ACHIN ERY

RLLOGA N, JR
THE PENNSYLVANIA
STATE UNIVERSITY
LIBRARIES
PEROT RIMIMOTEUARA O1APE
UNIVERSITY | IBRARIFS
 Digitized by the Internet Archive
in 2022 with funding from
Kahle/Austin Foundation

https://archive.org/details/turoomachineryba0O000loga
Turbomachinery
 MECHANICAL ENGINEERING
A Series of Textbooks and Reference Books

EDITORS
L. L. FAULKNER S. B. MENKES
Department of Mechanical Engineering Department of Mechanical Engineering
The Ohio State University The City College of the
Columbus, Ohio City University of New York
New York, New York

Spring Designer’s Handbook, by Harold Carlson

Computer-Aided Graphics and Design, by Daniel L. Ryan
Lubrication Fundamentals, by J. George Wills
ke
Sag
Cae Solar Engineering for Domestic Buildings, by William
A. Himmelman
Applied Engineering Mechanics: Statics and Dynamics,
by G. Boothroyd and C. Poli
Centrifugal Pump Clinic, by Igor J. Karassik
Computer-Aided Kinetics for Machine Design, by
Daniel L. Ryan
Plastics Products Design Handbook, Part A: Materials
and Components, edited by Edward Miller
Turbomachinery: Basic Theory and Applications, by
Earl Logan, Jr.

OTHER VOLUMES IN PREPARATION

Turbomachinery
BASIC THEORY AND APPLICATIONS

Earl Logan, Jr.

Arizona State University
Tempe, Arizona

MARCEL DEKKER, INC. New York and Basel

 Contents

Preface

Types of Turbomachines

1.1 Introduction
1.2 Geometries
1.3 Practical Uses

Basic Relations

2.1 Velocity Diagrams

2.2 Mass Flow Rate
2.3 Energy Equation
2.4 Momentum Equation
2.5 Applications
References
Problems

Dimensionless Quantities

3.1 Introduction
3.2 Turbomachine Variables
3.3 Similitude
Reference
Problems
4 Centrifugal Pumps and Fans 31

4.1 Introduction ot
4.2 Impeller Flow By
4.3 Efficiency 34
4.4 Performance Characteristics 35
4.5 Design of Pumps 38
4.6 Fans 41
References 43
Problems 43

Centrifugal Compressors 45
5.1 Introduction 45
5.2 Impeller Design 47
5.3 Diffuser Design 49
5.4 Performance 50
References ae
Problems 53

Axial-Flow Pumps and Fans ae)

6.1 Introduction 55
6.2 Stage Pressure Rise 57
6.3 Losses 60
6.4 Pump Design 61
6.5 Fan Design 65
Reference 66
Problems 66

Axial-Flow Compressors 67

7.1 Introduction 67
7.2 Basic Theory 68
7.3 Cascade Tests 69
7.4 Performance 71
References 73
Problems 74
 8 Gas Turbines 75

8.1 Introduction is
8.2 Basic Theory 77
8.3 Design 81
8.4 Radial-Flow Turbine 85
References 86
Problems 86

9 Steam Turbines 87

9.1 Introduction 87
9.2 Impulse Turbines 88
9.3 Reaction Turbines 93
9.4 Design 95
References 95
Problems 95

10 Hydraulic Turbines 97

10.1 Introduction 97
10.2 Pelton Wheel 98
10.3 Francis Turbine 100
10.4 Kaplan Turbine 102
10.5 Cavitation 103
References 104
Problems 104

11 Wind Turbines 105

11.1 Introduction 105

11.2 Actuator Theory 106
11.3 Horizontal-Axis Machines 110
11.4 Vertical-Axis Machines 114
References 115
Problems lS)

Index 117
Turbomachinery
 l Types of Turbomachines

1.1. Introduction

Turbomachines constitute a large class of machines which are found virtually

everywhere in the civilized world. This group includes such devices as pumps,
turbines, and fans. Each of these has certain essential elements, the most impor-
tant of which is the rotor, or rotating member. There is, of course, attached
to this spinning component a substantial shaft through which power flows to
or from the rotor, usually piercing a metallic envelope known as the casing.
The casing is also pierced by fluid-carrying pipes which allow fluid to be ad-
mitted to and carried away from the enclosure bounded by the casing. Thus a /

