor. Therefore the
maximum speed of the motor has to be reached before the acquisition of signal.
Steady state analysis comprises of no-load and load tests. A physical layout is as
shown in the block diagram figure 6.9.

The actual experimental setup is shown as below in the figure 6.10.

(a) encoder connected (b) A proximity sensor to
to the motor check start and stop

(c) The complete set-up with ball valve (Circled)

 VETOMAC 2010

Due to noise and disturbance present in the lines, distorted pulses are generated.
The peaks are not horizontal line, but for torsional vibration analysis the amplitude of
the signal is not important. Only the phase contains the angular speed details. The
data acquiring and processing using LabVIEW is displayed in figure 6.12.

The ball-valve is used to apply the load. When the opening in the valve reduces,
the load increases. In the timer/counter-based method, counter input is used. Angular
speed is the topic interest here. Therefore direct frequency measurement will reduce
time and unnecessary calculations.

4.2.1.2 Transient-state analysis

 VETOMAC 2010

The shafting of the compressor, encounters rapid fluctuation of speed during

start-up and start-down. Torsional vibration during this period is visible for the naked
eye. This reveals the behavior of the system under transient conditions, importantly
about the torsional resonance frequencies. Under normal working conditions, the rate
of speed change from idle state to steady state operation is rapid. Such that, it seems
that the system operates smoothly without sign of torsional vibration. The block
diagram lay-out of transient state analysis is same. The difference between this lay-
out and steady state lay-out is that the motor is operated through Variable Frequency
Drive (VFD).
As the behaviour of the system is to be known, transient analysis is performed for
no load condition. Consequently, counter method of an angular speed measurement is
used. It provides good measurement accuracy over a wide speed range, from
standstill to thousands of revolutions per minute. During the transient period the
amplitudes of vibration are high due to torsional resonance.

4.2.3 Coupling selection and analysis

Couplings are compared based on different working parameters. Since, the effect
of misalignment is considered; the Jaw type coupling is selected. It is capable of
accommodating both the misalignments, parallel (radial) and angular. The jaw
couplings are highly reliable for light, medium and heavy duty electrical motor and
internal combustion power transmission applications. This coupling consists of three
parts: driving siding hub, elastomer and driven side hub. The hubs are considered for
inertias, and it is simple. Maximum attention is paid in the selection of elastomer. It
has to be selected considering, the hardness, stability, and stiffness (radial and
torsional). All these in turn are material properties. Torsional vibration must be
controlled to prevent equipment failure.
From the manufactures catalogue, four standard elastomers (TPU material) with
different hardness are selected for the analysis. It is not possible to perform analysis
for the entire range of hardness. It is not needed and also time consuming. The four
choices are made as per the industry requirements. It covers the hardness range, at the
higher end. Elastomers are designed to accept a certain amount of misalignment;
when this is exceeded, identifiable symptoms are usually recognizable. For
understanding purpose, the force induced due to parallel misalignment is discussed.
Table 7.4 gives the details of force and moment due to parallel offset for the
elastomers selected.

Hardness Parallel Misalignment, e = 0.2 mm

Force, N Moment, Nm
Sh 92 A 1880.00 0.376
Sh 98 A 3520.00 0.704
Sh 64 D 5179.20 1.036
Sh 72 D 9140.00 1.828
 VETOMAC 2010

With the new data available the free vibration and un-damped forced vibration
results are shown in the table 7.5.
Natural Coupling
Frequency,
rad/sec Existing Sh 92 A Sh 98 A Sh 64 D Sh 72 D
ωn1 0 0 0 0 0
ωn2 266 243 331 398 522
ωn3 4545 13787 13829 13869 13953
ωn4 4581 15211 15260 15312 15446

The hardness of the elastomer and the inertia of the hubs influence the
interference diagram. It is seen from the diagram (for Sh 92 A), that the system is free
from any operating resonance. Moments due to the misalignments are less compared
to actual exciting torque at the compressor end. Due to the damping, a certain
quantity of the vibration energy is transferred into heat (PV) in the elastomer. The loss
should be minimal in fraction to the input power. Damping power loss has to be
calculated for two reasons: to know about the magnitude of the loss, and to check
correct elastomer is selected. The power loss is calculated for each order and added
according to the following formula:

