Condition Monitoring Lectures

CONTENTS
Module A

1. Introduction
2. Failure
3. Single-Degree-of-Freedom Vibration
4. Multi-Degree-of-Freedom Vibration
5. Balancing
6. Steam Turbine
7. Gas Turbine
8. Generator
9. Pumps

Module B

10. Hydraulic Turbines
11. Fans
12. Wind Turbines
13. Gearboxes and Rolling Element Bearings
14. Condition Monitoring
15. Sensors and Instrumentation
16. Analysis Techniques
17. Fault Diagnosis

MECH7350 is delivered in two intensive modules over one semester. This volume contains the
course notes for Module A. Supplementary material will be provided through lectures from
industry experts.
MECH7350 Rotating Machinery 1. Introduction
1-1
1. INTRODUCTION

(This section is based largely on Black and Veatch)

Fig. 1.1 is a diagram of a typical pulverised coal-fuelled electrical generating facility. In this
section the main components are identified and an overview of the course is presented.
Throughout the course typical rotating machinery is explained. It is left to students to
identify features of particular plant in their situations.

1.1 Coal Handling and Pulverising
Coal is usually delivered to the facility by trains or long conveyor belts. The coal handling
system unloads the coal, then stacks, reclaims, crushes and conveys it to a storage silo. Coal
is fed from the storage silos, pulverised to a powder, and blown into the steam generator.
Fig. 1.1 Typical pulverised coal fuelled electrical generating facility (from Black and Veatch)
1-2
There is rotating machinery in the coal handling and pulverising systems but it will not be
considered explicitly in this course. There are electric motors, gearboxes and mills that have,
for example, generic bearing and vibration issues but such issues will be addressed when
other parts of the overall facility are addressed.

1.2 Steam Generator
Pulverised coal is mixed with air in the steam generator and combusted, and the combustion
energy is used to produce, to superheat and to reheat steam. The only parts of the steam
generator that will be addressed are the forced draft fans.

1.3 Turbine
The steam turbine converts the thermal energy of the steam to rotating mechanical energy,
and the generator which is coupled to the turbine, converts the mechanical energy to
electrical energy. Aspects that are addressed include:
Configurations
Speed of rotation
Design features
Casings
Couplings
Alignment
Rotors
Balancing
Blading
o Types of blade; impulse and reaction stages
o Materials
o Losses
o Blade attachment
o Blade vibration
o Blade erosion
Bearings
Bearing vibration
Instrumentation
Condition monitoring
1-3

1.4 Generator
Aspects addressed include:
Basics of synchronous generator theory
Three-phase windings
Rotor design features
Bearings
Rotor winding
Sliprings, brushgear and shaft earthing
Stator design features
Stator winding
Cooling
Excitation systems
Vibration
Balancing
Instrumentation

1.5 Condenser and Cooling Towers
Steam exhausted from the low-pressure section of the steam turbine is condensed to liquid in
the condenser. The condensed water is moved from the condenser by condensate pumps
through low-pressure regenerative feedwater heaters to a deaerator. Boiler feed pumps move
the deaerated water through high pressure regenerative heaters to the steam generator. Other
pumps are used for circulating water through the cooling towers, and for oil lubrication.
Aspects of pumps that will be addressed include:
Design features
Performance
Configurations
Operating characteristics
Cavitation

1-4
1.6 Forced Draft, Primary Air and Induced Draft Fans
Combustion air is supplied to the steam generator by forced draft fans. Primary air fans
transport pulverised coal into the steam generator. Induced draft fans remove the flue gases
from the steam generator and exhaust them to the stack. Aspects of fans to be addressed
include:
Design features
Performance
Configurations
Operating characteristics

1.7 Electric Motors
The generator is the largest rotating electrical machine in a power generation plant.
However, many electric motors, large and small, are used. Issues to be addressed include:
Rolling element bearings
Balancing

1.8 Gearboxes
There are a number of large gearboxes in a modern power station, e.g. on the coal pulveriser.
Issues to be addressed include:
Design of gear trains
Rolling element bearings
Fatigue
Lubrication

1.9 Other Types of Rotating Plant
Other types of rotating plant to be addressed are:
Gas turbines, current and future
Wind turbines
Hydraulic turbines
1-5
1.10 Fundamentals
Rotating machinery can fail catastrophically or wear to such an extent that performance is
degraded or there is a risk of catastrophic failure. To enable an appreciation of failure modes,
the course includes an introduction to the phenomena of wear and fracture mechanics of
materials.

Component failures are often caused by the cyclic forces that can develop in rotating
machinery. A fundamental understanding of the following is presented:
Newtonian dynamics in rotating machinery
Vibration theory
Balancing
Vibration measurement
Signal processing
Fault diagnosis using vibration analysis

Basic understanding of the design, performance and behaviour of turbines, fans, pumps and
hydrodynamic lubrication requires some knowledge of fundamentals of fluid mechanics.
This is presented.

MECH7350 Rotating Machinery 2. Failure
2-1
2. FAILURE MECHANISMS
(This section is based largely on J uvinall and Marshek)

When machines in generating plant fail there is a cause and a failure mechanism which
involves materials and their properties. This section addresses failure mechanisms.

2.1 Some Terms for Elastic Materials
Elastic materials deform linearly under load and return to their original shape after the load is
removed. Consider a bar of elastic material deformed under a tensile load P.
=axial stress =
P
A
(pascals, Pa)
=axial strain =
L
L
(dimensionless)
E =
=modulus of elasticity, or Youngs modulus (Pa)

Consider a cube of elastic material deformed under shear forces.

=shear stress =(shear force)/area, (Pa)
=shear strain, (dimensionless)
G =
xy
xy
=modulus of rigidity, or shear modulus (Pa)

P
P
Area,
A
yx

xy

yx

xy

1
2
xy

1
2
xy

2-2
2.2 Fracture Mechanics
2.2.1 Types of Failure
Failure of a loaded member can be regarded as any behaviour that renders it unsuitable for its
intended function. Static loading can result in objectionable deflection and elastic instability,
as well as plastic distortion and fracture. Distortion, or plastic strain, is associated with shear
stresses and involves slipping along natural slip planes. Failure is defined as having occurred
when the plastic deformation reaches an arbitrary limit. Fracture, on the other hand, is clearly
defined as the separation or fragmentation of a member into two or more pieces. It normally
constitutes a pulling apart, associated with tensile stress.

In general, materials prone to distortion failure are classed as ductile, and those prone to
fracture without significant prior distortion as brittle. Unfortunately, there is an intermediate
grey area wherein a given material can fail in either a ductile or a brittle manner, depending
on circumstances. Normally ductile materials can fracture in a brittle manner at sufficiently
low temperatures. Other factors promoting brittle fracture are sharp notches and impact
loading.

2.2.2 Basic Concepts
The fracture mechanics approach begins with the assumption that all real materials contain
cracks of some size even if only submicroscopic. If brittle failure occurs, it is because the
conditions of loading and environment (primarily temperature) are such that they cause an
almost instantaneous propagation to failure of one or more of the original cracks. If there is
fatigue (cyclic) loading, the initial cracks may grow very slowly until one of them reaches a
critical size, at which time total fracture occurs

2.2.3 Stress Concentration
g
= gross-section tensile stress

2
P
wt
=

P
P
t
Stress concentration
2w
2-3
However, stress is much higher at the base of a crack (stress concentration factor). If radius
of crack root approaches zero, stress concentration factor approaches infinity. This means
that ductile yielding will occur in a small volume of material at the crack root, and the stress
will be redistributed. Thus the stress concentration factor is considerably less than infinity.

In the fracture mechanics approach, a stress intensity factor K is evaluated theoretically (more
soon) and is compared with a limiting value of K that is found from standard tests to be
necessary for crack propagation in that material. The limiting value is called fracture
toughness or critical stress intensity factor
c
K . Failure occurs when K exceeds
c
K .

Values of
c
K are substantially lower for thick members (plane strain) than for thin members
(plane stress), so it is conservative to assume thick members. Thick members offer less
opportunity for redistributing high crack root stresses by shear yielding.

Table 2.1 contains typical mechanical properties of 25.4 mm thick plates made of common
aircraft structural materials. Note:
The relatively high fracture toughness of the titanium alloy in comparison to its
ultimate strength
u
S ;
The room temperature comparison of
c
K for the two steels of nearly equivalent
ultimate strengths; and
The reduction in
c
K with temperature for the high-toughness D6AC steel.
Yield strength
y
S is the tensile stress at which plastic yielding first occurs in a specimen
tensile test. Ultimate stress
u
S is the stress in the tensile stress specimen when it is carrying
the maximum possible load before failure. S
y
and S
u
are defined below.

2-4
Table 2.1 Strength properties of 25.4 mm thick plates values of Ultimate Stress
u
S , Yield Stress
y
S and
Critical Stress Intensity Factor
c
K .

Material Temperature
u
S
MPa
y
S
MPa
c
K
MPa(m)
1/2
7075-T651 aluminium alloy Room 538 483 29.67
Ti-6A1-4V (annealed) titanium alloy Room 896 827 71.44
D6AC high toughness steel Room 1517 1310 76.90
D6AC high toughness steel -40C 1565 1358 49.46
4340 steel Room 1793 1496 57.15

Cracks generally begin in thick members at the surface, and have a somewhat elliptical form,
as shown adjacent. Research has established that if:
6
2
>
t
w

2
a
c
= about 0.26
3 >
c
w

0.5
a
t
<
0.8
g
y
S
<
Then at the edge of the crack, K is given approximately by:

( )
2
0.39 0.053 /
g
g y
a
S

Fracture would be predicted for K >
c
K .

2.3 Fatigue
Fatigue failure might better be described as progressive fracture under fluctuating or repeated
loading. Fatigue fractures begin with a minute (usually microscopic) crack at a critical area
t
2-5
of high local stress. This is almost always at a geometric stress raiser. Fatigue failure results
from repeated plastic deformation, such as breaking of a wire by bending it back and forth.
Whereas a wire can be broken after a few cycles of gross plastic yielding, fatigue failures
typically occur after thousands or even millions of cycles of minute yielding. Fatigue failure
can occur at stress levels far below the conventionally determined yield point or elastic limit.

The initial fatigue crack usually results in an increase in local stress concentration. As the
crack progresses, the material at the crack root at any particular time is subjected to the
destructive localised reversed yielding. As the crack deepens, thereby reducing the section
and causing increased stresses, the rate of crack propagation increases until the remaining
section is no longer able to support a single load application and final fracture occurs.

Engineering practice relies on empirical fatigue data from the standardised R.R. Moore
fatigue test rotating beam, shown diagrammatically below.

S-N curves are generated for materials. The figure below is typical. (S =stress, N =number
of cycles to failure at the amplitude S of oscillatory stress.)

Notch
Small region behaves
plastically
Main body behaves
elastically

2-6

The adjacent figure is a typical S-N curve for
steel and shows an endurance limit. This is the
stress below which fatigue failure does not
occur, even for an indefinitely large number of
loading cycles. For a low carbon steel the
endurance limit is about one-half of the ultimate
strength.

2.4 Surface Damage
More machine parts fail through surface damage than breakage. Various mechanisms for
surface damage are described briefly.

2.4.1 Corrosion
Corrosion is the degradation of a material (normally a metal) by chemical or electrochemical
reaction with its environment. It can combine with static or fatigue stresses to produce a
more destructive action than would be expected by considering the actions of corrosion and
stress separately.

2.4.2 Cavitation Damage
Cavitation damage is the formation of bubbles in a liquid that is moving with respect to a
nearby solid surface. Bubbles are formed when the liquid pressure drops below its vapour
pressure. When these bubbles subsequently collapse at or near the solid surface, pressure
waves impinge upon the surface causing local stresses that can be great enough to cause
plastic deformation of many metals. A surface damaged by cavitation appears roughened,
with closely spaced pits. In severe cases, enough material is removed to give the surface a
spongy texture. Cavitation commonly occurs in centrifugal pumps and turbine blades.

2.4.3 Adhesive Wear
When two surfaces slide across each other, the contact pressure and frictional heat of sliding
are concentrated at the small local areas of contact (asperities). Local temperatures and
pressures are extremely high and conditions are favourable for welding at these points. These
welds fail in shear, and new welds form, and so on. This is called adhesive wear. Loose
particles resulting from the wear can cause further damage.
2-7

2.4.4 Surface Fatigue
When curved elastic bodies, such as parts of a rolling-element bearing, are pressed together,
finite contact areas are developed because of deflections. These contact areas are so small
that very high compressive stresses can result in a cyclic manner. Fatigue failures can be
initiated by minute cracks that propagate to permit small pieces of material to separate from
the surfaces. This is pitting or spalling.

MECH7350 Rotating Machinery 3. Single Degree-of-Freedom Vibration
3-1
3. SINGLE DEGREE-OF-FREEDOM VIBRATION

Understanding of complex vibration problems begins with understanding of the vibration of a
system with a single degree-of-freedom, i.e. a system for which the motion can be described
by the time variation of a single coordinate. In many situations the important features of the
behaviour of complicated multi-degree-of-freedom systems can be described adequately with
a single degree-of-freedom system.

3.1 Notation
Vectors are used in this section. A vector quantity is represented by an underscore. For
example, F
can be a force which has a scalar magnitude F and a direction of action.

The notation F
indicates the sum of vectors, i.e.,

1 2
.....
n
F F F F = + + +

This is shown diagrammatically in Fig. 3.1.

Displacement x
, velocity v
and acceleration a
can also be regarded as vector quantities,

although a scalar representation is adequate in systems with a single degree-of-freedom. The
following notation is used for time-derivatives.
dx
v x
dt
= =
dv
a v x
dt
= = =
Fig. 3.1 vector notation

1
F

2
F

3
F

n
F

3-2
3.2 Units
mass kilograms, kg
time seconds, s
displacement metres, m
velocity m/s
acceleration m/s
2

force kg m/s
2
(from Second Law), called newtons, N
angle radians, rad

/ s r =
For a full circle,
0
2
2 360
r
rad
r
= = =

d
dt
= angular velocity, rad/s

2/60 rad/s = 1 rpm

3.3 Some Fundamentals of Dynamics
Here some important dynamics results are summarised in the context of rotating machines.
Full detail is available in Bedford and Fowler.
3.3.1 Newtons Laws
Newtons laws of motion were first enunciated in Sir Isaac Newton (1687) Philosophiae
Naturalis Principia Mathematica. They are written in modern language as follows (from
Bedford and Fowler).
First Law: When the sum of the forces acting on a particle is zero, its velocity is constant.
In particular, if the particle is initially stationary, it will remain stationary.
Second Law: When the sum of the forces acting on a particle is not zero, the sum of the
forces is equal to the rate of change of momentum of the particle. If the mass is constant, the
sum of the forces is equal to the product of the mass of the particle and its acceleration.
mv =
momentum

( ) d mv dv
F m
dt dt
= =

if m is constant

2
2
d r
m ma
dt
= =

r
s

3-3
Third Law: The forces exerted by two particles on each other are equal in magnitude and
opposite in direction; e.g. gravitational attraction, swinging a mass on a string.

3.3.2 Rotational Equivalent
Consider holding a shaft with two hands and twist each in opposite directions. This is
equivalent to applying a moment (or torque) to the shaft with zero resulting force.

T = rF + rF = dF
Strictly, T is a vector. It acts about an axis which has direction. Consider again the Second
Law for translation;
F = ma
Mass might well have been called inertia. The rotational equivalent is:
T = I
Where = angular acceleration
=
d
dt
2
2
d
dt
=
I = moment of inertia, kgm
2
For a particle of mass m rotating in a circular path of radius r, I = mr
2
about the axis.
For a uniform rigid disc of mass m and radius r, I = mr
2

Translational (linear) momentum of mass m is mv

Angular momentum of a rigid body about a fixed axis is I

Kinetic energy of translation of mass m is mv
2
.
Kinetic energy of rotation is I
2
. (Detail is in Bedford and Fowler.)
T
F
F
is equivalent to
radius r
j
m
k
m
jk
f
kj
f
jk kj
f f =
3-4
3.3.3 Centre of Mass
Consider a system of two particles.
Define centre of mass as;
1 2
1 2
m r m r
r
m m
+
=
+

1
r =
if m
2
is small.
Differentiate twice with respect to time.
1 1 2 2
1 2
m r m r
r
m m
+
=
+

Consider both external and internal forces acting on each particle.
For
1
m , Second Law gives
1 21 1 1
F f mr + =

For m
2
,
2 12 2 2
F f m r + =

Add:
( )
1 2 21 12 1 2
0 Total mass of system Sum of external forces on system
F F f f m m r
=
+ + + = +

So;
Sum of external forces on a system of particles = total mass of system acceleration of
mass centre
F ma =

3.3.4 Vibration of Single Degree-of-Freedom Systems
Consider, for example, the simplest model of vibration of
an unbalanced turbine (rotor + casing) on elastic
mountings (Fig. 3.2). For simplicity, turbine is
constrained to move only vertically. A similar model
could be set up for horizontal vibration. Assume:
M = total mass of turbine plus casing
m = equivalent unbalance point mass
k = spring stiffness of support; (spring force = kx)
c = viscous damping coefficient of support;
(damping force = c x )
Viscous damping is a convenient approximation to more realistic non-linear damping because
it leads to analytical solutions that approximate actual systems well.

y
x
m
2
m
2
1
r

1
r

Fig. 3.2 (from Thompson).
3-5
In the equilibrium position (rotor stationary), 0 x x = =
And spring force balances weight. Only changes from equilibrium
are considered, so weight (gravity) can be ignored.

Treat as a system of two particles, m and M-m, in displaced position shown.
Position of centre of mass is given by:

( ) ( sin )
( )
M m x m x e t
x
M m m
+ +
=
+

Apply Second Law to motion of mass centre, recognising that x is positive upwards.

2
2
( ) ( sin ) d M m x m x e t
kx cx M
dt M
+ + (
=
(

[ ]
2
2
sin
d
Mx me t
dt
= +

2
sin Mx me t =
Rearrange to get differential equation of motion of turbine casing.

2
0
sin sin Mx cx kx me t F t + + = = (3.1)
where F
0
= me
2
= centripetal force (constant for constant ).
me = unbalance.
Same unbalance can be caused by larger m at smaller e.
Some special cases are now considered.

3.3.5 Free Vibrations
No rotation of turbine rotor; = 0.
Equation of motion (3.1) becomes:
0 Mx cx kx + + =
0
c k
x x x
M M
+ + =
Rewrite as:

2
2 0
n n
x x x + + = (3.2)
where
n
k
M
= and
2
n
c
M
=
The physical significance of
n
and will become apparent.
System motion is caused by initial conditions,

Mg
kx
o

3-6
(0) x = initial displacement at t = 0, and
(0) x = initial velocity.
Solution of (3.2) is:
2 2
2
( ) ( )
sin 1 ( ) cos 1
1
n
t n
n n
n
x o x o
x e t x o t
(
+
= + (
(

(3.3)

This can be obtained from the theory of
differential equations, Laplace transforms, or by
substituting (3.3) into (3.2). Solution is the sum
of oscillatory sine and cosine terms, multiplied by
a time-decaying exponential (Fig. 3.3).

For no damping, c = 0 and = 0. The
solution of (3.2) is:
( )
sin ( ) cos
n n
n
x o
x t x o t
(
= +
(

The system vibrates at undamped natural frequency
n
k
M
= .
For < 1, the frequency of damped oscillation is
2
1
d n
= .
For > 1, the solution of (3.2) is:
2 2
( 1) ) ( 1) )
n n
t t
x Ae Be
+
= + where:
2
2
(0) ( 1) (0)
2 1
n
n
x x
A

+ +
=

2
2
(0) ( 1) (0)
2 1
n
n
x x
B

=

The motion is an exponentially decreasing function
of time, as shown in Fig. 3.4.

Fig. 3.3

Fig. 3.4
3-7
For
2
1, 1 0 = = and the solution of (3.2) is:
[ ] { }
(0) (0) (0)
n
t
n
x x x x t e

= + +
Three types of responses are shown in Fig. 3.5 for
different initial velocities.
This is called critical damping. If damping is any
higher, exponential decay is slower.
For 1, 1
2
n
c
M
= = so 2
critical n
c M =
Then is the ratio of actual damping present, c, to damping, 2
n
M needed to achieve
critical damping. is called the damping ratio.

Summary = 0 undamped free vibrations
< 1 underdamped oscillations
1 = critical damping
> 1 overdamped
The frequency of damped vibrations is
2
1
d n
= rad/s. This differs by only a small
amount from
n
for small . Hence
n
can be estimated well from the observed frequency
in transient bump tests.