turbomachine always involves an energy transfer between a flowing fluid and

a rotor. If the transfer of energy is from rotor to fluid, the machine is a pump,
efan, or compressor; if the flow of energy is from the fluid to the rotor, the
machine is a turbine.
The purpose of the process described above is either to pressurize the
fluid or to produce power. Useful work done by the fluid on the turbine rotor
appears outside the casing as work done in turning; for example, it can turn
the rotor of a generator. A pump, on the other hand, receives energy from an
external electric motor and imparts this energy to the fluid in contact with
the rotor, or impeller, of the pump.
The effect on the fluid of such devices is that its temperature and pres-
sure are anereased by apum umping-ttype turbomachine, andd the same properties
2 TYPES OF TURBOMACHINES

are reduced in passage through a work-producing turbomachine. A water pump

might be used to raise the pressure of water, causing it to flow up into a reser-
voir through a pipe against the resistance of frictional and gravitational forces.
On the other hand, the pressure at the bottom of a reservoir could be used to
produce a flow through a hydraulic turbine, which would then produce a
turning moment in the rotor against the resistance to turning offered by the
connected electric generator.

1.2 Geometries

A typical turbomachine rotor, a centrifugal pump impeller, is shown schemati-

cally in Figure 1,1. Liquid enter the eye E of the impeller moving in an axial
direction, and then turns to a radial direction to finally emerge at the discharge
D having both a radial and a tangential component of velocity. The vanes V
impart a curvilinear motion to the fluid particles, thus setting up a radial cen-
trifugal force which is responsible for the outward flow of fluid against the re-_
sistance of wall friction and pressure forces.
The vanes of the rotor impart energy to the fluid by virture of pressure
forces on their surfaces, which are undergoing a displacement as rotation takes
place. Energy from an electric motor is thus supplied at a constant rate through
the shaft S which is assumed to be turning at a constant angular speed.
If the direction of fluid flow in Figure 1.1 is reversed, the rotor becomes
part of a turbine, and power is delivered through the shaft S to an electric
generator or other load. Typically, hydraulic turbines have such a configuration
(see Figure 1.4) and are used to generate large amounts of electric power by ad-
mitting high-pressure water stored in dams to the periphery of such a rotor. A

rotation
i? 2 4D

Figure 1.1 Pump impeller.

 GEOMETRIES 33

pressure drop occurs between the inlet and the outlet of the turbine; the water
exits axially and is conducted away and discharged at atmospheric pressure.
If the substance flowing through the impeller of Figure 1.1 were a gas,
then the device would be a centrifugal compressor, blower, or fan, depending
on the magnitude of the pressure rise occurring during transit from inlet to
outlet. For the reversed flow case, i.e., a radially inward flow, the machine
would be called a radial-flow gas nuine or turboexpander.
A different type of turbomachine is shown in Figure 1.2. Here the flow
. direction is generally axial, i.e., parallel to the axis of rotation. The machine
shown in this figure represents an axial-flow compressor or blower, or with a
different blade shape an axial-flow gas or steam turbine, depending on the
direction of energy flow and the kind of fluid present.
In all of the machines mentioned thus far, the working fluid undergoes
a change in pressure in flowing from inlet to outlet, or vice versa. Generally,
pressure change takes place in a diffuser _oor nozzle, and in the rotor as well.
However, there is a class of turbines in which pressure change does not occur in
the rotor. These are called impulse, or Zero-reaction, turbines, as distinquished
from the so-called reactionn turbine, which allows a pressure decrease in both
nozzle and rotor. A hydraulic turbine with zero reaction is shown in Figure
1.3, and a reaction-type hydraulic turbine appears in Figure 1.4.

Rotor vanes

Stator vanes

Figure 1.2 Axial-flow blower.

Water nozzle

vane :

Figure 1.3 Pelton wheel.

4 TYPES OF TURBOMACHINES

Centrifugal machines are depicted in Figure 1.5 through 1.7, and axial-
flow turbomachines are indicated in Figures 1.8 through 1.10. A mixed-flow
pump is shown in Figure 1.11. This class of machine lies part way between
the centrifugal, or radial-flow, types and the axial-flow types.

Rotor vanes

Stationary
guide vanes

| Draft tube

‘Tail race

Figure 1.6 Centrifugal compressor.

 GEOMETRIES

Casing

Figure 1.7 Centrifugal blower.

Guide vanes

Figure 1.8 Kaplan turbine.

Wi ‘Casing

Steam —>

SY
Rotor Rotor Axis

Figure 1.9 Steam turbine.

6 TYPES OF TURBOMACHINES

Rotor blades
Stator blades

Flow ——>

Jf Rotor axis

Figure 1.10 Axial-flow compressor.

Impeller
Stator vanes

eae

ee Rotor axis

Figure 1.11 Mixed-flow pump.

