CHAPTER 1

INTRODUCTION

1.0 Concept
Vibration is one of the most important, informative but annoying parameters which
indicates the dynamic condition of rotating machinery. Analogous to the human health
being reflected in one's pulse condition, the inherent machine health status is explicitly
exhibited by the vibration characteristics. The constituents of vibration not only indicate the
severity of the machine dynamic condition but also help in fault assessment and
diagnosis.

Critical and high speed rotating machinery especially those in hydro, thermal and nuclear
power plants require systems for continuous monitoring and analysis of vibration signals for
preventive maintenance and minimization of break down times. It is not only to ensure
production capacities and prevent unscheduled breakdowns, but also to safeguard the
machine itself and the mankind around it. Such continuous monitoring involves high
speed processing of large data, advanced analysis techniques and tools, quicker and efficient
interpretation and diagnosis. This can best be achieved by a computerized approach with
the aid of microprocessor based high speed Data Acquisition (ND)
cards, state of the art
software development environments and analysis tools. Following sections give a brief
outline of fundamentals of vibration, modes of vibration measurement, vibration analysis
techniques and their utility, rotor dynamic characteristics, condition monitoring
methodologies and the integrated approach followed in this thesis.

1.1 An Overview Of Vibration

The back and forth motion of a machine or its part from the position of rest can be
simply stated as vibration (1:igure I. I). The parameters which determine the characteristics of
vibration are
Mass of the vibrating system
Stiffnessof the vibration system
 h n p i n g characteristics of the vibrating system
Excitation forces such as unbalance, misalignment, looseness,
aerodynamiclhydraulic,etc.

The mass (m), stiffness(k) and darnping(c) forces oppose the excitation force(F,Sinwt) to
minimize or control the vibration in a system[ll9]. The fundamental equations of motion for
free and forced vibration are respectively as follows:

m i + & + k x = F~ Sinwt

Vibration is mostly caused due to mechanical problems associated with machinery and
some of the common problems that are known to produce vibration are
Residual unbalance in rotating parts
Eccentricity
Residual misalignment in couplings and bearings
Shaft bend
Component looseness
Worn or damaged parts
Bad drive belts, chains and gears
Bad bearings
Torque variations
Electromagnetic forces
Hydraulic forces
Aerodynamic forces
Resonance
Rubbing
 The three parameters required to define vibration are amplitude, frequency and phase. The
modes of amplitude of vibration measurement are displacement, velocity and acceleration
expressed in microns (peak to peak). m d s e c (RMSor peak) and m/secA? or 'g'
(RMS Or peak) respectively. While frequency is measured in Hz or RPMICPM, phase is
measured in degrees. Displacement, velocity and acceleration are commonly used for low-
to-medium, medium-to-high and high-to-very-high frequency applications respectively.
Generally the failure mode due to dynamic stress, fatigue and force is associated with
displacement, velocity and acceleration respectively. The transducers used for vibration
measurement are proximity probes, velocity seismoprobes and accelerometers, each
measuring the corresponding basic mode, though other modes can also be obtained from the
same transducer by conversion using the frequency content of vibration. Phase is obtained
with a reference on the rotating object using a tacho probe.

Mass, Stiffness and damping characteristics idaround the structure interface, dictate a
component's resonant frequencies, at which it offers minimum resistance to excitation. This
phenomenon when applied to rotating machinery is termed as critical speed. The rotor support
systems depending on whether critical speeds are outside or within the operating speed, are
termed rigid or flexible rotor systems respectively. Most of the turbo generator utility sets
and critical auxiliaries have rotor support systems based on oil lubricated journal bearings, in
which the bearing characteristics also play an important role in rotor response behaviour in
addition to rotor characteristics. The rotor characteristic is based on the material, cross
sections, lengths, additional masses and inertia due to impellers or wheels, while the bearing
properties are guided by the geometry, clearance, lubricant viscosity, load and rotational
speed.

Inspite of every care being taken during the design/manufacture/assembly of these

machinery, the overall tolerances that finally set in may result in typical vibration response
characteristics. The wear and tear due to long and unintermpted runs and interaction due to
the process parameters like liquidlfluid temperature, pressure, flow, thermal expansion, load
etc result in change of system behaviour, which in turn results in changed vibration response.
This necessitates the requirement of continuous vibration monitoring, analysis, assessment
and diagnostic systems not only to ensure the vibration severity to be within the specified
limits but also to aid in assessment on the extent and type of severity and to give msximum
possible guidance for necessary course of action either immediately or in the long run.

1.2 Vibration Monitoring Techniques

Condition monitoring methodologies generally adopted are
breakdowdpreventive/condition-based depending on the category of machinery like general-
purposelimportant~criticalrespectively. Of the various condition monitoring techniques like
aural, visual, temperature, wear-debris, operational variables etc, vibration monitoring is
considered to be the most important, especially for critical and high speed machinery. Other
parameters certainly provide corroborative evidence for high vibration problems, their
sources and effects.

Vibration measurements in rotating machinery are made on bearing housings and shaft
locations nearest to the bearings. While shaft vibrations are measured in two radial
directions mutually perpendicular to each other, bearing housing vibrations are taken in three
directions viz. horizontal, vertical and axial.