3.3.6 Complex Frequency Response
In this section the system is forced to vibrate at frequency . It is convenient to use the
concept of phasors. Complex variable theory gives:
( )
0 0
cos sin
i t
X e X t i t
= + where 1 i =
So
( )
0 0
sin Im
i t
X t X e
=
Recall that

( )
1 2 1 2
1 2 1 2
.
i i i
Ae A e A A e
+
=
and
( )
1
1 2
2
1 1
2 2
i
i
i
Ae A
e
A e A
=
Return to (3.1).
2
0
sin sin Mx cx kx me t F t + + = = where
2
0
F me =

Fig. 3.5
3-8
Rewrite as
2 0
2 sin
n n
F
x x x t
M
+ + = (3.4)
Solution is the sum of a constant amplitude forced vibration (particular integral) and a
transient response (complementary function) which decays to zero with time. We are
interested in the forced response after transients have died away.
Write [ ]
0 0
sin Im cos sin
F F
t t i t
M M

= +
`
)

0
Im
i t
F
e
M

=
`
)

We could have equally validly assumed a cosine forcing function in (3.5);
0 0
cos Re
i t
F F
t e
M M

=
`
)

So put
0 0
sin
i t
F F
t e
M M
= in (3.4).
Now vibration x will also be a sine (or cosine) function but out-of-phase with
0
sin
F
t
M
.
So let
( )
0
i t
x x e

= (x lags
0
F by )

i t
xe
= where
0
i
x x e

= = complex amplitude of vibration
Then (3.4) becomes
2 2 0
2
i t i t i t i t
n n
F
xe i xe xe e
M

+ + =
( )
2 2 0
2
n n
F
i x
M
+ + =
2
2 2 2
0
1/ 1
/ 2 1 2
n
n n
x
F M i i

= =
+
where
n
=
Take amplitudes (moduli).
( ) ( )
2
0
1
2 0 0 2 2
2
1/
/ /
1 2
n
x x
F M F M

= =
(
+
(

(3.5)
Using
2
/
n
k M = gives,

( ) ( )
1
2 0 2 2
2
1
/
1 2
static
x x
F k x

= =
(
+
(

and
2
2
tan
1

3-9
Case A Rotating unbalance

2
0
F me =
So
( ) ( )
2
0
1
2
2 2
2
1 2
Mx
me

=
(
+
(

(3.6)
0
Mx
me
and are plotted on Fig. 3.6.

Case B
0
F = constant, independent of . It follows that
( ) ( )
0
1
2 0 2 2
2
1
/
1 2
x
F k

=
(
+
(

and (3.7)
2
2
tan
1
(3.8)
(3.7) and (3.8) are plotted in Fig. 3.7.

Fig. 3.6
3-10

Case C Vibration Isolation
Spring and damper transmit forces to foundation that are 90
o
out-of-phase.
So magnitude of force transmitted is
( ) ( ) ( )
1 1
2 2 2
2 2
0 0 0
1 2
T
F kx c x kx
( (
= + = +

(3.9)
Combine (3.7) and (3.9) to get force transmissibility,
0
T
F
TR
F
= =
( )
( ) ( )
1
2 2
2
2
2
1 2
1 2

(
+
(
(
+

(3.10)
This is plotted in Fig. 3.8.

Fig. 3.7
3-11

3.3.7 Vibration Measuring Instruments
Consider a vibrating mass m suspended inside the casing of a vibration measuring instrument,
which in turn is attached to a vertically vibrating surface (Fig. 3.9).

Fig. 3.8

Fig. 3.9.
3-12
Displace m in positive y-direction with positive velocity and draw a free-body diagram.
Assume weight of m is balanced by initial spring
force.
Equation of motion is
( ) ( ) c x y k x y mx =
Assume instrument gives an output signal
proportional to relative motion z = x-y.
If
0
sin y y t =
then
2
0
sin mz cz kz m y t + + =
This is identical to the case of rotating unbalance, with z and m
2
y replacing x and me
2
.
So, if ( )
0
sin z z t = , then
( ) ( )
0
1 2
2 0 2 2
2
1
1 2
z
y

=
(
+
(

(3.11)
When the natural frequency
n
of the instrument is high compared to that of the vibration
0
sin y y t = to be measured, / 1
n
= and
( ) ( )
1
2
2 2
2
1 2
(
+
(

approaches unity.
Then
2
2 0 0
0 0 2 2 2
n n n
y y acceleration
z y

= = = =

Hence z is proportional to the acceleration to be measured. Fig. 3.11 is a plot of
0
2
0
/
n
z
y
for
various damping ratios.

m
( ) k x y
( ) c x y
+ve y

Fig. 3.10.
3-13
If we can choose 0.7 = , we can get a useful frequency range 0 / 0.20
n
with a
maximum error of 0.01 percent.
But the widely used piezoelectric crystal accelerometers (barium titanate) have 0 and can
operate up to 0.06
n
= .
/ 2
n n
f = can be as high as 50,000 Hz, so instruments can operate to 3000 Hz. Fig. 3.11
shows the rugged construction of a piezoelectric accelerometer.

The crystal produces a charge q
proportional to z and so to
acceleration
2
. y But crystal has a
very small capacitance C, so voltage
V=q/C is greatly reduced if output
cable has a high capacitance. This is
overcome by using a charge
amplifier.

3.4 Whirling
(This section is based largely on Thompson)

Rotating shafts, with or without rotors, tend to bow out at
certain critical speeds and whirl. Whirling is the rotation of the
plane made by the bent shaft and the line of centres of the
bearings. The shaft and rotor (and SG in the figure below)
rotate with angular velocity . Whirling (and rotation of OS)
occurs at angular velocity
which may or may not be equal to

and can be in the same or the opposite direction to .

Whirling can be caused by mass unbalance, eccentricity, cyclic
fluid friction in bearings and gyroscopic forces. Synchronous
whirl is when =
.

Fig. 3.11 Construction of a piezoelectric accelerometer.

3-14
If
k
m
= , the critical speed, the amplitude of whirl is limited only by the damping which is
usually small. Severe damage can occur. If rotor operation is above critical speed, run-up
must pass quickly through that speed.

Long shafts without a rotor can whirl. This is a complicated problem to analyse because of
its distributed-parameter nature.

3.5 Spring Stiffness

Only linear springs are considered in this course. For small displacements, most materials
behave approximately linearly. This is an adequate assumption when calculating or
measuring natural frequencies.

Force, f kx =
k = spring stiffness
x = displacement from unstretched position

Moment, T =
t
k

t
k = torsional spring stiffness
= rotational displacement (rad)
from unstrained position

Springs in series

1
k
2
k
f
x

x x
time time
Fast run-up Slow run-up

f
x

T

1 2
1
1/ 1/
k
k k
=
+

3-15

Example

Shafts in bending are springs for unbalance in rotor. Are springs in series or parallel?

Shafts in torsion (length L)

Torsional spring stiffness,
t
T JG
k
L
= =
J = second polar moment of area.

Shafts in bending
I = second cross-sectional moment of area

1
k
2
k
f

1 2
k k k = +

Rotor

d
4
32
d
J

= ( )
4 4
32
o i
J d d
=

h
b
Neutral
axis
3
12
bh
I =

Circular shaft, diameter d
4
64
d
I

=

3-16
Some material properties
Material Modulus of Elasticity, E Modulus of Rigidity (shear
modulus), G
Alloy or carbon steel
Copper
207 GPa
121 GPa
79 GPa
46 GPa

The following are spring stiffnesses for a range of configurations of beams in bending (from
Thompson).

MECH7350 Rotating Machinery 4. Multi-Degree-of-Freedom Systems
4-1
4. MULTI-DEGREE-OF-FREEDOM SYSTEMS

(This section is based largely on Thompson)

4.1 Translational Systems
In the following example (Fig. 4.1) damping is assumed to be negligible.

Draw free-body diagrams for each mass displaced in the positive direction.

Apply Second Law to each mass.

Look for modes of vibration where each mass vibrates harmonically at the same frequency
and passes through equilibrium at the same time (in-phase or 180 degrees out-of-phase).
Put;

1 1 1
2 2 2
sin
sin
i t
i t
x A t or Ae
x A t or A e
=
=

Then the two equations of motion become;

( )
( )
2
1 2
2
1 2
2 0
2 2 0
k m A kA
kA k m A
=
+ =
(4.1)
Or, in matrix notation;

2
1
2
2
0 2
0 2 2
A k m k
A k k m

=

For non-trivial solutions (i.e. not
1 2
0 A A = = ),
( )
( )
1 1 2 1 1
1 2 2 2
2
kx k x x m x
k x x kx mx
=
+ =

Fig. 4.1 (from Thompson).

m 2m
1
kx ( )
1 2
k x x
2
kx

4-2

2
2
2
0
2 2
k m k
k k m

=

Put
2
= and expand;

2
2
3 3
0
2
k k
m m

+ =

This is called the characteristic equation of the system. The two roots,
1
and
2
, are the
eigenvalues of the system.
Here
1
2
0.634
2.366
k
m
k
m
=
=

and the natural frequencies are;

1 1
2 2
0.634
2.366
k
m
k
m

= =
= =

From (4.1) the ratio of the amplitudes of vibration can be found;

1
2
2
0.731
2
A k
A k m
= =
for
1
=
= -2.73 for
2
=
These are plotted in Fig. 4.2.
These are the natural modes of vibration.
At
1
, masses vibrate naturally in phase.
At
2
, masses vibrate naturally out of phase (or in opposition).
The actual amplitudes of vibration at natural frequencies depend on the magnitude, a, of
initial conditions.

1
x

a

=
2
Initial
displacement
=
1

a 0.731a
1
x 2
x
2
x
2.73a

Fig. 4.2
4-3

If any general set of initial conditions are applied, the system will vibrate in a combination of
natural modes and beating will occur, as shown below.

4.2 Rotational Systems
Consider the rotational system in Fig. 4.3, with three degrees-of-freedom.
t
k = torsional stiffness (torque T = k
t
)
I = moment of inertia

Draw free-body diagrams for each rotor, with a positive .

1
I 2
I
3
I
1
t
k
2
t
k

Fig. 4.3.

( )
1
1 2 t
k
( )
1
1 2 t
k
( )
2
2 1 t
k
( )
2
2 1 t
k

4-4
For rotor 1: ( )
1
1 2 1 1 t
k I =

For rotor 2: ( ) ( )
2
1
1 2 2 3 2 2 t t
k k I =

For rotor 3: ( )
2
2 3 3 3 t
k I =

Put;

)
1 1 1
)
2 2 2
)
3 3 3
sin ( )
sin ( )
sin ( )
i t
i t
i t
t or e
t or e
t or e

=
=
=

In matrix form;

( )
1 1
1 1 2 2
2 2
2
1
1
2
2 2
2
3
3
0
0
0
0
0
t t
t t t t
t t
I k k
k I k k k
k I k

Form is 0 A =
For non-trivial solutions, 0 A =
This gives the three eigenvalues,
1 2 3
0, , =
For
1 1 2 3
0, 0 and = = = =
This is a rigid body rotational mode.
The two natural frequencies of vibration are
1 2
and

Packages such as MATLAB can be used to do the modelling and the matrix algebra, and
generate the natural frequencies and mode shapes.

4.3 Vibration of Continuous Systems
Because mass and elasticity are distributed continuously, an infinite number of coordinates
are needed to describe these systems, i.e. they have an infinite number of natural frequencies
and mode shapes. It is necessary to solve partial differential equations as opposed to the
ordinary differential equations for lumped parameter systems.

Some low-order natural frequencies are shown in Table 4.1 for transverse vibration of a
number of beams with different end conditions.

4-5
Table 4.1.
Here:
2
4
EI
= = density = frequency, rad/s

MECH7350 Rotating Machinery 5. Balancing
5-1
5. BALANCING

(This section is based largely on Rao)

5.1 Types of Rotor Unbalance
In a thin rotor disc, unbalance mass lies in a single plane and the resultant unbalance is a
single radial force. Circumferential location of the balance mass can be found by letting the
rotor rotate under gravity on a shaft on knife edges or in
low friction bearings (Fig. 5.1), until it finds a rest point.
To balance, remove mass by trial and error on line OP, or
add mass by trial and error on line OQ. Balance is
achieved when the rotor comes to rest with no preferred
orientation. This is called static or single plane
balancing.

When the unbalance is in more than one plane, e.g. on a long rotor, the
resultant is a force and a rocking moment. A long rigid rotor, such as a motor armature, can
be considered to be a series of thin discs, each with some unbalance (Fig. 5.2). This is
equivalent to a single unbalance force plus a moment, or an unbalance force in each end
plane. The rotor must be spun in flexible supports to detect and correct the unbalance. This
is called dynamic balancing.

5.2 Single-Plane Balancing
Eccentricity or unbalance can cause transmission of unacceptable vibration to shafts and
bearings. Some rotating bodies can be treated as a thin circular disc, e.g. a fan or a pump
impeller, but not a long turbine or generator rotor. Eccentricity is caused by inaccurate
alignment of a disc on a shaft, e.g. inaccurate machining of a bore. Unbalance can be caused
by wear of a part of a disc, or by inaccurate machining. Both eccentricity and unbalance can
be represented by a point mass
u
m at radius
u
r on an otherwise uniform disc; i.e. unbalance

Fig. 5.1 Static balancing.

Fig. 5.2.
5-2
u u
m r causes an unbalance centripetal force
2
u u
m r . Balance can be achieved by removing
material
u
m at
u
r or by adding material diametrically opposite. Note that correcting
unbalance
u u
m r can be obtained by choosing a convenient
u
r and adjusting
u
m . The disc can
be rolled on knife edges but it is more accurate to spin it in the following trial-and-error
method.

Strobe light is fired at frequency of rotation (Fig. 5.3) at a time in the cycle governed by the
peak in the measured oscillatory vibration at the bearing, which in turn is influenced by the
circumferential location on unbalance on the disc.

Step 1: With the disc stationary, put O reference marks on both the disc and the
stationary support (Fig. 5.4(a)).

Step 2: Run unbalanced disc and measure:
magnitude of vibration
u
A

Fig. 5.3

Fig. 5.4 (from Rao).
5-3
angle between reference marks on disc and support.
Step 3: Add a known trial mass
t
m at known radius
t
r and known angle to reference
mark. Run disc and measure:
magnitude of vibration
u w
A
+

angle between reference marks on disc and support.

Step 4: Construct a vector diagram (Fig. 5.5).
Calculate from to know where to add balancing mass, as follows.
Cosine rule gives:
( )
1
2 2
2
2 cos
w w u u w u u w
A A A A A A
+ +
= = +

2 2 2
1
cos
2
u w u w
u w
A A A
A A
+
+
=

The magnitude of the original unbalance mass is
u
o t
w
A
m m
A
= . It is located at the same radial distance
t
r as the trial mass. The procedure can
be repeated to get a finer (more accurate) balance.

5.3 Dynamic or Two-Plane Balancing
Any unbalance mass in a long rotor is equivalent to two unbalance masses in any two planes.

For equivalence of forces,

2 2 2
1 2
m R m R m R = +

1 2
m m m = +
For equivalence of moments, sum moment about axis into plane of diagram through O.

2 2
1 1
3
3
l
m R m Rl m m = =

Fig. 5.5 (from Rao).

is equivalent to
O

5-4
Hence,

1 2
2
3 3
m m
m and m = =
This argument can be extended to show that a distribution of unbalance masses (distributed
along rotor and at different angular positions) is equivalent to unbalance masses in any two
planes and at generally different angular positions.

Now consider use of transducers on each bearing A and B, a vibration analyzer and a strobe
light. While stationary, put reference marks on each end of rotor and on stator.

Step 1: At operating speed, measure amplitude and phase due to original unbalance at
each bearing. Measurement of amplitude and phase at bearing A is in some way due to
equivalent unbalance at both left and right plane.

A AL L AR R
All unknown
V A U A U = +

(5.1)
Similarly

B BL L BR R
All unknown
V A U A U = +

(5.2)
Step 2: Add a known trial weight
L
W
in left plane at a known angular position and

measure displacement and phase at the two bearings.
( )
A AL L L AR R
V A U W A U = + +

(5.3)
( )
B BL L L BR R
V A U W A U = + +

(5.4)
(5.3) (5.1)
A A
AL
L
V V
A
W

=

(5.5)
(5.4) 5.2)
B B
BL
L
V V
A
W

=

(5.6)
Step 3: Remove
L
W
and add known

R
W
.
Measure
( )
A AR R R AL L
V A U W A U = + +

(5.7)

5-5
( )
B BR R R BL L
V A U W A U = + +

(5.8)
(5.7) (5.1)
A A
AR
R
V V
A
W
(5.9)
(5.8) (5.2)
B B
BR
R
V V
A
W

=

(5.10)
Having calculated
AL
A
,
BL
A
,
AR
A
and
BR
A
we can substitute into (5.1) and (5.2) to find the

original unbalances;

BR A AR B
L
BR AL AR BL
A V A V
U
A A A A

(5.11)

BL A AL B
R
BL AR AL BR
A V A V
U
A A A A

(5.12)
The rotor can be balanced by adding equal and opposite balancing weights in each plane.

Useful Vector Algebra:
If;

A
A a =
and
B
B b =

1 2
a ia = +
1 2
b ib = + 1 i =
where;

1
cos
A
a a =
1
cos
B
b b =

2
sin
A
a a =
2
sin
B
b b =
Then;
( ) ( )
1 1 2 2
A B a b i a b = +

( ) ( )
( )
1 1 2 2 2 1 1 2
2 2
1 2
a b a b i a b a b A
B b b
+ +
=
+

( ) ( )
1 1 2 2 2 1 1 2
A B a b a b i a b a b = + + i

5.4 Balancing of Flexible Rotors
If the flexibility of a rotor is significant, 2-plane balancing can be ineffective and balancing
must be applied to three or more planes. This is a specialised procedure. It is visited in
Section 8 on generators and addressed in detail in Harris and Piersol (2002).
MECH7350 Rotating Machinery 6. Steam Turbines
6-1
6. STEAM TURBINES

6.1 Turbine Types
Large steam turbines are all of the axial-flow type
(Fig. 6.1). They may use single flow, double flow or
reversed flow (Fig. 6.2, where blades are not shown).
Double flow avoids excessively long blades and can
reduce axial thrust. Steam enters and leaves cylinder
radially, so design must leave space for flow to turn to
axial direction with minimum losses.

The limit of a single-cylinder turbine is about 100
MW. Multi-cylinder designs are used in large plant,
e.g. one high pressure (HP) turbine, one intermediate
pressure (IP) turbine and two low pressure (LP)
turbines (Figs 6.3 and 6.4 show various multi-cylinder
turbine arrangements). The IP and LP turbines are
usually double flow.

Cross compound machines avoid long shafts and can
enable fewer LP turbines if LP turbine shafts are run at
different speeds. Mainly used with 60Hz grid
frequency.

6.2 Speed of Rotation
Speed of shaft rotation is f = pn
f = grid frequency (Hz)
p = number of generator pole pairs
n = rotational speed (Hz)
Machine type Rotational speed (rpm)
Two-pole (full-speed) 3000 (50Hz)
Four-pole (half-speed) 1500 (25Hz)
Turbines to drive boiler feed pumps operate at variable speeds, as high as 8500 rpm, to
accommodate the operational range of the driven machine.

Fig.6.1 Axial-flow turbine (from MPSP).

Fig.6.2 Direction of flow in turbines
(from MPSP).
6-2

Fig. 6.3 Multi-cylinder turbine arrangements (from MPSP).
6-3

Fig. 6.4 Tandem-compound and cross-compound machines (from MPSP).
6-4
6.3 Turbine Stages
An impulse stage consists of stationary blades forming nozzles through which the steam
expands, increasing velocity as a result of decreasing pressure. The steam then strikes the
rotating blades and performs work on them, which in turn decreases the velocity (kinetic
energy) of the steam. The stream then passes through another set of stationary blades which
turn it back to the original direction and increases the velocity again though nozzle action.

Ideal reaction stages would consist of rotating nozzles with stationary blades (buckets) to
redirect the flow for the next set of rotating nozzles. The expansion in the rotating blades
causes a pressure force (reaction) on them that drives them. However, it is impractical to
admit steam to rotating nozzles. The expansion of steam in the stationary nozzles of a
practical reaction turbine is an impulse action. Therefore, the reaction stage in actual turbine
actions is a combination if impulse and reaction principles.

A reaction stage has a higher blade aerodynamic efficiency than an impulse stage, but tip
leakage losses are higher because of the pressure drop across the rotating stage. This is
significant for short blades (HP) but becomes insignificant for long blades (LP).

Modern turbines are neither purely impulse nor purely
reaction. They are a combination of both, with a
highly twisted profile so that the inlet and outlet angles
conform to the three-dimensional flow characteristics
at all blade heights, e.g. Fig. 6.5.

Blade efficiencies are not ideal. Profile loss is due to formation of a boundary layer on the
blade surface. Secondary loss is due to friction on the casing wall and on the blade root. It is
a boundary layer effect. Tip leakage is due to steam passing through the necessary small
clearance between the moving blade tip and the casing, or between the end of the fixed blades
and the rotating shaft.

A shroud band extends around the entire circumference of the moving blades, joining the tips.
The shroud is sealed against the casing by several knife edges (Fig. 6.6).

Fig. 6.5 LP last stage moving blade
(from MPSP).
6-5

The long fixed blades of an LP cylinder are stiffened towards their tips with lacing wires.
These damp vibrations and raise the resonant frequency of the blade so that it does not
coincide with any exciting frequencies. Lacing wires cause some aerodynamic losses.

Wetness loss is associated with moisture droplets entrained in the low pressure steam near the
exit plane of the LP turbines. Droplets absorb energy and can erode the leading edges of
moving blades.

6.4 Turbine Blading
6.4.1 Impulse Stages
In impulse stage, most of the heat drop occurs in the
stationary blading (acting as nozzles to increase
velocity and kinetic energy). Driving force arises
from change in momentum (direction) across the
moving blades.
6.4.1.1 Moving Blades
Moving blades are subject to turbulent disturbances
from the nozzle wakes formed by the fixed blades, so
resonances must be avoided. Also, momentum
change across moving blades is high, so moving

Fig. 6.6 Reaction and impulse turbine interstage sealing (from
MPSP).