Sizes vary from a few inches to several feet in diameter. Fluid states vary
widely as well. Steam at near-critical conditions may enter one turbine, while
cool river water enters another. Room air may enter one compressor, while cold
refrigerant is drawn into a second. The materials encountered in the machines
are selected to suit the temperatures, pressures, and chemical natures of the
fluids handled, and manufacturing methods include welding, casting, and
machining.
Our consideration herein of the subject of turbomachines includes a
wide variety of forms and shapes, made of a variety of materials using a num-
ber of techniques. This book does not attempt to deal with all the problems
encountered by the designer or user of turbomachines,
but only with the most
general aspects of the total problem. The present treatment is concerned with
specification of principal dimensions and forms of those turbomachines en-
countered frequently in industry.
PRACTICAL USES 7

1.3 Practical Uses

The importance of the turbomachine to our way of life cannot be overem-

phasized. The steam power plant, which is responsible for the generation of
most electrical power in the world, can be used to illustrate this basic fact. The
steam power plant consists of a prime mover driving a large electric generator.
A steam turbine is usually used as the prime mover. Steam for the turbine is
supplied from a boiler at high pressure and temperature. Water for making the
steam is forced into the boiler by means of a multiple-stage centrifugal pump.
Fuel for creating the heat in the boiler is supplied by a pump, compressor, or
blower, depending on the nature of the fuel. Air for combustion of the fuel
enters the boiler through a large centrifugal fan. After the steam has been
generated in the boiler and has expanded in the turbine, it is exhausted into a
condenser where it is condensed and collected as condensate. Pumps are used to
remove the condensate from the condenser and deliver it to feedwater heaters,
from which it is drawn into the boiler feed pumps to repeat the cycle. The
condensation process requires that large amounts of cooling water be forced
through the tubes of the condenser by large centrifugal pumps. In many cases
the cooling water is itself cooled in cooling towers, which are effective because
large volumes of outside air are forced through the towers by axial-flow fans.
Thus we see that many turbomachines are required to operate the simplest
form of modern steam-electric generating station. It is clear that modern in-
dustry and the entire economy depend upon such generating stations, and
hence we are all dependent upon turbomachines in this and in many other
applications.
Before we consider the specifics of pumps and turbines, we must deal
with the fundamentals underlying their design and performance. In Chapters
2 and 3 we will develop these underlying principles of all turbomachines by
starting our discussion with the first principles of physics, i.e., conservation
of mass, momentum, and energy.
 : Pa
Ob ot ae oe -¢
‘ " iow es v i ae ;
' a 7 Ve Wwe - -:
it ir
a if a ht an nn ; z
+ ents Ly ys Poet ‘)
Wis Givi pa re yak i
#7) mesh aes: Real ey oma a
a 5 enh yoy Re ge; ot ha OS
bald bP PN aD -Dicer ais
eee ae
Ne eatr ?

1 OD EA ee
» -
ee
a
ee
z

Help
gtae
7
a 00d gy
bg aii ce Sa he “i
: iC, aie ie yas re aeMes

MARTA Age ye SEE 1 oie bepanera

ne : i
- :
~oe
.
Opt her iy
ayi es.18 :
1) PS Lar
=
fe, | eee
i ;
ae. eo 7
a Ae oe ; si a el ri A 8 2 AG] Vig? \ nic

j Des ae Decay) i
it ~~ ae TY ie % q~ : ' ' , SOL mi
J
se ee ee
.
7 7 : 7

% re _ ‘ ’ a bd. of
rc j ; :
A NAb Uibge Cee he |
7 i= rie ihe dilice nade U , Seth
ml r = a. © 4 7 we = eae ow

, .
'

: May i eaaes

' ;a]
: ! iy 1 Le ’ beep.
i P s

— 1
i ye ae

ty
.

oP re

, od “ 7
2

wi), | a /
re g ce " —

¥ ea
ee w ve we ‘

{pa t\ , na e a ;
- 7
ha ‘Gl be a
oa :
' a x -
 Z Basic Relations

2.1 Velocity Diagrams

The rotor shown in cross section in Figure 2.1 will have fluid flowing in the
annulus bounded by abcda. Although fluid velocity varies radially from a tob,
it is assumed to have a single value over the entire annular section ab, namely,
the velocity V, at point 1. Similarly, at the rotor exit the velocity V, is taken
as the average of the velocity along cd. Points 1 and 2 lie on the line 1-2 which
denotes an element of a stream surface which exactly divides the flow into
two equal parts.
Figure 2.2 shows a velocity diagram at point 1. The blade, or vane, ve-
locity is calculated from

U=Nr (2.1)

for any point on the blade a radial distance r from the axis A-A of rotation.
The preilat ee vee of the rotor is denoted by N. For point 1,(2.1)
becomes ,

U,= NG (2.2)
2.2 Mass Flow Rate

The relative velocity W of the fluid, with respect to the moving vane, is added
vectorially to the blade velocity U to obtain the absolute fluid velocity V. The
relation can be expressed by
 10 BASIC RELATIONS

Figure 2.1 Turbomachine rotor.