The domains of vibration signal monitoring are basically divided into time, frequency and
modal. While frequency domain is the crisp information drawn by tunable filtering or FFT
operation on time signals, the latter also gives enormous amount of information related to
the presence of spikes, beats, randomness, shaft orbit etc in addition to being the source to
statistical analysis like probability density functions, kurtosis etc. Figure 1.2 is the
classification of time domain techniques [120]. Indices in time domain consist of peak level
indicating the maximum level above base line, RMS level indicating the spread of the time
waveform, and their ratio called crest factor. Synchronous averaging is the removal of
aperiodic noise or machine speed unrelated signal with reference to the fundamental
frequency.

Orbits are lissajous figures indicating the loci of the resultant of two mutually
perpendicular probe output waveforms. Statistical analysis in time domain is mostly
applicable to anti frktion bearing condition monitoring for early prediction of impulsive
nature of excitation. The shape of the probability density cuwe is uselvl in the diagnosis of
machine condition which is also described by a series of single-number indices. These are
moments of the curve and are analogous to mechanical moments about the centroid of a
plane. The first and second moments are well-known as the mean and the mean square
values. Odd moments relate information about the position of the peak density relative to the
median value. Even moments indicate the spread of distribution. Moments greater than two
are usually normalized by removing the mean and dividing by the standard deviation raised
to the order of the moment. The third moment is Skewness and the fourth moment is
Kurtosis.

Frequency domain data obtained most commonly by FFT algorithm presents the time
domain signal in the form of amplitudes versus various frequency contents due to various
excitation forces (Figure 1.3). Signature spectrum monitoring either directly or through
masksfenveloping, aids in looking for changes in certain excitation frequencies from time to
time. The classification of frequency domain techniques are given in Figure 1.4. Other
representations in frequency domain are cascade and waterfall plots (Figure 1.5) which are
spectrums stacked against timelspeed, which indicate the changes in the frequency contents
of the spectra with any operating parameter. Plotting of fundamental harmonic magnitude
and phase against speed during coast up and coast down are called Bode-plots or Nyquist
plots comesponding to rectangular or polar form respectively. These plots indicate the
critical speeds of the rotor system.

The third domain known as the modal domain (Figure 1.6) is the pictorial representation
of spectral domain; for each fnquency. the amplitude of vibration is plotted versus the
length/breadth/depth (geometry can be 1Dl2D13D)of the machinery under observation with
reference to a common phase reference. This indicates the nodes and anti nodes of vibration
amplitudes on the machinery under consideration.

With advancement in microprocessor technology and advent of desktop computers and

internet t e c h ~ l o g y ,dramatic changes have taken place in on-line vibration monitoring
systems and methodologies. Figure 1.7 shows the advancement of vibration monitoring
systems over the years in power plants. Though all the methods are still applicable,
appropriate selection of the monitoring methods is needed to suit the requirement based on
the criticality of the machinery.

13 Integrated approach
In view of the reasoning given above, there exists a need for an integrated approach, to
build an on-line vibration monitoring, analysis, assessment and diagnosis guidance system
considering the following aspects
Selection of appropriate vibration transducer systems
Time and frequency domain vibration analysis
Use of high speed data acquisition systems
Vibration parameter trending and severity assessment methodologies
Computing systems like desk top computers
Signal processing and display tools
Knowledge base for general rnalhnction diagnosis
Correlation analysis techniques between vibration and operating parameters
Rotor dynamic characteristics of rotor support Systems
Software tools for building operator guidance systems based on the above
Web site providing the core of the above for remote access
Testing in power plant sites

The present thesis is an attempt to design, develop and implement such an on-line system
for critical rotating machinery in power plants.
 Peak acceleration

/ Peak velocity

(Weight)
'W'
I,-----
PPFR

NEU_RA_L
POSITION
Force 'F' 1
e
iaJop
, .ebn
I
I
i ----point (Point
---- /Neutral
displacement
i
c) 01

/-I-----
---'em.

~orce'F' [ - - - l b--I----
~

r 0 * -,'1p-a-
I
LII-~
------
LOWER
LIMIT
Lc--4
I lme 4 I
~--Frequency (Pefiod)-l

FIGURE 1.2 TIME DOMAIN ANALYSIS TECHNIQUES

 Vibration a?mlysis

FIGURE 1.3 COMMON FAULTS AND THEIR INFLUENCE

ON COMPLEX VIBRATION SIGNAL

FREQUENCY DOMAIN ANALYSIS

II

FIGURE 1.4 FREQUENCY DOMAIN ANALYSIS

TECHNIQUES
 0 50 100 150 200
FREQUENCY: 10 Hcttzldiv

FIGURE 1.5 TYPICAL CASCADE PLOT OF

COMPRESSOR VIBRATION

A) THREE-DIMENSIONAL COORDINATES
SHOWINO FREQUENCY, DISTANCE
AND AMPLITUDE

I, I

B) FREQUENCY DOMAIN VIEW C) MODAL DOMAIN V I M

THE RELATIONSHIP BEWEEN THE FREQUENCY AND THE MODAL W W I N S

FIGURE 1.6 MODAL DOMAIN

TECHNIQUES