Fig.6.7 Modern impulse stage with
labyrinth sealing (from MPSP).
6-6
blades must be robust, and tend to be heavy. Moving blades are manufactured individually
and attached to a wheel disc which is part of the rotating shaft. Moving blade tips are riveted
to a coverband (or shroud band) which acts as a labyrinth seal and braces the moving blades
to reduce vibration (Fig. 6.7).
6.4.1.2 Fixed Blades
Fixed blades (Fig. 6.7) are manufactured by:
Machining from a solid disc;
Casting steel blades in to a cast iron diaphragm, and then machining; or
Electrochemical machining from solid.
Both the diaphragm and the blades need to be robust to withstand the large stage pressure
drag. Good labyrinth sealing is needed between the diaphragm and the rotor, again because
of the large pressure drop.

6.4.2 Reaction Stages
These stages employ equal degrees of impulse and reaction, so fixed and moving blades are
of identical section. This leads to economies of manufacture. Driving force arises from the
reaction force of the steam as it accelerates through the moving blades. Nozzle wakes from
the fixed blades are not strong so there is lower risk of fatigue failure due to vibration in the
moving blades. The pressure drop across fixed blades is small so a diaphragm is not needed.
Modern practice is to braze together short groups of blades before machining the
circumferential serrations (fir-trees) on the sides of the roots (Figs 6.8 and 6.9).

Fig. 6.8 Section through a reaction
stage (from MPSP).
Fig. 6.9 Brazed reaction blade group (from
MPSP).
6-7
Low Pressure Stages
Blades of up to one metre long can be used. A coverband or lacing wire must behave as a
beam spanning the blade pitch in resisting centrifugal loading, and must accommodate the
substantial circumferential strains due to elastic extension of the blades and the tendency of
the blades to untwist at speed. When lacing wires are used, they are usually of the loose
type with circumferential restraint on only one blade in each group, and are free to move
circumferentially in adjacent blades, centrifugal forces providing the necessary damping
through friction.

A coverband of conventional design is not feasible for slim sections and where the peripheral
speed might be approaching Mach 2, but a continuous ring of stiffening devices of sufficient
elasticity may be used to accommodate the circumferential strains. The elastic arch banding
shown in Fig. 6.10 braces the blade tips and provides some resistance to blade untwist as well
as permitting circumferential strain.

Zigzag spool rods are sometimes used at the tips of last-stage LP blades (Fig. 6.11). They
provide no restraint against circumferential expansion or centrifugal untwist, but the reduced
sections at the ends of the rods are forced against the holes in the blades by centrifugal action

Fig. 6.10 Arch coverbands (from MPSP).
6-8
and the sliding friction provides effective damping, minimising blade vibration and high
frequency flutter at the blade tip.

Fig. 6.11 Zigzag spool rod tip-ties (from MPSP).
6-9

6.4.4 Moving Blade Root Attachments
Last stage blades develop centrifugal forces of hundreds of
tonnes when running. Strong methods of attachment are
needed. Fir-tree roots are widely used (Fig. 6.12). There is
some looseness in fir-tree roots for assembly but this
becomes rigid when the blades rotate. However, it is not
possible to measure the zero-speed vibration characteristics
of the blades. Pinned roots overcome this but blade
replacement is not easy.

6.4.5 Blade Materials
Blade material must have some or all of the following properties, depending on the position
and role.
corrosion resistance (especially in the wet LP stage)
tensile strength (to resist centrifugal and bending stresses)
ductility (to accommodate stress peaks and stress concentrations)
impact strength (to resist water slugs)
material damping (to reduce vibration stresses)
creep resistance
12% Cr stainless steels are a widely used material. Their weakness is at very high
temperatures (> 480C). A typical high temperature steel is 12% Cr alloyed with
molybdenum and vanadium (to 650C).

Titanium has some attractions but it is expensive and material damping is low. It has poor
vibration characteristics. Because of its high strength/weight ratio, titanium is used in lacing
wire and for coverbands and shrouding.

6.4.6 Blade Vibration Control
Blade vibration characteristics under operating conditions are very complex and difficult to
predict by calculation such as finite element analysis because:
an individual blade has a very complex geometry

Fig. 6.12 Types of fir-tree roots (from
MPSP).
6-10
there are vibration interactions among the blades through the blade disc,
diaphragms, coverbands and lacing wires.
The vibration of a fully-bladed disc is much more complicated than is suggested by the
characteristics of a single cantilever blade. There is a multiplicity of modes of vibration in
the turbine working frequency range. For a single blade, there are only two or three.
Sources of vibration excitation are:
Non-uniform flow caused by:
o steam entering over only a portion of the circumference
o complex axial to radial flow behaviour (which is minimised with good design)
o flow distortion caused by steam extraction passages for feedheater tappings
Periodic effects due to manufacturing constraints, e.g.
o Inexact matching at fir-tree roots
o Eccentricity of diaphragms
o Ellipticity of stationary parts
o Non-uniformity of manufacturing thicknesses
o Moisture removal slots
All of the above sources cause excitation at the
rotation frequency or low multiples (harmonics) of
that frequency. Recall the Fourier content of a
non-sinusoidal periodic wave.

Nozzle wake excitation as a rotating blade passes a stationary blade.
Excitation frequency = rotational frequency number of stationary blades and its
multiples.

Some sources can cause excitation at frequencies that are unrelated to rotational frequency,
acoustic resonances in inlet passages, extraction lines and other cavities
vortex shedding from bluff bodies

time
Contains fundamental and harmonics

flow
Vortex street
6-11
unsteady flow separation from stationary blades.

unsteady shock waves in blade passages
surface pressure fluctuations from impingement of turbulent flow.

A mode will only be excited if:
the excitation frequency coincides with the resonant frequency of the mode; and
the loading has the necessary component of spatial distribution.

6.4.7 Vibration Testing of Blades
Predictions from a finite element analysis (resonant frequencies and mode shapes) are
verified against measurements with piezoelectric accelerometers on a single cantilevered
blade. If agreement is good, predictions for a fully bladed disc are compared with
experiments.

For short blade discs (all but the last LP stage) static testing is good enough to verify that all
natural frequencies are above the 8th engine order, i.e. 8 50 Hz = 400 Hz. This allows for
up to the 8th harmonic in the excitation.

Flow
Separated flow which might attach and separate in an
unsteady manner

mode shape mode shape
This spatial distribution of
fluctuating load will not excite the
mode
This spatial distribution of
fluctuating load will excite the mode

6-12
For the larger LP blades, natural frequencies are lower and may coincide with harmonics of
the rotation frequency below the 8th order. Testing must be at running speed because
centrifugal effects can change the stiffness of long blades. Testing is conducted at speed in a
vacuum wheel chamber. The presence of air would necessitate a huge amount of power to a
large electric motor to drive the disc. Windage near blade tips would cause overheating and
make results difficult to interpret.

The disc is run up to 115% of synchronous speed and blade vibration is detected with strain
gauges of piezoelectric crystals. A Campbell Diagram can be developed (Fig. 6.13).

Where lines cross there is a prospect of resonance in service. It is usual to confine attention
to 6% of synchronous speed (2820 to 3180 rpm). A common specification is that in this
range there be no resonances up to 8th order.

Problem modes can be tuned out by adjusting the blade mass near the tip, or by adding or
removing mass to or from the shrouding.

Fig. 6.13 Campbell diagram (from MPSP).
6-13
6.4.8 Blade Erosion
Water droplets in the last stages of a turbine can cause erosion at the leading edge of moving
blades, and cracks can form. Leading edges can be protected by surface hardening or by
welding a shield of hard material such as tungsten chromium tool steel or satellite (an alloy of
cobalt and chromium). Shields will probably need to be replaced once during the lifetime of
the turbine.

6.5 Turbine Casings
A turbine casing (cylinder) is a high pressure vessel with its weight supported at each end on
the horizontal centreline. It is designed to withstand hoop stresses in the transverse plane and
to be stiff in the longitudinal direction to maintain accurate clearances between the stationary
and rotating parts.

Casings are split along the horizontal centreline to allow internal access and insertion of the
rotor as a complete assembly.

High pressures necessitate very thick flanges and bolting. The temperature of these changes
more slowly than the rest of the casing during start-up so a flange warming system is used.

HP and IP casings are cast. LP casings can contain some fabrication. Casings are tested to
150% of highest working pressure.

6.5.1 High Pressure Casings
Cross-sections of a single-flow HP casing are shown in Figs 6.14 and 6.15. HP casings are
usually of a double shell design. The space between the shells is filled with steam at exhaust
conditions. Then each casing can be designed for smaller temperature and pressure
differentials. Some exhaust steam leaks past a baffle to fill the space between the shells. The
rotor is protected from high pressure steam at the inlet by a deflector ring. Steam leaking past
the gland at the HP end is piped to exhaust connections, so there is only a gentle flow
between the casings.

Triple casings are used in some machines to further reduce temperature and pressure
differentials.
6-14

?.2 Intermediate Pressure Casings
IP casings usually have a partial double-casing covering the first few stages. Subsequent
stages are supported by carrier rings.

Fig 1.94

Both the inner casing and the carrier rings reduce pressure and temperature loading on the
outer casing. IP cylinders are usually double flow.

Cooling flow through the outer casing can be achieved by slight differences in the blading in
the two flows to produce different pressures at each end. Fig. ?.? is a photo of HP and IP
cylinders.

Fig. 1.97

?.3 Low Pressure Casings
LP cylinders often have double casings and are usually double flow. Fig. ?.? shows a typical
cross-section.

6.5.2 Intermediate Pressure Casings
IP casings usually have a partial double-casing covering the first few stages. Subsequent
stages are supported by carrier rings (Fig. 6.16).

Both the inner casing and the carrier rings reduce pressure and temperature loading on the
outer casing. IP cylinders are usually double flow. Cooling flow through the outer casing
can be achieved by slight differences in the blading in the two flows to produce different
pressures at each end. Fig. 6.17 is a photo of typical HP and IP cylinders.

Fig. 6.14 Axial section of HP cylinder (from MPSP).
6-15

Fig. 6.15 Cross-section of HP cylinder (from MPSP).

Fig. 6.16 Axial section of IP turbine casing (from MPSP).
6-16

Fig. 6.17 HP and IP casings (from MPSP).
6-17
6.5.3 Low Pressure Casings
LP cylinders often have double casings and are usually double-flow. Fig. 6.18 shows a
typical cross-section.

6.6 Couplings
Shaft couplings are needed between the various stages of a turbine/generator set. Ideally,
couplings should:
Transmit torque
Allow angular misalignment
Transmit axial thrust

Fig. 6.18 Axial section of LP casing (from MPSP).
6-18
Ensure axial location or allow relative axial movement
Provide torsional resilience
Flexible or semi-flexible couplings can provide this but they are impracticable on large
turbines because of the high torque to be transmitted.

Rigid couplings are used in large turbines so that the joined shafts can behave as one
continuous rotor. They are either integral with the shaft forging (Fig 6.19) or shrunk on to
the shaft (Fig. 6.20). In the latter case, high pressure oil can be injected into annular grooves
to ensure correct seating during assembly, or to aid removal.

Couplings are designed to withstand a three-phase fault or out-of-phase synchronising
without damage (4-5 times full load torque).

6.19 Rigid forged coupling (from MPSP).
6-19

6.6.1 Rotor Alignment
Excessive misalignment of a multi-bearing shaft line can affect the vibration behaviour.
It causes bending moments at couplings which act like a rotating out-of-balance.
It can cause bearing unloading which alters shaft vibration behaviour.

A long shaft bends naturally under its own weight to form a catenary (Fig. 6.21) and revolve
around a curved centreline. The shape of the catenary depends on the masses and stiffnesses
of the rotors.
The aim of alignment is to ensure insignificant bending moments and shear at the couplings.
Bearing heights are adjusted so that coupling faces are square to each other, with centrelines
coincident and with the same slope where the faces meet. This is done by slightly separating
Fig. 6.20 Shrunk-on coupling (from MPSP).
6-20
the coupling and turning the rotor to different positions. Bearings are adjusted to get uniform
gap and concentricity (measured with a dial gauge).

Bending moment cyclic variations can be measured with strain gauges and optical
techniques. Lasers are used to set the catenary up initially, prior to adjusting it.

Outer bearings may be 25 mm above the level of central bearings. Changes in service to
pedestal bearings are monitored on-line with a manometric system.

6.7 Journal Bearings
Bearings on the shaft line of a large turbine/generator set are invariably white-metalled
journal bearings because of their:
high load capacity
reliability
absence of wear through use of hydrodynamically generated films of lubricating
oil (no metal-to-metal contact)

Turbine bearings have diameters up to 550 mm, with length/diameter (L/D) ratios of 0.5 to
0.7. Generator bearings have L/D ratios of 0.6 to 1.0 because of the weight of the generator
rotor. They are split in halves for assembly of the rotor, with bolts and local dowels (Fig.

Fig. 6.21 Shaft catenary for a large turbine-generator (from MPSP).
6-21
6.22). White metal is either cast into a mild steel liner or cast into the bearing body. The
bearing body is spherically seated into the pedestal for angular alignment. Shims are
available for vertical and horizontal alignment.

Fig. 6.22 Journal bearing (from MPSP).
6-22
White metal (or babbit) is usually composed of 80 to 90 % tin to which is added about 3 to
8% copper and 4 to 14% antimony. These alloys have very little tendency to cause wear to
their steel journals because of their ability to embed dirt. They are easily bonded, cast and
shaped, and can have good load-carrying and fatigue properties.

The bores of journal bearings are usually elliptical to provide the geometry for hydrodynamic
lubrication. A circular bore is machined with shims in the horizontal split. The shims are
removed in assembly to give typical diametrical clearance/diameter ratios of 0.001 vertically
and 0.00015 horizontally. Oil is fed into the bearing via lead-in ports at two diametrically
opposite points on the horizontal centreline. This is to cool and lubricate the bearings and
comes from the main turbine lubricating-oil pump.

Each bearing also has a high pressure jacking oil supply at the bottom. This lifts the shaft
when starting from rest, until speed is high enough for hydrodynamic lubrication to start-up.

Instrumentation at each bearing normally gives:
white metal temperature
lubricating oil outlet temperature and inlet pressure
jacking oil pressure
vertical and horizontal vibration

6.7.1 Hydrodynamic Lubrication
(This section is taken from Williams)

Hydrodynamic bearings depend on the presence of a converging, wedge-shaped gap into
which a viscous fluid is dragged by the relative motion of two surfaces. A pressure is
generated which tends to push the surfaces apart. This balances the load on the bearing.

Large rotating machinery utilises hydrodynamic journal and thrust bearings. An analytical
solution of their behaviour is complicated but the elements of behaviour can be understood by
studying the simple two-dimensional pad bearing in Fig. 6.23.

6-23

The bearing is long in the y-direction, so that there is no fluid flow normal to the plane of the
paper. The upper, inclined fixed member is of length B, while the lower flat slider moves
from left to right with velocity U. Fig. 6.23 also shows the pressure distribution in the
viscous fluid. The integral of this pressure distribution supports W/L, the load per unit length
into the page.

The angle of the wedge is greatly exaggerated in Fig. 6.23. It is typically only a quarter of a
degree.

Consider the equilibrium of the small element of fluid within the gap in Fig. 6.23. The local
film thickness is h and it varies in a known way from
i
h at the entry to
o
h at the exit. Gravity
and inertia (acceleration) forces can be neglected.
Then

p
x z z x
x z

=

where

Fig. 6.23 Two-dimensional bearing pad (from Williams).
6-24
p = pressure and = shear stress.
Hence
p
x z

=

.
But
u
z

=

where = Newtonian viscosity
u = local fluid velocity in x-direction.
So
2
2
p u
x z

=

But h B , so we can take p = constant across the film thickness, and p is a function of x
only.
Therefore
2
2
dp u
dx z
.
This can be integrated twice to get

2
2
dp z
u Az C
dx
= + + where A and C are constants of integration.
Setting boundary conditions, u U = at 0 z = and 0 u = at z h = enables evaluation of the
constants of integration and leads to:
( )
1
1
2
dp z
u z z h U
dx h

= +

(6.1)
The volumetric flow rate q through a unit width (in the Oz-direction) can be obtained by
integration of (6.1) across the film.

3
0
12 2
h
h dp Uh
q udz
dx
= = +
(6.2)
This is independent of x for an incompressible fluid.
Let h h = where 0.
dp
dx
=
Then
2
Uh
q Uh = = where / 2 U U = (6.3)
Combining (6.2) and (6.3) gives

3
12
dp h h
U
dx h

= (6.4)
This is the simplest form of Reynolds equation. If h is known as a function of x, (6.4) can be
integrated to get p(x) and the load carrying capability of the bearing. Fig. 6.24 shows the
6-25
geometry of a journal bearing and the exaggerated wedge of fluid. Analysis of this
geometry is complicated by the finite length of a journal bearing and the flow out of its ends.

Fig. 6.25 also shows the configuration of
an ideal steady state hydrodynamic film.
If a vibratory load (e.g. due to rotating
unbalance) is superimposed on the
steady load (weight) the oil thickness
can change and move around the
circumference. This can lead to whirling
of the journal (shaft) and affect the
vibratory behaviour of the whole rotor
line.

Fig. 6.24 Geometry of journal bearing (from Williams).

Fig. 6.25 Possible oil fil
configurations (from MPSP).
6-26
6.7.2 Hydrodynamic Thrust Bearings
By pivoting the fixed part of the bearing in Fig. 6.23, the angle of tilt will vary with load so
that it is at the optimum for load-carrying capacity. This was the discovery of the Australian,
A.G.M. Michell, and led to his famous thrust bearing design which is used in
turbine/generator shaft lines. This is illustrated in Fig. 6.26.

A turbine thrust bearing is used to provide axial location for the turbine rotors relative to the
cylinders. Because solid couplings are used, only one thrust bearing is used in the shaft line.
It is usually located near areas where blade/cylinder clearances are a minimum. It is in two
halves for ease of assembly.

Thrust bearings are of the Michell tilting pad design (Fig 6.27).

Fig. 2.26 Thrust bearing configuration (from Stachowiak and Batchelor).
6-27

Fig. 6.27 Mitchell thrust bearing (from MPSP).
6-28
In normal operating conditions, net thrust is always in the one direction. However, a set of
surge pads is incorporated on the opposite side of the shaft collar to take account of transient
reversals of thrust which can occur during load reduction and a turbine trip.

6.8 Pedestals
Pedestals support the turbine rotor via the journal bearings in a fixed axial relationship with
each cylinder so that gland clearances are maintained. They are usually fabricated in steel
and stiffened by ribs and gussets plates.

In the LP area, pedestals are normally bolted and dowelled to the foundations. At the HP
end, provision is made for the cylinders to expand by way of sliding points at the top or
bottom of the pedestals.

6.9 Turning Gear
Turbine rotors must be turned continuously during warming up and cooling down processes.
Otherwise distortion can occur. Usually there are two independent turning gear systems
provided:
A hand barring arrangement. A lever operates on a toothed wheel between two
turbine rotors.
Electric turning gear. An electric motor turns the rotors through a wormshaft and
a wormwheel at less than 30 rpm.

6.10 Turbine Rotors
The shaft of each turbine rotor is a single, high quality alloy steel forging, machined to
provide the required contours and functioning parts. Each end contains an integral coupling,
gland seal area and bearing area. For HP and IP reaction turbines, axial grooves are
machined into the rotor for the blades. For impulse HP and IP turbines and for LP turbines,
wheels are machined or shrink-fitted onto the rotor with the blades mounted in grooves in the
wheels.

The rotors of HP and IP turbines may have a centre bore machined in the shaft to remove
impurities formed during the forging, and to allow access for ultrasonic inspection.

6-29
Alloy steels are chosen to have good creep resistance and high temperature and high fracture
toughness.

6.10.1 Overspeed Testing
A 20% proof overspeed test is specified on all large turbine/generator rotors at the time of
manufacture. This tests the forging against spontaneous fast fracture and confirms its
balance.

6.10.2 Rotor Balancing
With the blade discs assembled, the rotor is balanced both statically and dynamically. Each
blade disc is balanced individually before assembly. Rotors are dynamically balanced at low
speed (400 rpm) with weight adjustments made in two planes, one at each end of the rotor.
Provision is made to vary screwed plugs in tapped holes, or to add weights. The aim is to get
25 m < amplitude of vibration at the bearing pedestals.

Modal behaviour must be understood for long rotors. These rotors are often balanced at
running speeds and critical speeds in a vacuum chamber. When rotor flexibility is important,
balancing is done at three or more planes.

On-site vibration testing can be done but it is affected by variations in the stiffness of the
bearings, possible shaft misalignment and the coupling of the individually balanced rotors to
form the complete shaft system. Access holes are provided in the casing.

6.10.3 Critical Speeds
A stationary shaft and rotor between bearings has a
natural frequency of vibration with the shaft in
bending. If the speed of rotation coincides with this
natural frequency, any small unbalance can cause
dangerous vibrations. This is a critical speed.

If the critical speed is below the running speed, the shaft is regarded as flexible. Care is
needed to run up through this critical speed quickly. Modern units have rigid shafts, with

Rotor
Shaft Bearing

6-30
critical speed above operating speed. With the long shafts in large units, large shaft
diameters are needed.