=Vibratory Torque order, kNm

i = order number
n = speed, rpm and
=Dynamic torsional stiffness,

Taking n = 980 rpm, , i = 12, ψ = 0.8, and from

decomposition (Static torque)

displacements for a horizontal Jeffcott rotor are considered. A transverse surface

crack is assumed at the center of the span. Consider a mass-less elastic shaft of
diameter D and length L with a disk of mass m mounted at mid-span. The transverse
crack at the centre of span has a depth a. The eccentricity of the centre of the disk
mass from the geometric centre of the disk is ε and β represents the orientation of
the eccentricity in the direction of the shaft rotation from the ξ axis. The damping
coefficient is c. θ (t) is the instantaneous angle of rotation and ω is the constant
rotational speed. Considering direct stiffnesses kξ , kη in the ξ , η directions, and
cross coupled stiffnesses kξ η , kη ξ , which come into play due to partial opening of
the crack, the equations of motion can be expressed in the rotating co-ordinates as
 VETOMAC 2010

m(ξ&
&−2ωη&
−ω ξ2 )+ c ξ & (
−ωη + kξ +ξ kη =)
ξη εω
m 2
β−
cos mg θcos (
1
&+ 2ωξ &
m(η& −ω η2 )+ cη( &
+ωξ +
) kξ +
ηξ η
k =η εω
m 2
β+
sin mgθsin )

Citations to be referred by numbers and

4. Results and Discussion listed in the order of occurrence in text,

A program in MATLAB (version 5.3) is written to implement the numerical

solution procedure for the above equations [1]. Stiffnesses are assumed to be constant
for one degree of rotation (π /180 radian), for which the integration of Eq. (1) is
carried out with sufficiently small time step (∆ t=0.000002666 seconds for runb=0.8)
for accurate solution. Table 1 shows the degree of ration with the stiffness,
Table Caption 10 pt
Table 1: Variation of stiffness with Crack Orientation Bold, Title Case

Angle (deg.) Stifness(N/m) Crack Depth(mm)

2.3 50.234 2.3
4.5 45.123 3.4
20.4 23.45 3.5
23 56.00 3.0
34 45.23 2.3
60 22.45Second Level Heading 12 pt3.4 Underlined
leave one line before and after
3.1 Response for one degree

The response obtained at the end of one degree of rotation is stored and again
used to re-evaluate new stiffness values to be used in equation of motion to get next
set of displacements (ξ and η ). Thus response is used to evaluate stiffnesses which
in turn give next set of response.
 VETOMAC 2010

Fig. 1. Frequency domain plot of the response of the cracked rotor during unstable crack
propagation (runb=1.5, β =00) a) horizontal and b) vertical direction.

4. Conclusions Figure Caption 10 pt Bold

The unbalance lateral vibration response of a Jeffcott rotor with a single centrally
situated crack is analysed for unstable crack propagation. It has been observed that
for the supercritical region the response does not show diverging rotor response and
the frequency response also shows only a dominant running frequency component.
However, for the subcritical speed range, the response shows diverging vibration
amplitude and the frequency response indicates a noticeable bending natural
frequency component. The unbalance orientation angle has substantial influence over
the vibration response with the bending natural frequency component is more
prominently observed when the unbalance is in opposite phase with the crack.

References

1. Darpe A.K., Gupta K. and Chawla A, (2003), Experimental Investigations of the

Response of A Cracked Rotor to Axial Excitation, Journal of Sound and
Vibration, vol. 260, pp. 265-286.
2. Whitten, K. W., Davis, R. E., Peck, M. L. and Stanley, G. G. (2004), General
Chemistry, 7th edition, Brooks/Cole-Thomson Learning, pp152-153

Pl follow the style of citation reference

(JSV Style)