However, each turbine does not act independently of others. There might be up to seven
individual rotors in a shaft line. The bearings are hydrodynamic and so have flexibility which
might increase with wear. As a result, there are then a number of critical speeds, two or three
of which can be below the operating speed. A typical speed-vibration curve from an
instrumented bearing housing is shown in Fig. 6.28.

Fig. 6.28 Typical speed-vibration curve at a pedestal (from MPSP).
MECH7350 Rotating Machinery 7. Gas Turbines
7-1
7. GAS TURBINES

(This section is based largely on Black and Veatch)

Gas turbines are only briefly covered in these notes because (a) many of their mechanical
fundamentals are similar to those for steam turbines, and (b) they are the topic of a guest
lecturer.

Gas turbine technology is used in a
variety of configurations for electric
power generation. Conventional
applications in power stations are
simple cycle and combined cycle.
Simple cycle operation is used
primarily for peaking power
generation. Smaller units (about 15
MW) are used for black starts.
Fig 7.1 is a schematic of simple
cycle operation.

Combined cycles combine the gas turbine and steam turbine cycles into more efficient power
plants by utilising the gas turbine exhaust gas heat. This is shown schematically in Fig. 7.2.

Fig. 7.1 Simple gas turbine cycle (from Black and Veatch).

Fig. 7.2 Combined cycle gas turbine (from Black and Veatch).
7-2
Gas turbine applications generally rely on natural gas, with its environmental benefits, or fuel
oil for fuel. However, future power plants can be expected to use the integrated gasification
combined cycle (IGCC) in which coal is partially combusted in oxygen to produce syngas,
which in turn is burned in a combined cycle process to produce electricity.

7.1 Gas Turbine Components
The main components of a gas turbine (Fig. 7.3) are:
Inlet air system
Compressor
Combustion systems
Turbine
Exhaust system
generator

When the gas turbine is started, ambient air is drawn through the inlet air system, where it is
filtered and then directed to the inlet of the compressor where it is compressed and directed to
the combustion system. Inside the combustion system the air is mixed with fuel and the
mixture is ignited by a spark plug ignition system. The compressed and heated combustion
gases then flow to the turbine, expanding and causing it to rotate. The rotating turbine drives
Fig. 7.3 Major sections of a gas turbine assembly (from Black and Veatch).

7-3
the compressor and accessory equipment, such as the main lube oil pump. The number of
stages within the compressor and turbine may vary.

After leaving the last stage of the turbine, the exhaust gases are either released to the
atmosphere or directed through an exhaust system to heat recovery equipment.

7.2 Multi- and Single-Shaft Plants
In a multi-shaft combined-cycle plant, there are generally several gas turbines with heat
recovery generators producing steam for a single steam turbine. The steam and gas turbines
use different shafts and generators. With the largest gas turbines on the market, one steam
turbine per gas turbine or one steam turbine for two gas turbines is common.

If one steam turbine per gas turbine is installed, the single-shaft application is most common
gas turbine and steam turbine driving the same generator. A plant with two gas turbines
can be built either in a multi-shaft or a single-shaft configuration.

7.3 Sequential Combustion
In a gas turbine with sequential combustion, air enters the first combustion chamber after the
compressor. There, fuel is burned, raising the gas temperature to the inlet temperature for the
first turbine. The hot gas expands through this turbine stage, generating power before
entering the second combustion chamber where additional fuel is burned to reach the inlet
temperature for the second part of the turbine. There, the hot gas is expanded to atmospheric
pressure. Sequential combustion increases efficiency.

7.4 Materials
Gas and steam turbines experience similar problems. However, the magnitude of these
problems is different, leading to more demands on the materials in gas turbines. In modern,
high performance, long-life gas turbines, the critical components are the combustor liner and
the turbine blades.

Creep and corrosion are the primary failure mechanisms in a gas turbine blade, followed by
thermal fatigue. The first-stage blades must withstand the most severe conditions of
temperature, stress and environment. Temperatures can be as high as 1500C. Nickel-base
alloys are widely used with coatings to protect against hot corrosion.
7-4
7.5 An Example
An example of a large gas turbine used in power generation is the Alstom Model GT26
shown in Fig. 7.4. It has the following features:
Sequential combustion
Turbine speed of 3,000 rpm (50 Hz)
Gross electrical output of 288 MW
Natural gas primary fuel, with fuel oil as a backup
Blade cooling with air extracted from the compressor
Hydrodynamic journal and tilting-pad thrust bearings
Turbine outer casing and compressor casing split horizontally

Fig. 7.4 Alston GT26 gas turbine.
MECH7350 Rotating Machinery 8. Generators
8-1
8. GENERATORS

Fig. 8.1 is a reminder of the components of a turbine/generator system. Here it includes the
exciter which provides DC current to the rotor of the generator.

8.1 Synchronous Generator Theory
Synchronous means that the generators rotor runs at the constant mains frequency as load
varies.
8.1.1 Electromagnetic Induction
The instantaneous voltage induced in a stator conductor is given by;

dB
e kl
dt
=
where;
k = constant
l = length of conductor

dB
dt
= rate of change of magnetic flux density
If e is to be a sinusoid, dB/dt must also be sinusoidal. Excitation coils are wound on the rotor
to produce a flux with a density which varies approximately sinusoidally around the
circumference. A two-pole arrangement (one pole pair) is shown in Fig. 8.2. The magnitude

Fig. 8.1 Components of a turbine-generator system (from Black and Veatch).
8-2

of the flux density can be changed by varying the direct current to the excitation coils on the
rotor.

Speed, Frequency and Pole Pairs
If f = frequency, Hz
n = rotational speed, r/s
p = number of pole pairs
Then f = pn
In Australia, generators are 2-pole (i.e. p = 1). Hence generators run at 3000 rpm.
Three-phase Windings
Fig. 8.3 shows how three phases are generated by having three winding spaces on the stator.
This is still a 2-pole generator. It is economical to have many stator conductors in parallel so
the individual conductor voltages are additive. Each go conductor is connected to a return
conductor, acted on by the pole of opposite polarity, and thence to a third conductor adjacent
to the first, and so on through the phase. The return conductors are disposed in a layer
displaced radially from the go conductor, both in the slots and in the end region. The
discrete nature of the windings gives rise to generation of harmonics.

Fig. 8.2 Production of sinusoidal
voltage (from MPSP).
8-3

Torque
The mechanical torque provided by the turbine is balanced by an electromagnetic torque
caused by the interaction of the magnetic flux and the current flowing in the stator windings.

8.2 The Rotor
(This section is mainly from MPSP)
The rotor must:
carry the excitation windings
provide a low reluctance path for the magnetic flux
transfer the rated torque from the turbine to the electromagnetic reaction at the air
gap
resist large centrifugal forces.
Steel is the most suitable material. A steel
forging is used with machined slots (Fig. 8.4).

Fig. 8.3 Arrangement of stator
conductors (from MPSP).

Fig. 8.4 Rotor section (from MPSP)
8-4
The winding slots are cut in diametrically opposite pairs over about two-thirds of the
circumference, leaving the pole faces without winding slots (Fig. 8.6). Equalising slots are
cut across the pole faces to avoid a difference in stiffness in two orthogonal directions which
would produce a twice-per-revolution vibration.

The rotor shaft and the coupling must be designed to withstand four to five times full-load
torque in the event of a sudden short circuit.

Journal bearings are used.

8.2.1 Rotor Winding
Full detail of the winding of a rotor is shown in Fig. 8.6. Winding coils are assembled into
diametrically opposite pairs of rotor slots symmetrically distributed about the pole axis, but in
the opposite senses in the two poles; i.e. clockwise current for the north pole and
counterclockwise current for the south pole.

Each coil must be assembled as a half-turn, with brazed joints in the end turns (end
windings). There are usually one or two turns in the width of a slot. Axial passages and
radial holes in the copper conductors provide for forced cooling with hydrogen. The high
rotational speed causes hydrogen to flow axially from both ends and radially outwards. Fans
mounted on the rotor, primarily to circulate hydrogen though the stator, help the flow through
the rotor (Fig. 8.7). Insulation is achieved with slot liners of glassfibre material

Fig. 8.5 Stiffness compensation (from MPSP).
8-5

Thick end rings are used to restrain the rotor end winding under the action of centrifugal
force. The ends of the windings are connected to flexible leads and there are radial copper
studs to connect to the sliprings and thence to the exciter. The contents of the winding slots
are retrained by an aluminium wedge (Figs 8.4 and 8.8).

Fig. 8.6 Rotor winding (from MPSP).

Fig. 8.7 Axial flow fans on rotor (from MPSP).

8-6

8.2.2 Sliprings, Brushgear and Shaft Earthing
For a 660 MW generator the excitation current is about 5000 A. This must flow through
sliprings with a large area (Figs 8.9 and 8.10). Brushes only last about six months but they
can be changed while running on-load.

In normal operation there is 10-50 volts between the two shaft ends of a generator, due
mainly to magnetic dissymmetry. To stop an axial current from flowing and damaging
bearings, an insulation barrier is provided at the exciter end.

Fig. 8.8 Rotor slot (from MPSP).
8-7

Fig. 8.9 Sliprings and connections (from MPSP).

Fig. 8.10 Slipring brushgear and brushes (from (MPSP).
8-8
It is important that the shaft at the turbine end of the generator is maintained at earth
potential, using a pair of shaft-riding brushes connected to earth through a resistor. The
voltage at the exciter end is monitored as an indication that insulation is intact (Fig. 8.11).

8.3 The Stator
(This section is mainly from Klempner and Kerszenbaum)
The stator core with windings are assembled into a skeletal core frame which is inserted into
a strong outer casing.

The stator core provides paths for the magnetic flux from one rotor pole around the outside of
the stator winding and back into the other pole. It is made up from tens of thousands of
electrical grade steel laminates, each about 0.4 mm thick. This prevents large circulating
eddy currents with their associated losses. Each lamination is insulated on both sides with a
very thin layer of an organic or inorganic compound. Winding slots, location notches and
holes for ventilation are cut in one pressing operation. The laminate segments are fitted onto
key bars in a stator frame structure and clamped axially (Fig. 8.12).

The core is cooled with hydrogen which passes axially through ducts cut radially in each
laminate.

Fig. 8.11 Shaft earthing and monitoring (from MPSP).

8-9
The stator outer casing provides support for the stator core, which can be 500 tonnes, and acts
as a pressure vessel in case of an explosion of the hydrogen cooling gas. Casings are
fabricated steel cylinders of up to 25 mm thickness and reinforced externally (Fig. 8.13).

Fig. 8.12 Core frame (from MPSP).

Fig. 8.13 Outer stator casing (from MPSP).

8-10

8.3.1 Stator Windings
Insulated copper bars are distributed around the inside of the stator core. To reduce eddy
current losses, bars are made up of numerous copper strands that, within the core, are
insulated from each other and Roebel Transposed so that each strand occupies each position
in the stack at least once over the full length of the stator bar. This reduces the effect of
circulating currents caused by all strands in a bar being electrically connected at the ends of
the core.

Figs 8.14 and 8.15 show stator end-windings which must be strongly braced to resist peak
forces and to minimise 100 Hz vibration.

Fig. 8.14 Stator end windings (from MPSP).

8-11

Stator cooling is achieved indirectly if the strands within the conductor bars are all solid and
the heat generated (I
2
R) is removed by conduction to the stator core.

In directly gas-cooled bars, hydrogen passes from end to end in cooling ducts.

In direct water-cooled bars, the copper strands are made hollow to carry demineralised water.
These cooling designs are shown in Fig. 8.16.

Hydrogen is used (rather than air) because:
its low density minimises fan and rotor windage losses
its heat transfer coefficient is 50% more effective than that of air at the same
pressure.

Hydrogen is contained between the casing and the rotor shaft with journal type seals which
use a flow of seal oil which is supplied at a pressure slightly higher than the hydrogen
pressure.

Fig. 8.15 Bracing of end-windings (from Klempner and Kerszenbaum).
8-12

8.3.2 Generator Auxiliary Systems
There are five main auxiliary systems:
Lubricating oil system
Hydrogen cooling system
Seal oil system
Stator cooling water system
Excitation system

8.3.3 Exciter Systems
The exciter system provides the DC current to the generator rotor windings. The system is
designed to control the applied voltage, and thus the field current to the rotor, which in turn
gives control of the generator output. DC currents can be as high as 8000 amps. There are
three types of excitation system;
Rotating
Static

Fig. 8.16 Stator conductor bar cross sections. (a) Indirectly cooled stator
conductor bar; (b) Directly gas-cooled stator conductor bar; (c) directly water-
cooled stator conductor bar (from Klempner and Kerszenbaum).
8-13
Brushless

Rotating
The exciter is connected to the free end of the generator shaft. The AC output of an
alternator (stationary armature and rotating field) is rectified by stationary diodes located off
the generator and fed to the main rotor sliprings as DC current.
Static
An external source of power is used to supply AC power to an excitation transformer whose
output goes to a rectifier bridge for conversion to DC.
Brushless
A high frequency (multi-pole) AC generator is attached to the generator shaft. The diode
rectifying components rotate on the shaft.

8.4 Generator Vibration Issues
(This section is largely from MPSP and Kempner and kerszenbaum.)

8.4.1 The Rotor
Rotor mechanical unbalance is caused by:
loss of a balance weight
oil whirl in a journal bearing
bearing loading
coupling alignment
misaligned hydrogen seal or oil wiper rubs
foundation resonance
a general rotor structural problem
cracked forging
shorted turns
thermal effects (e.g. rotor shorted turns, blocked ventilation duct).
The level of severity is usually determined by the magnitude of the vibration present, and
may require an outage to correct the source of vibration, or to apply balance weights.
Mechanical vibration problems are usually independent of load. Thermal unbalance usually
worsens with load.

8-14
Bearing and shaft vibration on both ends of the generator may be monitored in terms of
magnitude, phase and frequency at variable load conditions. Accelerometers and proximity
probes are used in two sets, set 90 degrees apart. Sophisticated vibration analysers are
available. Each manufacturer gives its own recommendation for alarm and trip. A typical
maximum allowable amplitude is 1 mm at 3000 rpm.

Generator rotors are relatively flexible and pass through two main critical speeds during run-
up to rated speed of 3000 rpm. Two-plane balancing is inadequate. Facilities for balancing
are provided along the length of the rotor in the form of tapered holes in the cylindrical
surface.
Imperfect equalisation of the stiffness of the rotor in two orthogonal directions (associated
with the creation of poles) will cause 100 Hz vibration to occur, superimposed on the normal
50Hz. It is important to distinguish between these two components. A significant crack in
the rotor will have a comparatively greater effect on the double frequency vibration
component; run-down traces are recorded and analysed to provide assurance that no
significant change has occurred since the previous run-down.

Oil whirl in bearings can cause vibration at 25 Hz.

The torsional resonance of the generator rotor coupled to the turbine rotors is at about 13 Hz.
It is important that this is significantly different from the frequency of torsional exciting
influences such as the steam governor control (1-2 Hz).

Transient oscillations in torque occur during electrical disturbances (switching operations,
lightning strikes, imperfect synchronising events). Some of the torque cycles may be large
enough to cause plastic deformation of the turbine-end shaft and the generator/excitor
coupling (if there is one).

8.4.2 Stator End-Winding Vibration
Vibration of the stator end-windings must be minimised. It can cause fatigue cracking in the
winding copper. This is a serious problem if cooling hydrogen can leak into the cooling
water circuit. Resonances close to 100 Hz must be avoided, since both the core ovalising and
the winding exciting force occur at this frequency. Accelerometers in the end-winding
structure allow any increase in vibration due to support slackening to be monitored.
8-15
Looseness can be corrected by tightening bolts or by inserting tightening wedges. If
permanent mounting of accelerometers is not possible, a bump test, or impact frequency
spectrum analysis is done with temporary vibration transducers and a calibrated impact
hammer. Vibration amplitude is highly dependent on current.

The general aim is to keep the maximum amplitude of vibration of stator end-windings to less
than 50 m peak-to-peak, with no natural frequencies within the ranges 40-65 Hz and 80-120
Hz.

8.4.3 Stator Core and Frame
The plus/minus magnetic field revolves. This alternating effect causes vibration of the core
at the rotational frequency and with harmonics, due to the nature of the flux patterns. In a
two-pole generator, the driving frequency is 50 Hz and there is a 100 Hz (twice per
revolution) component due to the four-node pattern of the flux. This can be seen in Fig. 8.17
where there are two areas of high flux density and two of minimum density at any given point
in time, as the flux patterns rotate at the rated speed. This causes the core to be distorted
minutely by the electromagnetic pull into an oval shape, in and out, in the radial direction.
The result is vibration of the core and subsequently the frame.

Fig. 8.17 Rotor flux pattern (from Kempner and Kerszenbaum).
8-16
Because of the inherent vibration and the large forces involved, the core must be held solidly
together such that there are no natural frequencies near the once and twice per revolution
forcing frequencies. Designers take care to ensure that the natural frequencies of the core are
not near 50 or 100 Hz. It is desirable to keep the natural frequencies at least 20% away
from the once and twice per revolution frequencies. Damped spring mounting of the whole
generator on its foundation might be needed, or spring mounting of the core in the casing.

In addition to the vibration due to the alternating flux, there is a large rotational torque
created by the electromagnetic coupling of the rotor and stator, across the airgap. This is in
the direction of rotor rotation. The torque due to the magnetic field in the stator iron is
transmitted to the core frame via the keybar structure at the core back. Therefore the stator
frame and foundation must be capable of withstanding this torque and large changes in torque
when there are transient upsets in the system or the machine.

Vibration in the stator core is naturally produced by the unbalanced magnetic pull in the
airgap, origination from the unequal magnetic field distribution of the rotor. The core must
be maintained tight or fretting will occur between the laminates. If the core becomes too
loose, the laminates and/or the space blocks might even fatigue, with loose core material
breaking off. Monitoring of core vibrations can be done with accelerometers mounted on the
core back in strategic locations.

Frame vibration is also excited by unbalanced magnetic pull and by any vibration produced in
the core. There are known results of vibration resonance occurring on the frame as a result of
the frame having a resonant frequency near line or twice line frequency. Severe damage to
the frame can occur by the initiation of cracks in the frame welds. Good core-to-frame
coupling is required to ensure that the core and the frame move together. Such vibrations
have been corrected by spring mounting of the core to the frame or installing a damping
arrangement to de-tune the vibration modes. Monitoring of frame vibration can be done with
accelerometers mounted on the keybars, frame ribs or casing structure.

MECH7350 Rotating Machinery 9. Pumps
9-1
9. PUMPS

9.1 Fundamentals of Fluid Mechanics
9.1.1 Bernoulli Equation
(This section is mainly from White)

We consider only two-dimensional, incompressible, frictionless (inviscid) flow.
Consider an elemental fixed streamtube control volume of variable area A(s) and length ds
(Fig. 9.1), where:
s = streamline direction
= fluid density (constant)
p = pressure
v = streamtube velocity
A(s) = streamtube cross-sectional area at s.

Conservation of mass gives:
0
out in
m m = because there can be no accumulation of mass in the control volume if
density is constant. Hence at any s, m Av = .

Now consider Newtons Second Law applied to fluid in the control volume. Sum elemental
forces in the streamwise direction.
( ) ( ) ( )
s
in out
F mv mv d mv = =
(9.1)
Neglect shear forces on the walls (inviscid flow) so the forces are due to pressure and gravity.

,
sin
s grav
dF dW = (z positive up)
Fig. 9.1 (from White).
9-2
= sin g Ads
= gAdz
= ( ) Adz g =
To get pressure force, imagine pressure p subtracted from all faces of the control volume.
Then:
( ) ( )
,
1
0
2
s press
dF dp A dA A A dpdA = + + +
= Adp to first order.
Substitute into (9.1).
( ) gAdz Ad d mv Avdv = =
Divide by A.
0
dp
vdv gdz
+ + =
This is Bernoullis equation for steady, frictionless flow along a streamline. Beware of its
limitations.
We can integrate between any two points 1 and 2 to get:

2 2 1 2
1 1 2 2
1 1
2 2
p p
v gz v gz

+ + = + + (9.2)
or

2 2
1 1 2 2
1 2
2 2
p v p v
z z
g g
+ + = + + (9.3)

Summary of assumptions
1. Steady flow
2. Incompressible flow
3. Frictionless flow
4. Flow along a single streamline: different streamlines may have different
Bernoulli constants
2
0
2
p v
h z
g
= + +
5. No shaft work between 1 and 2: no pumps or turbines on the streamline
6. No heat transfer between 1 and 2.

9-3
9.1.2 Hydraulic and Energy Grade Lines
The energy grade line (EGL) shows the height of the total Bernoulli constant
2
0
.
2
p v
h z
g
= + + Fig. 9.2 shows that this is constant along a streamline and can be measured
with a pitot tube which causes v = 0 at its inlet. Then the pressure at its location in the
streamtube is given by
0
/ p h z = , where z is the height of the streamtube above an
arbitrary datum.

The hydraulic grade line (HGL) shows the height corresponding to elevation and pressure
head / z p + , that is, the EGL minus the velocity head ( )
2
/ 2 v g . The HGL is the height to
which liquid would rise in a piezometer tube attached to the wall of the stream tube. In Fig.
9.2 the HGL is rising due to a decrease in velocity as the tube expands.

A pump causes a jump in the head (or the EGL) and (9.3) can be written across the jump as

2 2
2 2
pump
in out
p v p v
z h z
g g

+ + + = + +

If
in out
v v = because pipe diameters are the same and
in out
z z across a pump, then

out in
pump
p p
h
=

Fig. 9.2 Hydraulic and energy grade lines for frictionless flow in a duct (from White).
9-4

9.2 Positive-Displacement Pumps
Positive-displacement pumps (PDPs) force the fluid along by volume changes. Some types
are shown in Fig. 9.3.

Fig. 9.3 Positive-displacement pumps: (a) reciprocating piston, (b) external gear
pump, (c) double-screw pumd (d), sliding vane, (e) three-lobe pump, (f) double
circumferential piston, (g) flexible-tube squeegee (from White).
9-5
Advantages Disadvantages
Generate high pressures Low flow rates
Can handle high viscosities Pulsating flow
Self-priming

9.3 Rotodynamic Pumps
(This section is mainly from White)

Fig. 9.4 is a schematic of a typical centrifugal rotodynamic pump. An impeller rotates within
a casing. Fluid enters axially through the eye of the casing. It is caught up in the impeller
blades and is whirled tangentially and radially outward until it leaves through all
circumferential parts of the impeller into the diffuser part of the casing. The fluid gains both
velocity and pressure while passing through the impeller. The doughnut-shaped diffuser, or
scroll, section of the casing decelerates the flow through area increase, and further increases
the pressure. The diffuser may be vaneless or fitted with fixed vanes to help guide the flow
toward the exit.

Fig. 9.5 is a comparison of typical performance of rotodynamic and positive-displacement
pumps running at a constant speed.

Fig. 9.4 Centrifugal pump (from White).
9-6

A rotodynamic pump basically increases the Bernoulli head of the flow between point 1, the
eye, and point 2, the exit.

2 2
2 1
2 2
p v p v
H z z
g g g g

= + + + +

Usually
2 1
v v and
2 1
z z . Then

2 1
p p p
H
g g

=
The power delivered to the fluid equals the specific weight times the discharge (Q m
3
/s) times
the net head change.

w
P gQH =
This is called the water horsepower. The power required to drive the pump is the brake
horsepower.
bhp T =
where is the shaft angular velocity and T is the shaft torque. If there are no losses (100%
efficiency),

w
bhp P =
But in practice,
w
P bhp <

Fig. 9.5 Comparison of performance curves of typical rotodynamic and positive-
displacement pumps at constant speed (from White).
9-7
and the efficiency of pump =
w
P gQH
bhp T
=
This equation shows that and T applied to the pump can generate Q and H.

9.3.1 Elementary Pump Theory
Fig. 9.6 shows idealised velocity vector diagrams at inlet and outlet of impeller blades.
Backward curved blades are shown and are usually used for stable flow without pump surge.

Fluid enters the impeller at
1
r r = with velocity component w
1
(tangent to the blade angle
1
)
plus circumferential speed u
1
= r
1
matching the tip speed of the impeller. Its absolute
entrance velocity is thus the vector sum of
1
and u
1
, shown as V
1
.

Similarly, the flow exits at r = r
2
with component w
2
parallel to the blade
angle
2
plus tip speed u
2
= r
2
, with resultant velocity V
2
.

9.3.1 Euler Turbomachine Equations
Consider the angular momentum L of an elemental mass of fluid.
dm
V
Vt
r
O

Fig. 9.6 Inlet and exit velocity diagrams for an idealised pump impeller (from White).
9-8

t
L dmrV =
Consider change of angular momentum as it crosses the impeller.

( )
2 1
2 1 t t
dL dm r V rV =
The time rate of change of angular momentum is:

( )
2 1
2 1 t t
dL dm
r V rV
dt dt
=
Now integrate across all of the impeller.
Applied torque T = rate of change of angular momentum

( )
2 1
2 1 t t
T Q r V rV = where Q = total volumetric flow (m
3
/s)
Thus:

( )
2 1
2 1 w t t
P T Q u V u V = =

( )
2 1
2 1
1
w
t t
P
H u V u V
gQ g
= =
These are the Euler turbomachine equations. For given blade inlet and exit tip angles
1
and
2
, they can be used for simple pump design or analysis. They are for flow without losses.

9.3.2 Measured Performance Curves
In practice, losses occur due to:
Leakage through impeller casing clearances
Non-ideal flow losses (viscous effects)
Friction in bearings and seals

Design of a practical pump is very complicated and is based very much on experimentation.
Fig. 9.7 shows the measured performance of two centrifugal pumps, each with a different
casing diameter (pump size) and three different impeller diameters. (Unfortunately, data is in
American units.)

NPSH is the Net Positive-Suction Head which is the head required at the pump inlet to
prevent the liquid from cavitating or boiling locally. The pump inlet is the low-pressure point
where cavitation will occur first. Apart from the damage caused to the impeller, cavitation
can also lead to rotating unbalance.
9-9
NPSH =
2
2
i i v
p V p
g g g
+ where p
v
= vapour pressure of liquid.
Given NPSH from the pump performance curves, we must ensure that the right hand side is
equal or greater in the actual system to avoid cavitation. This might involve putting the pump
in a pit to increase p
i
(from Bernoulli equation).

Fig. 9.7 Measured performance curves for two models of a centrifugal water pump: (a) basic
casing with three impeller sizes; (b) 20 percent larger casing with three larger impellers at
slower speed (from White).

9-10
Similarity
For a given geometry of pump (fixed casing/impeller diameter ratio, blade inlet and outlet
angles) a technique called dimensional analysis gives rise to a number of dimensionless
coefficients.
Capacity coefficient
3 Q
Q
C
nD
=
Head coefficient
2 2 H
gH
C
n D
=
Power coefficient
3 5 P
bhp
C
n D
=
where n = shaft speed in revs/s and D = impeller diameter. Consistent units must be used
throughout. It follows that:
Efficiency
H Q
P
C C
C
=
We can see how this agrees with the pumps in Fig. 9.7. They must be geometrically similar,
so we can look at only two pumps, viz. those with impeller diameters D = 38 in and D = 32
in. These have approximately the same ratio of casing/impeller diameter.

Fig. 9.8 shows that these dimensionless
coefficients work well. Then we can
answer questions such as:
What if we use a larger
pump?
What if we run the pump
away from the condition of
maximum efficiency?

It follows that if pump 1 and pump 2
are from the same geometric family
and are operated at homologous points
(the same dimensionless position on a
chart such as Fig. 9.8, their flow rates,

Fig. 9.8 Nondimensional plot of the pump performance
data from Fig. 9.7 (from White).
9-11
heads and powers are related by similarity rules as follows:

3
2 2 2
1 1 1
Q n D
Q n D

=

2 2
2 2 2
1 1 1
H n D
H n D

=

3 5
2 2 2 2
1 1 1 1
P n D
P n D

=

These similarity rules can be used to estimate the effect of changing the fluid, speed or size
on any rotodynamic pump, within a geometrically similar family. A graphic display of these
rules is given in Fig. 9.9, showing the effect of speed and diameter changes on pump
performance. The curves are plotted to scale but with arbitrary units. The speed effect is
substantial, but the size effect is even more dramatic, especially for power, which varies as
D
5
. Generally, a given pump family can be adjusted in size and speed to fit a variety of
system characteristics.

Strictly, there must be similarity of efficiencies also;
1 2
= , but it can be shown that viscous
and leakage effects are less significant in large pumps, leading to higher efficiencies.
Corrections for this have been developed (see White).

Fig. 9.9 Effect of change of speed on homologous pump performance: (a) 20 percent change in
speed at constant size; (b) 20 percent change in size at constant speed (from White).
9-12
9.4 Pumps in Power Generation Facilities
(This section is mainly from MPSP)

Table 9.1 is a summary of power station pump applications.

Table 9.1 Power station pump applications (from Black and Veatch).
9-13
9.4.1 Circulating Water Pumps
Vertical spindle pumps are generally used, with either metal casings or concrete volutes
which are more economical for the largest sizes. Speeds are normally in the range 150 to 300
rpm and it is usually economically justifiable to include a step-down gearbox between the
motor and the pump. Fig. 9.10 shows a concrete volute pump.

Fig. 9.10 Bottom inlet single-entry impeller concrete volute pump (from MPSP).
9-14
A cooling water pump gearbox is shown in Fig. 9.11.

Gearboxes, when included on vertical pump units, must be of the coaxial type with the pump,
motor and gearbox centrelines positioned vertically above each other to ensure equal load
distribution on the support ring. With an epicyclic design it is possible to combine the
gearbox and the pump by mounting the impeller on the low speed output shaft of the gear
unit. This eliminates the need for one journal bearing and reduces overall height.

Fig. 9.11 A cooling water pump epicyclic gearbox (from MPSP).
9-15
The gearbox is equipped with a complete forced-lubrication oil system for the gears and the
thrust and journal bearings.

9.4.2 Condenser Extraction Pumps
Two condenser extraction pumps are normally used with a standby unit. The extraction
pumps are designed to pump condensate from the condenser at a saturation temperature
corresponding to the sub-atmospheric pressure within the condenser. Under these conditions,
the available positive suction head is very low and suction conditions are very stringent. The
rotational speed of the pump and the flow of condensate into the first stage impeller should be
kept as low as possible and the impeller eye must have the maximum capacity for handling
water in order to provide the minimum NPSH requirements and thus avoid unnecessary depth
of the extraction pumps below the condenser.

Fig. 9.12 shows a horizontal split-casing extraction pump. One of the most important factors
is the avoidance of oxygen ingress into the condensate through glands which may be required
to seal against sub-atmospheric pressures. A secure supply of seal water is necessary.

In recent years the required pump generated head has risen significantly to over 250 m. This
causes design problems on large split-casing pumps. A multi-stage vertical caisson type
pump (Fig. 9.13) is often preferred, and installed in a pit to reduce the risk of cavitation
erosion.

9.4.3 Boiler Feed Pumps
Boiler feed pumps installed on early 500 MW units were typically as shown in Fig. 9.14.
They included a long flexible shaft with about six stages and the residual thrust from the
impellers taken by a conventional balance disc arrangement. The bearings were mounted on
separate pedestals which had to be removed before the main pump internals could be
withdrawn. Glands were of the fixed labyrinth type, and the main bolted casing joint was
subjected to full discharge pressure. With these designs, loss of water or reduction in NPSH
(even momentarily) is likely to lead to metallic contact resulting in pump seizure. The bolted
casing joint arrangement then results in a substantial outage for replacement of the pump
internals.

9-16
To overcome the shortcomings of the multi-stage flexible shaft design of feed pump, a new
concept was developed with the emphasis on achieving maximum availability coupled with a
design life of at least 45,000 hours for all components. The essential features were a rugged
high speed unit capable of surviving dry running and thermal shock operation without
damage, and having a cartridge construction that permitted rapid replacement of the complete
rotor/stator assembly. An example is shown in Fig. 9.15.

9.4.4 Slurry Pumps
A typical slurry pump (Fig. 9.16) differs from other centrifugal pumps used for clear liquids
in a number of ways. Wetted-end parts (casing, impeller) have walls that are thicker or are
equipped with hard metal or rubber liners. Slurry pumps are often vertically split to allow for
replacement of casing and impeller liners. Flow passages through both the casing and
impeller are large enough to permit solids to pass without clogging the pump. Since the gap
between the impeller face and suction liner increases with wear, the rotating assembly of the
slurry pump must be capable of axial adjustments to maintain the manufacturers
recommended clearance. This is critical if heads, capacities and efficiencies are to be
maintained.

9.4.5 Miscellaneous Pumps
There are a large number of smaller pumps in a power station for uses that include:
Water treatment plant. These are centrifugal with rolling element bearings.
Injection of chemicals into boiler feed water. A measured quantity is obtained
with positive-displacement pumps.
Fire pumps. These are simple and robust centrifugal pumps.

9-17

Fig. 9.12 Horizontal split-casing extraction pump (from MPSP).
9-18

Fig. 9.13 Vertical caisson extraction pump (from MPSP).
9-19

Fig. 9.14 500 MW main boiler feed pump (from MPSP).

Fig. 9.15 Modern feed pump construction (from MPSP).
9-20

Fig. 9.16 Typical slurry pump (from Black and Veatch).
MECH7350 Rotating Machinery 10. Hydraulic Turbines

10-1
10. HYDRAULIC TURBINES
(This section is taken mainly from White)
10.1 Introduction
Hydraulic turbines extract energy from water which has a high head. There are basically two
types, reaction and impulse, the difference being in the manner of head conversion. In
reaction turbines the water fills the blade passages and the head change or pressure drop
occurs within the impeller. They can be of radial, axial or mixed flow types. In impulse
turbines the high head is first converted through a nozzle into a high velocity jet which strikes
the blades at one position as they pass by. Reaction turbines are smaller because water fills
all the blades at one time.

10.2 Reaction Turbines
Reaction turbines are low-head, high-flow devices. The flow is opposite to that in a pump
(from volute to eye of impeller after transferring most of the energy of the water to the
impeller) but a difference is the important role stationary guide vanes play. Purely radial and
mixed flow designs are called Francis turbines. At even lower heads an axial flow, propeller
turbine is more compact. It can be fixed bladed but better efficiency is obtained over an
operating range by using adjustable vanes, in the Kaplan turbine. Various impeller
configurations are shown in Fig. 10.1.

Fig. 10.1 Reaction turbines: (a) Francis, radial type; (b) Francis, mixed-flow; (c) propeller axial-flow;
(d) performance curves for a Francis turbine, n = 600 rpm, D = 0.686 m, Nsp = 29 (from White).

10-2
10.3 Simple Radial Turbine Theory
The Euler turbomachine equations derived in Chapter 9 for pumps also apply to turbines if
we reverse the flow direction and reshape the blades. Fig. 10.2 shows a radial turbine runner.

Again assume one-dimensional frictionless flow through the blades. Adjustable inlet guide
vanes are essential for good efficiency. They bring the inlet flow to the blades at angle and
absolute velocity V
2
for minimum shock or directional miss-match loss. After vectorially
adding in the runner tip speed u
2
= r
2
, the outer blade angle should be set at
2
to
accommodate the relative velocity w
2
.

Consideration of angular momentum as for pumps (Chapter 9) gives an idealised formula for
the power P extracted by the runner:
( ) ( )
2 2 1 1 2 2 2 1 1 1
cos cos
t t
P T Q r V rV Q u V uV = = =
where V
t2
and V
t1
are the absolute inlet and outlet circumferential velocity components of the
flow.

The absolute inlet flow normal velocity V
n2
= V
2
sin
2
is proportional to the flow rate Q. If
the flow rate changes and the runner speed u
2
is constant, the vanes must be adjusted to a new
angle
2
so that w
2
still follows the blade surface.
Fig. 10.2 Inlet and outlet velocity diagrams for an idealised radial-flow reaction turbine runner (from White).

10-3
10.4 Power Specific Speed
Turbine parameters are similar to those of a pump, but the dependent variable is the output
brake horsepower which depends on the inlet flow rate Q, available head H, impeller speed n
and diameter D. The efficiency is the output brake horsepower divided by the available water
horsepower gQH. The dimensionless forms are C
Q
, C
H
and C
P
defined as for a pump
(Chapter 9 in Module A). If we neglect viscous and roughness effects, the functional
relationships are written with C
P
as the independent variable:
( ) ( ) ( )
2 2 3 H H P Q Q P P
gH Q bhp
C C C C C C C
n D nD gQH

= = = = = = (10.1)
Fig. 10.1 shows typical performance curves for a Francis radial turbine. The maximum
efficiency point is called normal power.

A parameter that compares the output power with the available head, independent of size, is
found by eliminating the diameter between C
H
and C
P
. It is called the power specific speed
sp
N .
( )
( )
1/ 2
1/ 2
5/ 4 5/ 4
1/ 2
P
sp
H
n bhp
C
N
C
gH
= = (10.2)
In lazy but common form this is written as:
( ) ( )
( )
1/ 2
5/ 4 sp
rpm bhp
N
H ft
=

(10.3)
Like pumps, turbines of large size are generally more efficient.

10.5 Impulse Turbines
For high head (typically above 250 m) and relatively low power (i.e. low N
sp
from (10.2)) not
only would a reaction turbine require too high a speed but also the high pressure in the runner
would require a massive casing thickness. The impulse turbine in Fig. 10.3 is ideal for this
situation. Since N
sp
is low, n will be low and the high pressure is confined to the small
nozzle which converts the head to an atmospheric pressure jet of high velocity V
j
. The jet
strikes the buckets and imparts a momentum change. The buckets have an elliptic split-cup
shape and are called Pelton wheels.


10-4

A simple analysis uses the Euler turbomachinery equation in Chapter 9, i.e.;
( )
2 1
2 1 w t t
P T Q u V u V = =

( )
2 1
2 1
1
w
t t
P
H u V u V
gQ g
= =
together with the velocity diagram in Fig. 10.3. Noting that u
1
= u
2
= u, we substitute the
absolute exit and inlet tangential velocities into the turbine power relation:
( ) ( )
{ } 1 1 2 2
cos
t t j j
P Q uV u V Q uV u u V u

= = +

or
( ) ( ) 1 cos
j
P Qu V u = (10.4)
where u = 2nr is the bucket linear velocity and r is the pitch radius, or distance to the jet
centreline. A bucket angle = 180
o
gives maximum power but is physically impossible
because water must clear the next bucket. In practice, 165
o
and 1 cos 1.966 or only
2 percent less than maximum power.

From (10.4) the theoretical power of an impulse turbine is a maximum when dP/du = 0, or

Fig. 10.3 Impulse turbine: (a) side view of wheel and jet; (b) top view of bucket; (c)
typical velocity diagram (from White).

10-5
* *
1
2
2
j
u n r V = =
For a perfect nozzle, the entire available head would be converted to jet velocity
( )
1/ 2
2
j
V gH = . Since there are 2 to 8 percent nozzle losses, a velocity coefficient C
v
is used:
( )
1/ 2
2 0.92 0.98
j v v
V C gH C = (10.5)
By combining (10.1) and (10.5) the theoretical impulse turbine efficiency becomes:
( ) ( ) 2 1 cos
v
C = (10.6)
where
( )
1/ 2
2
u
gH
= = peripheral velocity factor
Maximum efficiency occurs at
1
0.47.
2
v
C =
Fig. 10.4 shows (10.6) plotted for an ideal turbine ( = 180
o
, C
v
= 1.0) and for typical
working conditions ( = 160
o
, C
v
= 0.94). The latter case predicts
max
= 85 percent but
windage, mechanical friction, backsplashing and nonuniform bucket flow reduce this to about
80 percent. An impulse turbine is not quite as efficient as the Francis or propeller turbines at
their BEPs.

Fig. 10.5 shows the optimum efficiency of the three types of turbines, and the importance of
the power specific speed N
sp
as a selection tool for designers.
Fig. 10.4 Efficiency of an impulse turbine calculated from (10.6): solid curve = ideal, = 180
o
,
Cv = 1.0; dashed curve = actual, = 160
o
, Cv = 0.94; open circles = data, Pelton wheel,
diameter = 0.61 m (from White).

10-6

The water power available to a turbine may vary due to either head or flow rate changes, both
of which are common in hydroelectric plants. The demand for power also varies from light
to heavy, and the operating response is a change in the flow rate by adjustment of a gate
valve or needle valve (Fig. 10.3). As shown in Fig. 10.6, all three turbines achieve fairly
uniform efficiency as a function of the level of power being extracted. Especially effective is
the adjustable-blade (Kaplan-type) propeller turbine.

10.6 Some Practical Considerations
Cavitation must be avoided in hydraulic turbines. It can occur at turbine outlets where the
pressure is lowest.

More detail of a Pelton wheel turbine is shown in Fig. 10.7. Rapid shutdown of the turbine,
as would be required after loss of load from the driven machine, cannot be effected by rapid
Fig. 10.5 Optimum efficiency of turbine designs (from White).

Fig. 10.6 Efficiency versus power level for various turbine designs at constant speed and
head (from White).

10-7
closure of the spear (needle valve) due to water hammer effects. Shutdown is effected by
inserting into the jet either a cut-off to destroy its kinetic energy or a deflector to direct the jet
away from the bucket. The jet is then cut off at a suitably safe rate.

Fig. 10.8 shows a typical large Francis turbine in which water is fed radially to the runner
from guide vanes which are disposed around the full circumference. The angle of these vanes
can be varied to control machine output and the water is uniformly distributed to them by a
spiral casing. The loads on the guide vanes can be very large. The guide vanes are moved by
cranks attached to the end of one of the spindles, the other end of the crank being located in a
regulating ring. The ring is rotated by hydraulic rams or servomotors. Some form of slipping
device is incorporated so that if a foreign body jams between two vanes, the remainder can be
moved normally. A warning device indicates that the slipping device is operating. Large
axial thrusts are handled by admitting some high-pressure water to the underside of the
machine and by using a thrust bearing (usually of the tilting-pad type).

Fig. 10.9 shows details of a large Kaplan turbine through which the water flow is essentially
axial. The runner resembles a ships propeller whose blades (typically four to six in number)
are adjustable in pitch to present the optimum angle of attack to the water flow. The blades
are most conveniently adjusted by means of a lever arm within the hub.

Fig. 10.7 Pelton wheel turbine (from MPSP).

10-8
10.7 Pumped-Storage
Pumped-storage involves operations between two lakes; water is run through a turbine when
peak electricity generation is needed, and pumped from the lower to the higher lake to store
potential energy at periods of low demand. Separate pumps and turbines can be used, or a
reversible pump-turbine. Such a machine enables both pumping and turbining to be
performed by the same runners. It is effectively a Francis turbine whose runner geometry is a
compromise between the optimum for pumping and generation. During generation, the
turbine output is absorbed by the generator; whilst pumping, the generator acts as a motor,
driving the runner in the opposite direction. An advantage of the combined unit is the
reduction of equipment and installation costs.

Fig. 10.8 A large Francis turbine (from MPSP).

10-9

Fig. 10.9 A large Kaplan turbine showing detail of the hub mechanism for varying blade angle (from
MPSP).
MECH7350 Rotating Machinery 11. Fans

11-1
11. FANS
(This section is taken mainly from Black and Veatch)
11.1 Types of Fans
Fans are used in electricity generation to supply or exhaust air or flue gas. They increase the
pressure of a flow stream to offset the pressure losses that result from system resistance.
Centrifugal fans move air or gas perpendicular to the impeller shaft. Axial fans move air or
gas parallel to the impeller shaft.

11.1.2 Centrifugal Fans
Figures 11.1 and 11.2 show the construction and components of centrifugal fans. Centrifugal
fan blades are mounted in an impeller that rotates within a spiral housing. They are designed
with either one or two inlets, and normally run at constant speed with flow and pressure
controlled with inlet dampers.

Fig. 11.1 Double-inlet centrifugal fan (from Black and Veatch).

11-2
The performance of centrifugal fans is highly dependent on the type of blades used; backward
curved, straight or radial tip. This is shown in Fig. 11.3.

Fig. 11.2 Centrifugal fan components (from Black and Veatch).

Fig. 11.3 Fan blade types and performance curves (from Black and Veatch).

11-3
Backward curved blades have an aerofoil shape and are most widely used in power plant
operations. Their advantages are:
Highest efficiency, over 90 percent
Very stable operation
Low noise
Capacity for high speed operation
They should be avoided where large or adhesive particles are present.

11.1.3 Axial Fans
Single-stage (one rotor and one set of blades) axial fans are typically used in a forced draft
situation on a balanced draft steam generator. When axial fans are designed for induced draft
service, the higher pressure requirements normally dictate use of a two-stage fan such as in
Fig. 11.4. Axial fans are driven by single-speed or two-speed motors, with flow and output
pressure controlled by varying blade pitch with a hydraulically actuated mechanism. Axial
fans can maintain higher efficiencies at various steam generator loads than can constant-
speed centrifugal fans controlled with inlet dampers.

Use of lightweight fan blade materials such as aluminium or magnesium reduces the strength
requirements of the fan hub and the fan blade thrust bearing loads. This reduces equipment

Fig. 11.4 Two-stage axial fan (from Black and Veatch).

11-4
costs but in erosive situations blade protection might be needed, such as a hardened steel strip
on the leading edge.

11.2 System Resistance
A loss of pressure occurs when gas is forced through a duct system. This loss of pressure is
called system resistance, and is composed of two parts; friction losses and dynamic losses.
Friction losses mainly occur at the walls of the duct system and can be quantified by the
following empirical equation:
2
friction
f V L
P
Dg
= (11.1)
where

friction
P = frictional pressure loss
f = a dimensionless friction factor
= gas density
V = gas velocity
L = duct length
D = duct diameter
g = gravitational constant

Dynamic losses occur at changes of direction in gas flow and at sudden duct enlargements
and contractions. They are sometimes called velocity pressure losses. Provided that the gas
flow is turbulent, as is the case for nearly all combustion air and flue gas handling systems
and equipment, dynamic losses can be empirically quantified by the following equation:
2
1
2
dynamic
KV
P
g
= (11.2)
where
K = system constant based on geometry of duct and determined experimentally.
Then
friction dynamic
P P P = + (11.3)
When the system geometry is not modified (i.e. when f, L, D, g and K are held constant) only
changes in gas density and flow rate cause the system resistance to change. Once the
pressure test has been made at a known gas flow rate and density, the system resistance can

11-5
be predicted for any flow condition. If the gas density is not changed, system resistance can
be plotted against flow Q which is proportional to V.

11.3 Fan Performance
Fans are used to provide the pressure necessary to overcome system resistance. Fan
performance characteristics are developed from test data. A typical fan performance curve
(in American units; in. wg = inches water gauge, acfm = actual cubic feet per minute) for a
centrifugal fan operating at a given speed and gas density is shown in Fig. 11.5.

Once the fan is installed into a ductwork system, the intersection of the system resistance
curve and the fan characteristic curve defines the system operating point, as shown in Fig.
11.6.

Fig. 11.5 Typical centrifugal fan performance curve (from Black and Veatch).

Fig. 11.6 Typical installed centrifugal fan performance curve (from Black and Veatch).

11-6
An alternative to plotting fan brake horsepower requirement is to show fan efficiency as a
function of flow, as in Fig. 11.7. Fan efficiency is defined similarly to efficiency for a pump;
air power
PAV PQ
shaft bhp
= = = (11.4)

11.3.1 Similarity Rules
Fan performance curves are developed by testing model fans. The results of these tests are
used as a basis for determining the performance capabilities of full-size geometrically similar
fans. If gas compressibility effects are neglected (and they are usually small) then the
similarity rules based on dimensionless coefficients that were developed in Module A,
Chapter 9 for pumps apply.

11.4 Fan Operating Characteristics
In applications that require fans to maintain variable and controlled pressure or flow
conditions, the fans are equipped with flow or pressure regulating capability. With
centrifugal fans, this capability is normally provided by either inlet dampers, outlet dampers,
inlet vane control (located in the fan inlet cone) two-speed control, or variable-speed control.
With axial flow fans, adjustable blade pitch control is used. Fig. 11.8 illustrates typical
impacts of dampers or inlet vanes on centrifugal fan performance. The fan provides flow for
the system ABC. With full open inlet dampers or inlet vanes, the fan operates at point A with
bhp of A'. With three-quarter-open inlet dampers or inlet vanes, the fan operates at point B
with bhp of B'.

Fig. 11.7 Fan efficiency curve (from Black and Veatch).

11-7

Axial flow fans are normally controlled by operating the blade pitch. A characteristic
performance field for a typical constant-speed, variable pitch axial flow fan is shown in Fig.
11.9.

Fig. 11.8 Inlet vane control of centrifugal fans (from Black and Veatch).

Fig. 11.9 characteristic performance field for a typical constant-speed, variable pitch axial
flow fan

11-8
Each blade pitch angle results in a specific flow versus head characteristic curve. As the
blade pitch angle is increased, the pressure and flow capacity of the fan also increase.
However, axial fans are more prone to stall than are centrifugal fans. In stall, the pitch is so
high that the blades are required to generate such a high pressure rise that flow separation
occurs. Operation in this condition produces noise and vibration and can lead to fatigue
failure.

11.5 Other Practical Issues
11.5.1 Fan Vibration
Fan vibration is primarily due to rotor unbalance and balancing can be undertaken. However,
some unbalance always remains. If the fan rotor and bearing system natural frequency are
designed to be above the fan speed, the vibration sensitivity of the fan rotor to the remaining
unbalance can be reduced to acceptable levels.

11.5.2 Fan-induced Duct Vibrations
Pressure pulsations of sufficient magnitude to damage ductwork systems are normally
attributable to the associated fan and fan inlet flow conditions and are eliminated during fan
development by the introduction fins at various places near the inlet.

11.5.3 Fan Noise
Fan noise consists of two separate components. Single-tone noise results from the flow
leaving the blades passing by stationary objects such as flow straightening devices. The
blade passing frequency and its first harmonic are usually the most dominant. Broad band
noise is produced by the high-velocity air stream passing through the stationary objects in the
ductwork, dampers and fan housing. The noise generated by both of these components
travels out of the inlet box, through the discharge duct, and also through the fan housing.

Forced draft (FD) and primary air (PA) fan inlets normally have absorptive-type inlet
silencers. These consist of several rows of panels in the airstream that have perforated plate
skin and are filled with acoustically absorptive material. Induced draft (ID) fans do not
normally require noise reduction equipment.


11-9
11.6 Fan Selections
FD fans supply combustion air to the steam generator. The fans must have a pressure
capability high enough to overcome the total resistance of inlet silencers, air preheat coils, air
ducts, air heaters, wind boxes, burner registers, and any other resistance between the air
intake of the fan and the furnace. ID fans exhaust combustion products from the steam
generator. They maintain furnace pressure slightly below atmospheric. The condition is
known as balanced draft operation (balance between the FD and ID fans). The trend in fan
technology for large coal-fired power stations is toward axial-flow FD and IP fans. Although
they have complicated mechanical components, axial-flow fans offer lower operating costs
through higher efficiencies. If strong erosion resistance is needed for the flue gases, ID fans
can be of the radial tip centrifugal design.

PA fans carry the pulverised coal from the pulveriser to the furnace. This is a relatively low
flow with a high pressure differential, and robust, high speed centrifugal fans are generally
used.

Gas recirculation (GR) fans draw gas from a point between the economiser outlet and the air
heater inlet and discharge it into the bottom of the furnace or near the furnace outlet.
Recirculated gas introduced in the vicinity of the initial burning zone of the furnace is used
for steam temperature control, while recirculated gas introduced near the furnace outlet is
used for control of gas temperature. The combination of heavy dust loads and rapid
temperature changes demands the use of a rugged centrifugal fan.

Cooling tower fans are propeller-type fans designed to produce air velocities that are as
uniform as possible across the effective area of the fan. Blades have an aerofoil cross-section
and are tapered and twisted. Fibreglass-reinforced plastics are the most commonly used
blade material because of their light weight and exceptional corrosion resistance.

11.7 Fan Drives
Steam generator draft fans are normally driven with one of the following drive systems:
Single-speed induction motor
Two-speed induction motor (normally operated at the lower speed)
Adjustable speed electric motor

11-10
Stream turbine
Variable-speed fluid drive.
Fig. 11.10 shows a typical hydrokinetic fluid coupling. A runner on the driven (output) shaft
receives a vortex of oil from the impeller attached to the driving (input) shaft. There is no
mechanical connection between the runner and the impeller, and the two are almost identical
in shape. Kinetic energy is imparted to the oil by the impeller, and the oil flows radially
outwards and into the vanes of the runner. The oil then flows through the runner and
transfers its energy to the runner, similar to steam imparting energy to a turbine blade. The
speed of the runner and the energy transferred to the runner by the impeller are controlled by
the amount of oil flowing through the impeller.

Fig. 11.10 Hydrokinet coupling (from Black and Veatch).
MECH7350 Rotating Machinery 12. Wind Turbines
12-1
12. WIND TURBINES

(This section is based largely on White and Rixon)

12.1 Introduction
Some examples of wind turbine designs are shown in Figures 12.1 to 12.4. HAWT and
VAWT are vertical axis wind turbine and horizontal axis wind turbine respectively.

Fig. 12.1 American multiblade farm HAWT. Fig. 12.2 Darrieus VAWT.
Fig. 12.3 Savonius VAWT
Fig. 12.4 Modern large 3-blade propeller
HAWT.
12-2
A summary of the advantages and disadvantages of VAWTs and HAWTs follows.

Advantages of vertical wind turbines
Easier to maintain because most of their moving parts are located near the ground.
This is due to the vertical wind turbines shape. The airfoils or rotor blades are
connected by arms to a shaft that sits on a bearing and drives a generator below,
usually by first connecting to a gearbox.
As the rotor blades are vertical, a yaw device is not needed, reducing the need for this
bearing and its cost.
Vertical wind turbines have a higher airfoil pitch angle, giving improved
aerodynamics while decreasing drag at low and high pressures.
Hilltops, ridgelines and passes can have higher and more powerful winds near the
ground than up high because of the speed up effect of winds moving up a slope or
funneling into a pass combining with the winds moving directly into the site. In these
places, VAWTs placed close to the ground can produce more power than HAWTs
placed higher up.
Low height useful where laws do not permit structures to be placed high.
Smaller VAWTs can be much easier to transport and install.
Does not need a free standing tower so is much less expensive and stronger in high
winds that are close to the ground.
Usually have a lower Tip-Speed ratio so less likely to break in high winds.

Disadvantages of vertical wind turbines
There may be a height limitation to how tall a vertical wind turbine can be built and
how much sweep area it can have.
Most VAWTS need to be installed on a relatively flat piece of land and some sites
could be too steep for them but still be usable for HAWTs.
Most VAWTs produce energy at only 50% of the efficiency of HAWTs in large part
because of the additional drag that they have as their blades rotate into the wind.

Advantages of horizontal wind turbines
Blades are to the side of the turbine's center of gravity, helping stability.
Ability to wing warp, which gives the turbine blades the best angle of attack.
Allowing the angle of attack to be remotely adjusted gives greater control, so the
turbine collects the maximum amount of wind energy for the time of day and season.
Ability to pitch the rotor blades in a storm, to minimise damage.
Tall tower allows access to stronger wind in sites with wind shear. In some wind shear
sites, every ten meters up, the wind speed can increase by 20% and the power output
by 34%.
Tall tower allows placement on uneven land or in offshore locations.
Can be sited in forests above the treeline. .
Most are self-starting.
12-3
Can be cheaper because of higher production volume, larger sizes and, in general
higher capacity factors and efficiencies.

Disadvantages of horizontal wind turbines
HAWTs have difficulty operating in near ground, turbulent winds because their yaw
and blade bearing need smoother, more laminar wind flows.
The tall towers and long blades (up to 60 m long) are difficult to transport on the sea
and on land. Transportation can now cost 20% of equipment costs.
Tall HAWTs are difficult to install needing very tall and expensive cranes and skilled
operators.
The FAA has raised concerns about tall HAWTs effects on radar in proximity to air
force bases.
Height can be a safety hazard for low-altitude aircraft.
Offshore tower can be a navigation problem.
Downwind variants suffer from fatigue and structural failure caused by turbulence.

Large scale electricity generation is dominated by the rugged three-bladed HAWT type that
has been championed by Danish and German turbine manufacturers. These have high tip
speeds of up to 6 times wind speed, high efficiency, and low torque ripple which contributes
to good reliability. The world's largest turbines deliver up to 6 MW , have an overall height
of 186 m and a diameter of 126 m.

12.2 Idealised Theory
The performance of an ideal, frictionless, propeller wind turbine was predicted in 1920 by
Betz, using the simple model shown in Fig. 12.1.

Fig. 12.5 Idealised actuator disc and streamtube analysis of flow through a propeller-type wind
turbine (from White).
12-4
The propeller is represented as an actuator disc which creates across the propeller plane a
pressure discontinuity of area A and local velocity V. The wind is represented by a
streamtube of approach velocity V
1
in the x-direction and a slower downstream wake velocity
V
2
. The pressure rises to p
b
just before the disc and drops to p
a
just after it, returning to free-
stream pressure in the far wake. To hold the propeller rigid when it is extracting energy from
the wind, there must be a force F acting to the left on the support.

Consider Newtons Second Law applied to a control volume in the streamtube between
sections 1 and 2 (sum of external forces equals rate of change of momentum).
( )
2 1 x
F F m V V = =
(12.1)
where m AV = is the mass flow through the propeller.
Similarly, for a control volume enclosed by sections b and a just before and just after the
disc,
( ) ( ) 0
x b a a b
F F p p A m V V = + = =
(12.2)
Equating (12.1) and (12.2) gives the propeller force:
( ) ( )
1 2 b a
F p p A m V V = = (12.3)
Assuming ideal flow, the pressures can be found from the incompressible Bernoulli equation
up to the disc. (Bernoulli does not apply across the disc because of the presence of an
external force F.)

From 1 to b:
2 2
1
1 1
2 2
b
p V p V
+ = +
From a to 2:
2 2
2
1 1
2 2
a
p V p V
+ = +
Subtracting these two equations and noting that m AV = through the propeller, we can
substitute for p
b
- p
a
in (12.3) to get
( ) ( )
2 2 2 2
1 2 1 1 2
1 1 1
2 2 2
b a b
p p V V p V p V V V
= = + = +
or
( )
1 2
1
2
V V V = + (12.4)
Continuity and momentum thus require that the velocity V through the disc equals the
average of the wind and far-wake speeds.
12-5
Finally, the power P extracted by the disc is the difference between the rate of energy flow
upsteam and downstream of the disc.
( )
( )( ) ( ) ( )
2 2 2 2
1 2 1 2
2
1 2 1 2 1 2 1 2
1 1 1
2 2 2
1 1
2 4
P mV mV m V V
AV V V V V A V V V V
= =
= + = +

( ) ( )
2 2
1 2 1 2
1
4
V V V V = + (12.5)
For a given wind speed V
1
, we can find the maximum power possible by differentiating P
with respect to V
2
and setting this equal to zero. The result is
3
max 1
8
27
P P AV = = at
2 1
1
3
V V = (12.6)
which corresponds to V = 2V
1
/3 through the disc.
The maximum power available to the propeller is the mass flow through the propeller times
the total kinetic energy of the wind:
2 3
1 1
1 1
2 2
avail
P mV V = =
Thus the maximum possible efficiency of an ideal frictionless wind turbine is usually stated
in terms of the power coefficient:
3
1
1
2
P
P
C
AV
= (12.7)
Incorporating (12.6) gives the maximum power coefficient as
,max
16
0.593
27
P
C = = (12.8)
This is called the Betz number and serves as an ideal with which to compare the actual
performance of real wind turbines.

Fig. 12.6 shows the measured power coefficients of various wind turbine designs. The
independent variable is not V
2
/V
1
(which is artificial and convenient only in ideal theory) but
the ratio of blade-tip speed r to wind speed. The Darrieus rotor has the many advantages of
a vertical axis but has little torque at low speeds and also rotates more slowly at maximum
power than a propeller, thus requiring a higher gear ratio for the generator. The Savonius
rotor produces power at very low wind speeds, but it is inefficient and susceptible to storm
damage because it cannot be feathered in high winds.

12-6

12.3 Some Practical Aspects
Fig. 12.7 shows diagrammatically the main features of a large modern propeller turbine.
Cyclic stresses fatigue the blade, axle
and bearing material, and were a major
cause of turbine failure for many years.
Because wind velocity often increases at
higher altitudes, the backward force and
torque on a horizontal-axis wind turbine
(HAWT) blade peaks as it turns through
the highest point in its circle. The tower
hinders the airflow at the lowest point in
the circle, which produces a local dip in
force and torque. These effects produce
a cyclic twist on the main bearings of a
HAWT. The combined twist is worst in
machines with an even number of blades,
where one is straight up when another is
straight down. To improve reliability,
teetering hubs have been used which

Fig. 12.7 Diagram of a large propeller turbine.

Fig. 12.6 Performance of various wind turbine designs as a function of blade-tip speed ratio
(from White).
American multi-
blade farm
Large
propeller
HAWT
Darrieus
VAWT
Dutch four-arm
Savonius
rotor
High speed
HAWT
Grumman
windstream
12-7
allow the main shaft to rock through a few degrees, so that the main bearings do not have to
resist the torque peaks.

When the turbine turns to face the wind, the rotating blades act like a gyroscope. As it pivots,
gyroscopic precession tries to twist the turbine into a forward or backward somersault. For
each blade on a wind generator's turbine, precessive force is at a minimum when the blade is
horizontal and at a maximum when the blade is vertical. This cyclic twisting can quickly
fatigue and crack the blade roots, hub and axle of the turbine.

Fig. 12.8 and Table 12.1 show detail of a Danish Vesta 3 MW propeller turbine.

Fig. 12.8 Details of a Vesta 3 MW propeller turbine.

1. Hub controller 11. High voltage transformer
2. Pitch cylinders 12. Blade
3. Blade hub 13. Blade bearing
4. Main shaft 14. Rotor lock system
5. Air cooler 15. Hydraulic unit
6. Gearbox 16. Machine foundation
7. Mechanical disc brake 17. Yaw gears
8. Service crane 18. Composite disc coupling
9. VMP-Top controller with converter 19. Generator
10. Ultrasonic sensors 20. Air cooler for generator
12-8
Table 12.1 Technical specifications of Vesta 3 MW propeller turbine.
Rotor
Diameter: 90 m
Area swept: 6,362 m
2

Nominal revolutions: 16.1 rpm
Operational interval: 8.6 - 18.4 rpm
Number of blades: 3
Power regulation: Pitch control
Air brake: Full blade pitch by three separate hydraulic
pitch cylinders
Tower
Hub height: 80 - 105 m
Operational Data
Cut-in wind speed: 4 m/s
Nominal wind speed (3,000 kW): 15 m/s
Cut-out wind speed: 25 m/s
Generator
Generator: Asynchronous with speed control
Rated output: 3,000 kW
Operational data: 50 Hz
1,000 V
Gearbox
Type: Two planetary and one helical stage
Control
Type: Mikroprocessor-based control of all the
turbine functions with the option of remote
monitoring. Output regulation and
optimisation via OptiSpeed and OptiTip
pitch regulation.
Weight
Nacelle: 70 t
Rotor: 41 t
Tower
Hub height:
80 m 160 t
105 m 285 t

An anemometer measures the wind speed and transmits wind speed data to the controller to
orient the turbine properly with respect to the wind. A disc brake can be applied to stop the
rotor in emergencies. The controller starts up the machine at wind speeds of about 4 m/s and
shuts off the machine at about 25 m/s. Turbines cannot operate at wind speeds above about
25 m/s because their generators could overheat. The gear box connects the low-speed shaft to
12-9
the high-speed shaft and increases the rotational speed to that required by the generator to
produce electricity. The gear box is a costly and heavy part of the wind turbine. Blades are
turned, or pitched, out of the wind to keep the rotor from turning in winds that are too high or
too low to produce electricity. Towers are made from tubular steel (shown here) or steel
lattice. Because wind speed increases with height, taller towers enable turbines to capture
more energy and generate more electricity. Fig. 12.8 shows an upwind turbine, so-called
because it operates facing into the wind. The yaw drive is used to keep the rotor facing into
the wind as the wind direction changes. A control system allows the rotor speed to vary
within a range of approximately 60 per cent in relation to nominal rpm. Thus the rotor speed
can vary by as much as 30 per cent above and below synchronous speed. This minimises both
unwanted fluctuations in the output to the grid supply and the loads on the vital parts of the
construction.

The above systems are shown schematically in Fig. 12.9.

Fig. 12.9 Schematic of a propeller turbine.
12-10
12.4 Control Methods
The operation of a wind turbine involves starting the turbine from rest, regulating the power
while the system is running, and stopping the turbine if and when the wind speed becomes
excessive. Start-up of most turbines usually means running the generator as a motor to
overcome initial resistive torque until sufficient power is generated at cut-in speed assuming
that a power source is available.

The angle of the rotor blades is actively adjusted by the machine control system. This, known
as blade pitch control, has the advantage that the blades have built-in braking which brings
the blades to rest. Pitching the whole blade requires large actuators and bearings, increasing
the weight and expense of the system. One solution to this problem is to use partial span
blade pitch control where only the outer one third of the blade span is pitched.
MECH7350 Rotating Machinery 13. Rolling-Element Bearings and Gearboxes

13-1
13. ROLLING-ELEMENT BEARINGS AND GEARBOXES
(This section is taken from Juvinall and Marshek)
13.1 Rolling-Element Bearings
Rolling-element bearings are used to locate shafts in many fans, pumps and gearboxes where
the shaft and the outer members are separated by balls or rollers. The sliding friction in
journal bearings is replaced by rolling friction. Fig. 13.1 shows the main features of a radial
ball bearing. Small contact areas imply high stresses and the use of hard, high strength steel.

Fig. 13.1 Radial deep-groove ball bearing (from Juvinall and Marshek)

13-2
Rolling-element bearings have the following features:
Their major advantage is their low starting friction. This is not the case with
hydrodynamically lubricated journal bearings.
They are not particularly suited to high rotating speeds because:
o fatigue cycles are accumulated rapidly
o centrifugal forces on rolling elements are high
Rolling-element bearings take up more radial space around the shaft than plain journal
bearings, but less axial space.
They generate more noise than journal bearings
They do not require an expensive lubrication system
They can operate with no radial clearance. This is important if a shaft must be located
precisely.

13.2 Design of Rolling-Element Bearings
Design involves sophisticated engineering analysis, computation and experimentation. Users
are only interested in bearing selection but a few design features are worth noting.

Fig. 13.1(c) shows the geometry of ball-bearing contact surfaces. The selection of the
curvature of the bearing race is critical. If it is only slightly larger than the radius of the ball,
the contact area after local deformation is relatively large. This lowers contact stresses (or
increases load capacity). But different parts of this contact area are at various radii from the
axis of rotation. This causes sliding and in turn, friction and wear. A compromise is for the
radius of curvature of the race to be about 104 percent of the ball radius.

Material selection is critical. High-carbon chrome steels are generally used.

The design of the inner and outer rings for rigidity is important. Manufacturing tolerances
are critical.

13.3 Types of Rolling-Element Bearings
Rolling-element bearings are either ball bearings (capable of higher speeds) or roller bearings
(capable of higher loads). Both types can be classified as:
radial

13-3
thrust
angular-contact (for carrying both radial and axial loads)
These types are shown in Fig. 13.2.

Fig. 13.2 Representative types of ball bearings, with deep-groove shown in
Fig. 13.1 (from Juvinall and Marshek).

13-4
Roller bearings are classified by rolling element configuration as:
cylindrical
spherical
tapered
needle
These are shown in Figures 13.3 to 13.5.

Fig. 13.3 Spherical roller bearings (from Juvinall and Marshek).

Fig. 13.4 Tapered roller bearings (from Juvinall and Marshek).

13-5

Ball bearings are made in various proportions in order to accommodate various degrees of
loading (Fig. 13.6).

Fig. 13.5 Needle roller bearings (from Juvinall and Marshek).

Fig. 13.6 Relative proportions of bearings of different series (from
Juvinall and Marshel).

13-6
Bearings with filling notches (or loading grooves) permit assembly with the races concentric
and therefore contain more balls than the deep-groove type. This gives 20 to 40 percent
greater radial load capacity at the expense of thrust capacity.

Generally, ball bearings are not separable so they are adaptable to the use of seals and self-
contained grease lubrication (Fig. 13.7). Friction is higher than for open bearings with oil
mist lubrication.

13.4 Fitting
Normal practice is to fit the stationary ring with a slip or tap fit and the rotating ring with
enough interference to prevent relative motion during operation. Information is supplied by
the manufacturers. Too tight a fit can cause internal interference that shortens the life of the
bearing.

Fig. 13.7 Bearings with seals and shields (from Juvinall and Marshek).

13-7
13.5 Catalogue Information
Manufacturers catalogues, hard copy and web-based, provide complete information on:
dimensions
rated load capacities
mounting, lubrication and operation.
Dimensions of a representative set of bearings are given in Fig. 13.8 and Table 13.1. Rated
load capacity C is shown in Table 13.2. These values correspond to a constant radial load
that 90 percent of a group of nominally identical bearings can endure for 9 10
7
revolutions
without onset of surface fatigue failures.

Fig. 13.8 Shaft and housing shoulder dimensions (from Juvinall and Marshek).

13-8

Table 13.1 Bearing dimensions (from Juvinall and Marshek).


13-9

Table 13.1 continued.


13-10

Table 13.1 continued.


13-11

Table 14.2 Bearing rated capacities, C, for 90 10
6
revolution life with 90
percent reliability (from Juvinall and Marshek).

13-12
13.6 Bearing Selection
Select:
Type of bearing
Grade of precision
Lubrication and sealing
Closure (open, sealed or shielded
Load rating

Take account of:
Maximum speed. Limitation is linear surface speed, so small bearings can operate at
higher rpm than large ones.
Static load capacity in cases of heavy load at zero speed. Balls or rollers can then
indent the rings, resulting in noise.
The size of a bearing is usually influenced by the size of the shaft. However, the bearing
must have a high enough load rating to give an acceptable combination of life and reliability.
The major factors that influence load rating are now considered.

13.6.1 Life Requirement
Table 13.2 gives bearing load ratings for 90 10
6
revolutions. Applications will generally
require a different life. Manufacturers use an empirical relationship between bearing life L
and radial load F
r
. Equation 13.1 (a) and (b) is a typical relationship applied to both ball and
roller bearings.
( )
3.33
/
R r
L L C F = (13.1a)
or
( )
0.3
/
req r R
C F L L = (13.1b)
where
C = rated capacity (from Table 13.2) and C
req
= the required value of C for the
Application
L
R
= life corresponding to rated capacity (i.e. 9 10
7
revolutions)
F
r
= radial load involved in the application
L = life corresponding to radial load F
r
, or life required by the application
Doubling load reduces life by a factor of about 10.

13-13
Different manufacturers catalogues use different values of L
R
, e.g. some use 10
6
revolutions.
It is easy to show that the values in Table 4.2 must be multiplied by 3.86 to be comparable
with ratings based on a life of 10
6
revolutions.

13.6.2 Reliability Requirement
Tests show that the median life of a rolling-element bearing is about five times the standard
10 percent failure fatigue life. The standard life is commonly designated as the L
10
life.
Since this life corresponds to 10 percent failures, it also means that this is the life for which
90 percent have not failed, and corresponds to 90 percent reliability. Thus, the life for 50
percent reliability is about five times the life for 90 percent reliability.

Many designs require greater than 90 percent reliability. Fatigue lives characteristically have
a skewed distribution as shown in Fig. 13.9. Extensive experimental data has been used to
formulate life adjustment reliability factors, K
r
are is plotted in Fig. 13.10 for both ball and
roller bearings. The rated bearing life for any given reliability (greater than 90 percent) is
thus the product, K
r
L
R
. Incorporating this into (13.1) gives:
( )
3.33
/
r R r
L K L C F = (13.2a)
( )
0.3
/
req r r R
C F L K L = (13.2b)

Fig. 13.9 General pattern of bearing fatigue-life
distribution (from Juvinall and Marshek).
Fig. 13.10 Reliability factor Kr (from Juvinall and
Marshek).

13-14
13.6.3 Influence of Axial Loading
Properly aligned and lubricated, cylindrical roller bearings can carry thrust loads up to 20
percent of their rated radial capacities. The limitation is due to axial loading causing sliding
friction at the roller ends. Tapered roller bearings can carry substantial axial loads, as well as
radial loads. For ball bearings, any combination of radial load (F
r
) and thrust load (F
t
) results
in approximately the same life as does a pure radial equivalent load, F
e
, calculated from the
following equations. Load angle is defined in Fig. 13.2b. It is zero for radial bearings.
Standard values of are 15
o
, 25
o
, and 35
o
but only 25
o
is treated here.

( ) 0
o
radial ball bearings =
For 0 / 0.35,
t r e r
F F F F < < =
For 0.35 / 10, 1 1.115 0.35
t
t r e r
r
F
F F F F
F
( | |
< < = +
( |
\

For / 10, 1.176
t r e t
F F F F > =
( ) 25
o
angular ball bearings = (13.3)
For 0 / 0.68,
t r e r
F F F F < < =
For 0.68 / 10, 1 0.870 0.68
t
t r e r
r
F
F F F F
F
( | |
< < = +
( |
\

For / 10, 0.911
t r e t
F F F F > =

13.6.4 Shock Loading
Shock loading has the effect of increasing the nominal load by an application factor K
a
.
Representative values are in Table 13.3.
Table 13.3 Application Factors Ka (from Juvinall and Marshek).
Type of Application Ball Bearing Roller Bearing
Uniform load, no impact 1.0 1.0
Gearing 1.0 1.3 1.0
Light impact 1.2 1.5 1.0 1.1
Moderate impact 1.5 2.0 1.1 1.5
Heavy impact 2.0 3.0 1.5 2.0

13-15
Substituting F
e
for F
r
and adding K
a
modifies (13.2) to (13.4)
( )
3.33
/
r R e a
L K L C F K = (13.4a)
( )
0.3
/
req e a r R
C F K L K L = (13.4b)

13.7 Gearboxes
There are many gearboxes with a range of types and sizes in a power station. Almost all use
gears on shafts which are located with rolling-element bearings. The aim of this brief section
is only to illustrate the two basic configurations of gearboxes; simple and epicyclic (or
planetary). Gearboxes are sometimes called speed reducers.

Fig. 11.13 shows a triple-reduction speed reducer, with the upper half of the casing removed.
Note the use of helical and herringbone gear teeth and rolling element bearings. Note also
that the input and output shafts are parallel but not collinear.

Fig. 11.3 Triple-reduction speed reducer.

13-16
Fig. 11.14 shows a ball pulveriser which is driven by a crown wheel and pinion (bevel gear)
train to turn the drive through ninety degrees.

In some situations it is necessary for space reasons to have the input and output shafts of a
gearbox collinear. Then an epicyclic gearbox is used, such as that in the pulveriser in Fig.
11.15. In Chapter 12 it is shown how epicyclic gearboxes are used with wind turbines.

Fig. 11.14 Ball pulveriser with crown wheel and pinion drive (from Black and Veatch).

Fig. 11.15 Pulveriser with epicyclic
gearbox (from Black and Veatch).

13-17
13.7.1 Gear Trains
Simple Gear Trains
Several gears can be combined to for a gear train. Fig. 11.16 shows two simple gear trains in
which the shaft axes remain fixed; the two cases of a pinion (small) driving and external gear
(large) and an internal gear. In both cases the angular velocity ratio (gear ratio) is inversely
proportional to the number of teeth, N, indicated (and inversely proportional to the radii).
The external gears rotate in opposite directions and the internal gear rotates in the same
direction as the pinion.

Next consider the simple gear train in Fig. 11.17. Gear 2-3 is a compound gear. Again, all
three shaft axes are stationary. The angular velocity ratio is the product of a number of ratios.

( )
3 4 2 4
1 1 2 3
2 4
1 3
2 4
1 3
1

| | | || |
= +
| | |
\ \ \
| | | |
= +
| |
\ \
| | | |
=
| |
\ \

Here compound gear 2-3 gives some speed
reduction but also acts as an idler to give
the input and output shafts the same
direction of rotation.

Fig. 11.16 Simple gear trains with external and internal gears.
Fig. 11.17 Compound gear train.

13-18
Epicyclic Gear Trains
To obtain a desired gear ratio with the input and output shafts collinear, it is often
advantageous to design a gear train so that one of the gears will have a planetary motion.
With this motion, a gear will rotate about its own axis while its centre rotates about another
axis. An epicyclic gear train is shown in Fig. 11.18. Three planet gears are usually used for
balance.

For ease of analysis, consider yourself as an observer riding with the arm, but unaware of the
arms rotation. As an observer you see only two simple gear trains with angular velocity
ratios given by:
S P A
S A P
N
N

(13.5)
and
R A P
P A R
N
N

= +
(13.6)
Rotation of the planet gear,
P
is about an axis which is itself rotating. So the shaft of the
planet gear is of no use as an input or output.
P
can be eliminated by multiplying (13.5) an
(13.6) to get:
S R A
S A R
N
N

(13.7)

Fig. 11.18 Typical epicyclic gear train.

13-19
If the ring gear, sun gear or arm is held fixed (i.e. = 0), the other two can be used as input
and output with a gear ratio determined from (13.7). The other relation needed involves
either the teeth numbers, N, or the radii, r:
N
R
= N
S
+ 2N
P

r
R
= r
S
+ 2r
P
Lastly, epicyclic gear trains can be put in series to increase speed reduction, with the output
of one gear train becoming the input to the next.

MECH7350 Rotating Machinery 14. Condition Monitoring

14-1
14. CONDITION MONITORING

14.1 Introduction to Condition Monitoring Techniques
Maintenance is the management, control, execution and quality of those activities which will
ensure that optimum levels of availability and overall performance of plant are achieved, in
order to meet business objectives - The British Department of Trade & Industry (DTI) (Rao,
B.K.N.).

Maintenance strategies can be characterised as a) general purpose, b) essential and c) critical
(Scheffer and Gridhar).

a) General Purpose
Failure does not affect plant safety
Not critical to plant production
Machine has an installed spare or can operate on demand
These machines require low to moderate expenditure, expertise and time to repair
Secondary damage does not occur or is minimal

Fig. 14.1 Maintenance Strategies (from Scheffer and Gridhar).

14-2

b) Essential Equipment
Failure can affect plant safety
Machine that are essential for plant operation and where shutdown will curtail a unit
operation or part of the process
They may or may not have an installed spare available
Start-up is possible but may affect production process
High power and speed might not be running continuously
Some machines that demand time-based maintenance
These machines require moderate expenditure, expertise and time to repair

c) Critical Equipment
If their failure can affect plant safety
Machines that are essential for plant operation and where a shut-down will curtail the
production process
Machines which do not have spare parts
These machines have high capital cost, they are very expensive to repair or take a
long time to repair

14.1.1 Run-to-failure Maintenance
This applies to non essential equipment and machinery where shutdowns do not affect
production, materials and replacement are readily available. It allows the machinery to run to
failure and only repair or replace damaged components when the machine comes to a
complete stop.
Disadvantages:
Interrupt production
Large inventory of spare parts
Maintenance personnel have to work at odd time and interrupt normal activities and
tend to work overtime.

14.1.2 Preventive Maintenance
Preventive or time-based maintenance is to schedule maintenance at predetermined time
intervals, based on running hours of machines. In this case replacement of damaged

14-3
equipment is carried out before problems occur. This allows the machine to run continuously
and where the personnel have enough skill, knowledge and time to perform the preventive
work.
Disadvantages:
Performing maintenance tasks either too early or too late
Replacement of components before the end of residual life
Reduced production due to unnecessary maintenance
Possibility of diminished performance due to incorrect repair methods
Possibility good parts being disassembled and discarded and improper fixing of
replaced parts can lead to other problems

14.1.3 Condition-based Maintenance
Condition-based or predictive maintenance periodic monitoring involves periodic monitoring
on the health of the machine and scheduling maintenance only when a functional failure is
detected. This allows trends of the machine component be constructed and time to failure be
estimated. Maintenance can be conveniently planned and allows lead-time for organisation of
parts and maintenance personnel and be scheduled. This leads to full utilisation of the
machine and possible increase in production capacity.
Disadvantages:
Incorrect assessment of the deterioration of machines
Inaccurate prediction of the lead-time
Requires specialised equipment to monitor the trend and highly skilled personnel.

14.1.4 Proactive Maintenance
Proactive or prevention maintenance involves tracing all failures to their root cause and to
ensure that failures are not repeated. It utilises predictive/preventive maintenance techniques
in conjunction with root cause failure analysis (RCFA). RCFA detects and identify the cause
of failure and ensures that proper installation and repair techniques are used. It also identifies
need for redesign of machine to avoid future occurrence of the same problems and improve
the reliability of the machine.
Disadvantages:
Needs highly skill personnel with a vast knowledge of all aspects of maintenance
May require outsourcing to private consultants and problems with confidentiality

14-4
Requires specialised monitoring equipment and management support.

14.2 Introduction to Condition Monitoring
Condition monitoring and diagnostics of machines according to ISO, Sub-committee
9ISO/TC/108/SC5. The scope of this Sub-committee Standardisation of the procedures,
process and equipment requirement uniquely related to the technical activity of condition
monitoring and diagnosis of machines in which selected parameters associated with an
operating are periodically or continuously sensed, measured and recorded for interim purpose
of reducing, analysing, comparing and displaying the data and information so obtained and
for the ultimate purpose of using this results to support decisions related to the operation and
maintenance of the machine (Rao, B.K.N.).

Condition monitoring attempts to detect symptoms of eminent failure and approximates time
of a functional failure. It utilises a combination of techniques to obtain the actual operating
condition of the machines based on collected data such as vibration analysis, oil and wear
debris analysis, ultrasound, temperature and performance evaluation. The specific techniques
used depend on the type and operation of the machines.

Examples condition monitoring techniques (Scheffer and Gridhar):
(a) Vibration monitoring this is the most commonly used and effective technique to
detect internal defects in rotating machinery.
(b) Acoustic emission monitoring this involves detection and location of cracks in
bearings, structures, pressure vessels and pipelines.
(c) Oil analysis lubrication oil is analysed and the occurrence of certain
microscopic particles in it can be connected to the condition of bearings and gears.
(d) Particle analysis worn machinery components, whether in reciprocating
machinery, gearboxes or hydraulic systems, release debris. Collection and
analysis of this debris provides vital information on the deterioration of these
components.
(e) Ultrasonic monitoring this is used to measure thickness of corrosion or crack on
pipelines, offshore structures, pressure vessels.
(f) Thermography this is used to detect thermal or mechanical defects in generators,
overhead lines, boilers, misaligned coupling and cell damage in carbon fibre
structures on aircrafts.

14-5
(g) Performance monitoring this is used to determine the performance problems in
equipment. The efficiency of machines provides a good inside on their internal
conditions.

14.3 Relevant Industrial Standards
a) ISO 18436-1 Condition monitoring and diagnostics of machines Requirements for
training and certification of personnel Part 1: Requirements for certifying bodies and
certification process. This part of ISO 18436 defines the requirements for bodies operating
certification systems in no-intrusive machine condition monitoring, diagnostics and
correction technologies. General requirements for certification body personnel are contained
in this part of ISO 18436. Specific requirements for personnel in condition monitoring and
diagnostics will be contained in subsequent parts of ISO 18436.

b) ISO 18436-2 Condition monitoring and diagnostics of machines Requirements for
training and certification of personnel Part 2: Vibration condition monitoring and
diagnostics. The part of ISO 18436 defines the requirements against which personnel in the
non-intrusive machine condition monitoring and diagnostics technologies associated with
vibration analysis are to be carried and the methods of testing such personnel. Conformity
assessment for certification in vibration analysis will be performed by a body accredited to
the requirements of ISO 18436-3.

c) ISO 17359:2003(E) Condition monitoring and diagnostics of machines general
guidelines. This International Standard presents an overview of a generic procedure
recommendation to be used when implementing a condition monitoring programme and
provides further detail on the key steps to be followed. It introduces the concept of directing
condition monitoring activities towards root cause failure modes, and describes the generic
approach to setting alarm criteria, carrying out diagnosis and prognosis and improving the
confidence in diagnosis and prognosis, which are developed further in other International
Standards.

d) ISO 13379:2003(E) Condition monitoring and diagnostics of machines General
guidelines on data interpretation and diagnostics techniques. This International Standard
contains general procedures that can be used to determine the condition of a machine relative
to a set of baseline parameters. Changes fro the baseline values and comparison to alarm

14-6
criteria are used to indicate anomalous behaviour and to generate alarms: this is usually
designated as condition monitoring. Additionally, procedures that identify the cause(s) of the
anomalous behaviour are given in order to assist in the determination of the proper corrective
actions: this is usually designated as diagnostics.

e) ISO 13380:2002(E) Condition monitoring and diagnostics of machines General
guidelines on using performance parameters. This International Standard provides guidelines
for condition monitoring and diagnostics of machines using parameters such as temperature,
flow rates, contamination, power and speed, typically associated with the performance,
condition, safety and quality criteria. The evaluation of machine function may be based on
performance, condition, product quality or safety.

f) ISO 13374-1:2003(E) Condition monitoring and diagnostics of machines Data
processing, communication and presentation Part 1: General guidelines. This part of ISO
13374 establishes general guidelines for software specifications related to data processing,
communication and presentation of machine condition monitoring and diagnostics
information.

14.4 Vibration Monitoring
Vibration generated from a machine contains vital information on the health of the machine
and can be used to identify developing problems. Regular vibration monitoring can detect
deterioration or defective bearings, mechanical looseness, worn or broken gears,
misalignment and unbalance of rotor.

Fig. 14.2 Simple harmonic motion (from Scheffer and Gridhar).

14-7
All rotating machines produce vibrations that are a function of the machine operating
conditions and machine dynamics. The most classical example is that of a body with mass M
attached to a spring of stiffness K. Due to weight of mass M, the object will stabilised at an
equilibrium position at a distance x
o
. When the mass is displaced by a certain displacement
x and released, it moves up and down about the equilibrium position and reaches the top and
bottom limits. The motion can theoretically continue indefinitely if there is no damping and
is called periodic or harmonic motion. The relationship between the displacement of the mass
and time is expressed in the form of a sinusoidal equation:
X = X
0
sin t (14.1)
Where X displacement at any given time t; X
0
- maximum displacement; = frequency
(rad/s).
Velocity can be obtained by taking the first derivative of the displacement equation.
V = X
0

cos t (14.2)
Similarly, the acceleration can be obtained by taking the derivative of the velocity equation
or the second derivative of the displacement equation.
A = -X
0
2
sin t (14.3)

Fig. 14.3 Waveform of displacement, velocity and acceleration of mass in SHM
(from Scheffer and Gridhar).

14-8
Table 14.1 Some useful vibration parameters

Displacement (m) Velocity (m/s) Acceleration (m/s
2
)

Frequency (Hz) Bandwidth (Hz) Spike Energy (gSE)
Power Spectral Density Peak Value Root mean square (RMS)
Crest factor (CF) Arithmetic mean (AM) Geometric mean (GM)
Standard deviation (SD) Kurtosis (K) Skewness
Phase (deg)

Using Vibration to Machinery Fault Detection
A typical machine system is shown in Fig. 14.4. It consists of a driver, such as electric motor,
diesel engines, gas engines, steam turbines and gas turbines. The driven equipment could be
pumps, compressors, mixers, agitators, fans, blowers and others. The driven equipment is
connected to the prime mover via a gearbox, belt drive, coupling and other connectors.

Each of these rotating parts is further comprised of simple components such as:
Stator (volutes, diaphragms, diffuser, stator poles, etc)
Rotors (impellers, rotors, lobes, screws, vanes, fan blades, etc.)
Seals
Bearings
Couplings
Gears
Belts and pulleys

Fig. 14.4 A typical machinery system (from Scheffer and Gridhar).

14-9
All rotating and moving parts are prone to wear and failure after a period of service and when
mechanical defects occur, they generate high vibration levels. Some of the common faults are
listed in Table 14a.

Table 14.1a Common machine faults
Unbalance of rotating
parts
Misalignment of couplings
and bearings
Bend or bow shafts
Worn or damage gears
and bearings
Bad drive belts and
chains
Torque variations
Electromagnetic forces Aerodynamic forces Hydraulic forces
Looseness Rubbing Resonance

The causes of machinery vibration and resulting vibration characteristics can be classified in
terms of characteristics vibration frequencies and their harmonics. Table 14.2 shows the most
common causes of machinery vibration and the resulting characteristic frequencies. Table
14.3 shows possible causes of vibration from known characteristic frequencies. Some of the
common causes of bearing failure are shown in Table 14.4. It has to be pointed that these
faults are not easily identifiable and these tables are provided to be used as a reference guide.


14-10
Table 14.2 A guide to causes of vibration (from Bruel & Kjaer 2).


14-11
Table 14.3 Common faults from known vibration characteristic frequencies (from Rao, B.K.N.).


14-12
Table 14.4 Troubleshooting rolling element bearing failures (from Rao, B.K.N.).

MECH7350 Rotating Machinery 15. Sensors and Instrumentation

15-1
15. SENSORS AND INSTRUMENTATION

15.1 Vibration Parameters Acceleration, Velocity and Displacement
The form and period of vibration remain the same whether it is the displacement, velocity or
acceleration. The main difference is the phase difference between the amplitude-time curves
of the three parameters as shown in Fig. 15.1.

Fig. 15.1 Vibration Parameters Acceleration, Velocity and Displacement (from Bruel & Kjaer 2).

For sinusoidal signals, displacement, velocity and acceleration are related mathematically by
a function of frequency and time,

Displacement (x) = A sin t
Velocity (v) = A cost t
Acceleration (a) = - A
2
sin t

If the phase is neglected when taking the time-average measurements, the velocity level can
be obtained by dividing the acceleration by a factor proportional to the frequency (), and the
displacement can be obtained by dividing the acceleration by a factor proportional to the
2
.

15-2
The units for displacement, velocity and acceleration, according to ISO1000 are shown in the
table.

15.2. Choice of Measuring Parameters
Vibration measurements expressed in terms of the three different parameters are shown in Fig.
15.2 (plotted on logarithmic scale). The amplitude range required to displace the velocity
curve is smallest (about 10 to 1200 Hz). For displacement measurement, the frequency range
can go as low as 1 Hz and extended to about 500 Hz. The range can be further extended
beyond, but will depend on the characteristics of the transducer. Acceleration measurement
has the widest amplitude and frequency range and is especially useful for measuring high
frequency vibration.

Fig. 15.2 Amplitude and frequency range of vibration transducers (from Bruel & Kjaer 1).

15.2.1 Displacement
This is mainly used to measure low frequency vibration where small clearances between
elements are being considered. Displacement is often used as an indication of unbalance in
rotating machines. The vibration transducers for measuring vibration displacement are eddy
current transducers (proximity probes). These are the only transducers that provide
displacement of shaft or shaft-relative vibration measurements. Fig. 15.3 shows a typical
eddy current system, consisting of a probe, extension cables and an oscillator/demodulator.

15-3

Fig. 15.3 Operation of an eddy current system (from Scheffer and Gridhar).

Eddy currents are generated in the surface of the shaft. The oscillator/demodulator
demodulates the signal and provides a modulated DC voltage which is directly proportional
to the gap (displacement) and the AC portion is directly proportional to vibration. It can be
used for both radial vibration and distance measurements such as the axial thrust position and
shaft position.

(a) Mounting Method
On larger and more critical machines, 2 eddy current probes are normally mounted per
bearing (see Fig. 15.4).

(i) Internally mounted (ii) Internally/Externally mounted (iii) External mounted

Fig. 15.4 Mounting of eddy current probes (from Scheffer and Gridhar).

i. Internally mounted The transducers are mounted inside the machine on bearing
housing. The gap between the tip of the probe and the shaft needs to be carefully
calibrated. This allows the probe to have an unconstrained detection on the shaft

15-4
surface. However, there is no access to the probe while the machine is running
and requires extreme care to avoid cable whipping and oil leakage.

ii. Externally/internally mounted This allows the probes to be mounted with an
adaptor while the probe tip itself is inside the machine or bearing housing. This
allows probe replacement and gap adjustment while the machine is running. The
disadvantages are more machining process and long probe/stinger length may
cause resonance.

iii. Externally Mounted This is usually applied as the last resort installation where
there was no prior construction at the design stage and inadequate space available
with bearing housing and internal mounting. The advantage is that this is
relatively cheap to install. The disadvantage is that that it may record electrical
and/or mechanical run out of the shaft and requires special protection.

(b) Sensitivity
Eddy currents are sensitive to the permeability and resistivity of the shaft material and are
normally calibrated for Carbon Steel 4140. The field of the transducer extend into the surface
area of the shaft by approximately 0.4 mm and care must be taken to avoid non-homogenous
target materials such as chrome. Eddy probes are sensitive to the shaft smoothness for radial
vibration. The target area of the shaft is approximately 3 times the diameter of the probe.

(c) Calibration
All eddy current systems must be calibrated prior to commissioning. This is done using a
static calibration. The probe is installed in the tester with the target set against the probe tip.
The micrometer with the target attached is then rotated away from the probe in increments of
0.1 mm. The voltage reading is recorded and plotted as shown in Fig. 15.5.


15-5

Fig. 15.5 A typical calibration curve (from Scheffer and Gridhar).

When installed, the [robe must be gapped properly. In the figure a gap of set for -12 V DC
corresponding to an approximate gap of 1.5mm (60 mils). Proximity probes typically have a
sensitivity of 10mV/m (or 200mV/mil) with an approximate linear range of 2-2.5 mm (or
80-100 mils).

15.2.2 Velocity
This parameter is commonly used for monitoring the vibration of rotating machinery in the
frequency range between the displacement range and acceleration range. The constructions of
velocity transducers is relatively simple and have two basic types, magnet in coil and coil in
magnet, as shown in Fig. 15.6.

(a) Magnet in coil (b) Coil in magnetic

Fig. 15.6. Velocity Transducers (a) magnet in coil and (b) coil in magnet (from Scheffer and Gridhar).

The construction of magnet in coil has 3 components, a permanent magnet. A coil of wire and
spring supports for the magnet. The transducer is filled with oil to dampen the spring motion.
The construction of coil in magnet transducer had a permanent magnet surrounding the
moving coil, located in the centre cavity of the transducer. The motion of the coil is supported

15-6
by springs and damper system. The velocity transducer is a self-generating sensor and
requires no external devices to produce a voltage.

The relative motion between the magnet and the coil caused by the vibration motion produces
a voltage directly proportional to the velocity of the motion. The orientation of the spring-
mass system needs to be considered when applying to measure horizontal or vertical
vibration. The cross-axis vibration could damage the transducer.

(a) Mounting method
In general, three sensors are normally mounted to measure the vertical, horizontal and axial
vibrations. The sensors must be mounted on a flat surface with an area larger the diameter of
the sensors. The sensor can be mounted with a magnetic base for routine check up. Ideally,
the sensor is to be fixed on the machine surface using a mounting screw.

Environment factors could affect the vibration measurement include high temperatures,
magnetic interference, radiation, ground loops and corrosive materials.

(b) Sensitivity
The sensitivity of velocity transducers is dependent of inherent electrical noise. Higher output
sensitivity is required where the induced electrical noise is a problem. Sensitivities are
expressed in voltage/velocity (mm/s, m/s, etc.) and in the range of 20-50 mV/mm/s,
depending on the size of the spring/mass/damper of the system. It has a frequency range of
about 10 Hz to 1200 Hz. At low vibration frequency, the motion of the magnet/coil is
insufficient to generate sufficient output voltage and can affect the accuracy of the
measurement.

(c) Calibration
The frequency range of velocity transducers depend of characteristics of magnet/coil and
inertia of the systems. In general, most velocity transducers have a linear frequency range
from 10 to 1200 Hz. Calibration of transducer is normally carry at the commissioning stage
and follow by regular periodic calibration due to wear and tear of the moving parts.
Calibration is normally conducted using a shaker system to determine the output voltage
amplitude with varying frequency.


15-7
15.2.3 Acceleration
Acceleration transducers (or accelerometers) are the most popular transducers used in
condition monitoring of machinery. A typical construction of an accelerometer is shown in
Fig. 15.7. It consists of a mass acting on a layer of piezoelectric material and constrained in
position by a preloaded spring. The motion of the base due to a vibratory force will cause the
piezoelectric element to deflect thus producing an electric output proportional to the
acceleration of the spring/mass system.

The output from an accelerometer is dependent on the orientation and fixture of the PZT
element and the mass. The figure shows two basic configurations, the compression type and
the shear type. The former design has the mass located on top of the PZT element and the
output is due to a compressive force on the element. The shear type has the masses located
around the triangular PZT element. The motion of the masses will result in a shear force
proportional to the acceleration of the base.

Fig. 15.7. Construction accelerometer Fig 8. Mounting position for the accelerometer (from Bruel & Kjaer 2).

(a) Mounting method
Accelerometer should be mounted so that the desired measuring direction coincides with its
main sensitivity axis. They are also slightly sensitive in the transverse direction.

15-8
Table 15.1 Description of mounting positions
Accelerometer Description
A detects bearing vibration plus vibrations from other machine sources
B detects bearing vibration through modified paths plus mixed signals
from other parts of the machine.
C detects bearing vibration in the axial direction
D poor detection of bearing vibration

The method of fixing accelerometers on the machine surface plays a crucial role in detecting
the range of vibration frequencies. Poor mount of accelerometers can results in reduction of
the resonant frequency and can severely affect effective frequency range. The position on
which the accelerometer is to be mounted needs to be smooth and flat. Assuming an
accelerometer with a calibrated resonant frequency of 32 KHz, the Table 15.2 displays the
effect of resonant frequency related to accelerometer mounting types.

Accelerometers are sensitive to environmental effects, such as humidity, temperature
fluctuation, radiation, base strain (curvature surface), cable noise, magnetic interference and
acoustic noise.

Table 15.2 Mounting of accelerometer and corresponding resonant frequency
Mounting Method Resonant Frequency
Factory calibrated accelerometer on a smooth and flat surface 32 KHz
Threaded stud with tapped hole in the machine part 31 KHz
Bee-wax for sticking accelerometer. This become soft at about 40
deg C and is usable up to 100 m/s
2
29 KHz
Cemented stud can be attached to the measuring point by means
of glue.
28 KHz
Magnetic Base where an accelerometer can be attached. 7 KHz
Hand-held probe with the accelerometer mounted on the top of
the extension rod.
2 KHz


15-9
(b) Sensitivity
Accelerometer sensitivity is the output level for a given vibration acceleration. High
sensitivity is related to high output and normally entails a relatively large and heavy unit. In
practical applications, sensitivity can be increased with the application of modern
preamplifiers. In Fig. 15.9 it can be seen that the sensitivity is proportional to the
accelerometer mass. The accelerometer mass normally should be no more than 1/10 of the
dynamic mass.

Fig. 15.9 Sensitivity, mass and dynamic range Fig. 15.10 Accelerometer frequency range
(from Bruel & Kjaer 2). (from Bruel & Kjaer 2).

The dynamic range of accelerometers is crucial when measuring extreme low or high
acceleration. The lower range is determined directly by electrical noise from the connecting
cables and amplifier system. The upper range is determined by the accelerometers structure
strength. A general purpose accelerometer has an upper limit of to 100,000 m/s
2
. For the
shock measurement it may be linear up to 100,000g.

Accelerometer frequency range needs careful consideration when measuring mechanical
vibrations. General machines have a narrow frequency range between 10 Hz to 1000 Hz, but
other higher speed machines can have frequency over 10 KHz.. The frequency range is
directly related to the sensitivity of the accelerometer, see Fig. 15.10.

15-10
The lower frequency limit is about 1 Hz and is related to the amplifier cut-off frequency and
the ambient temperature fluctuations. The upper frequency limit is determined by the
resonant frequency of the spring/mass system of the accelerometer. In practical application of
accelerometer, the upper measuring frequency limit is set to about 1/3 of the accelerometers
resonant frequency.

(c) Calibration
Unlike velocity sensors, PZT accelerometers cannot be recalibrated or adjusted. The output
sensitivity of an accelerometer can change due to mishandling. It is therefore essential to
calibrate the accelerometer to determine the new sensitivity as it can be significantly
different from the original factory calibrated value.

Fig. 15.11 Calibration of accelerometer (from Bruel & Kjaer 2).

Accelerometers can be calibrated using a shaker system with constant or variable outputs and
the ratio of a reference voltage and an unknown voltage is proportional to the sensitivity. Fig.
15.11 shows a B&K accelerometer calibrator which generate a fixed periodic vibration of 10
m/s
2
. The output of the portable calibrator will produce a voltage or charge proportional to
the acceleration.

15-11
(d) Accelerometer preamplifier
Direct loading of an accelerometer output to a measuring device due to high impedance loads
can affect the accelerometers sensitivity and frequency response. A preamplifier is used to
convert the high impedance to a lower value which is suitable for analysis by measuring
equipment. This is achieved by (i) a constant current source or a regulated voltage source
preamplifier and (ii) charge mode type preamplifier. The voltage preamplifier is affected by
the cable length, as the length gets longer the output voltage gets smaller.

The charge preamplifier has a special adaptor which can be adjusted to match the impendence
of the accelerometer. Long cable length will not reduce the output of the accelerometer.

15.3 Waveform and Classification Vibration Signal
Vibration is the motion of a machine about its equilibrium position. Vibration waveforms are
described in terms of cycle, frequency, wavelength, amplitude and phase. A wavelength is
the time interval of one complete cycle and is sometime known as periodic time. A certain
number of individual waves pass a given period of time is called frequency (Hz- cycle per
second). The vibration waveform can be described in terms of displacement, velocity and
acceleration and a plot of their relationship is shown in Fig. 15.12.

Motion below 10 Hz (600 rpm) produces very little vibration in terms of acceleration,
moderate vibration in terms of velocity and relatively large vibrations in terms of
displacement. In the higher frequency range, acceleration values produce more significant
vibration level than velocity or displacement measurements.


15-12

Fig. 15.12 Relationship between acceleration, velocity and displacement at constant velocity (from Scheffer and
Gridhar.

15.3.1 Quantifying the Vibration Level
The severity of vibration is described in terms of vibration amplitude. The basic
characteristics of a periodic vibration signal are shown in Fig. 15.13. The relationship
between the peak-to-peak level, the peak level, the average level and the root-mean-square
(RMS) level is shown in the figure.

Fig. 15.13 Vibration level (from Rao, B.K.N).

15-13
a) Peak-to-peak level this value indicates the maximum excursion of the wave and is
useful for vibration displacement analysis related to maximum stress.
b) Peak level this value indicates what maximum level has occurred and takes no
account of the time history.
c) Average level this level indicates an average value over a period of time but has no
direct relationship with any useful physical quantities.
d) RMS level this gives an amplitude value which is directly related to the energy
content of the vibration.
e) Crest factor is the ratio of peak level to RMS level
f) Form factor - is the ratio of RMS level to average level

15.3.2 Vibration Limits
Vibration amplitude is a measure of the severity of the defect in a machine. A common
problem for vibration analysis is ability to judge the level of severity and establish an
acceptable level. It is not possible to set an absolute acceptable vibration level as the
operating condition, machine structure and foundation can vary from system to system. In
most condition monitoring program, the goal is to obtain a trend in vibration characteristics
that can provide warning of incipient failure. In general from historical records and skill of
personnel, it is possible to set up guidelines for vibration limits.

a) Vibration severity according to ISO 2372 (BS 4675).
This ISO Standard uses the vibration velocity (RSM level) to quantify the severity of the
range of machines. The letters A, B, C and D represents good, acceptable, still acceptable and
not acceptable respectively.


15-14
Table 15.3 ISO 2373 Guideline for machinery vibration severity (from Rao, B.K.N.).

Table 15.4 ISO 2372 machine classifications (from Rao, B.K.N.).
Class of Machines Description of Machines
Class I Individual parts of engines, machines and electric motors of up
to 15KW.
Class II Medium size machines (typically electric motors with output 15-
75 kW, without special foundations, rigidly mounted engines,
OR machine (up to 300 kW) on special foundation.
Class III Large prime movers with rotating masses mounted on rigid and
heavy foundations, which are relatively stiff in the direction of
vibration.
Class IV Large prime movers or machines with large rotating masses
mounted on foundations, which are relatively soft in the
direction of vibration measurement (eg. Turbogenerator sets,
especially those with lightweight structures).

b) American Petroleum Institute (API specifications)
The API Standards deal with turbomachines used in petroleum industry. These specifications
deal with the many aspects of machines design, installation, performance and support systems.
The specifications on vibration limits for turbo machines are widely accepted and have
developed limits for casing as well as shaft vibrations.


15-15

Fig. 15.14 Vibration limits API-610 centrifugal pumps in refinery service (from Scheffer and Gridhar).

c) IRD Mechanalysis vibration severity chart
This guideline is suitable for general machinery with operating speed beyond 300,000 rpm.
The horizontal lines represent acceleration measurement in g peak level and the diagonal
lines represent velocity measurement (in/s peak). For example, a machine operating at 60,000

15-16
rpm has an acceleration level of 2 g is considered fair. The same level is considered rough
or very rough for large machine with rotating speed less than 18,000 rpm.

Fig. 15.15 General machinery severity chart (from Rao, B.K.N.).

Condition Monitoring Lectures

Hochgeladen von

Dokumentinformationen

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

Condition Monitoring Lectures

Hochgeladen von

Copyright:

Verfügbare Formate

CONTENTS

=modulus of elasticity, or Youngs modulus (Pa)

=modulus of rigidity, or shear modulus (Pa)

can be a force which has a scalar magnitude F and a direction of action.

indicates the sum of vectors, i.e.,

can also be regarded as vector quantities,

= angular velocity, rad/s

which may or may not be equal to

= = density = frequency, rad/s

in left plane at a known angular position and

and add known

we can substitute into (5.1) and (5.2) to find the

Das könnte Ihnen auch gefallen