V=Wt+U (2.3)

The graphical representation of the addition of U, to W, is shown in Figure 2.2.

The corresponding velocity diagram at the rotor outlet is shown in Figure 2.3.
ee The mass flow rate m through the rotor is calculated by multiplying the
“ meridional velocity V,, by the area (normal) of the flow passage and by the
fluid density. For example, assuming that V,, is parallel to the rotor axis at
the rotor inlet,

m = 20, 71yhy Ving (2.4)

When the flow direction is at an angle to the rotor axis, a more complicated
expression is obtained, but the same principle applies. The general form of
. (24), applying to any station in the flow passage, is

m=pV,A (2.5)
where A is the area normal to the flow direction. Equation (2.5) is a'statement
of conservation of mass, i.e., the mass flow rate is the same at all stations.
It is assumed that each velocity on the central streamline of the flow
passage depicted in Figure 2.1 is the average value for the entire flow area at

Figure 2.2 Velocity diagram at inlet.

ENERGY EQUATION 11

Ue

Figure 2.3 Velocity diagram at outlet.

the position considered. The actual flow has a variable velocity across the pas-
sage. The variation can, of course, be handled mathematically through the use
of the integral form

m= f,PVm dA (2.6)
The latter form also allows for possible variations in density associated with
temperature, pressure, or concentration gradients.

2.3 Energy Equation

The principle of conservation of mass, expressed in (2.5), must be supplemented

by a steady-flow energy equation which expresses the conservation of energy.
The usual forms of energy per unit mass, which must be accounted for in a
turbomachine, are potential energy zg, internal energy e, flow work p/p, kine-
tic energy V/2, heat transfer q, and work w..A word statement of the energy
equation is the following:

Energy at section 1 + heat transfer = energy at section 2 + work done

between | and 2°

In equation form this is written as

2 2
Pt a. Po oy
zgte, +—+— +q=2ete,+—+—+w (2.7)
Pye, 2 Po

Such an equation has been derived in many thermodynamics treatises, such as

that by Jones and Hawkins (1960). Frequently, the internal energy is com-
bined with the flow work to form the enthalpy h. The equation is then

Pie
Vi
age cE NG
V2
ev (2.8)
12 BASIC RELATIONS

Usually, in turbomachinery applications the potential energy and the heat

transfer terms are neglected, and the specific work is denoted by E and called
energy transfer, with the result

i
h, +— =h, +—-+EV2 (2.9)
2 2D

In gas turbine or compressor applications the enthalpy and the kinetic energy
are combined to form the total enthalpy hy. Thus, (2.9) becomes

hy, = hg tE (2.10)
Compressors and pumps increase hy so that hy, >ho, , and the energy
transfer E is negative. On the other hand, turbines decrease hp, and E is posi-
tive. The work per unit mass calculated from (2.10) is also termed the head,
especially in pump or hydraulic turbine applications.

2.4 Momentum Equation

The moment of momentum equation is of particular interest in turbomachinery
applications. In its general form it states that the sum of the moments of ex-
ternal forces on the fluid in a control volume equals the rate of increase of
angular momentum in the control volume plus the net outlfow of angular
momentum from the control volume. Allen and Ditsworth (1972) give this in
equation form as
ae

2M = Saat nS ds Ga)
icanayWS tend'y ;

where c.v. and c.s. refer to intergration over the control volume or control
surface.
Applying (2.11) to a general turbomachine, the control volume is the
volume of fluid in the casing surrounding the rotor. Forces are applied to this
fluid along the surface of the rotor, and the sum of their moments about some
point on the rotor shaftis denoted by the term on the left side of (2.11). Assum-
ing steady flow through the control volume, the first term on the right side of
(2.11) is eliminated. Noting that the quantity pV ° dS is the mass flow rate
through an elemental area dS of the control surface, and that it has a positive
sign at the outlet, a negative sign at the inlet, and is zero elsewhere, we have

2M = fa Rx V dm - fy Rx V dm (2.12)
where A, and A, refer to the flow areas at the inlet and outlet, respectively.
MOMENTUM EQUATION 13

Figure 2.4 Velocity components.

Aligning the z axis with the rotor axis and taking the moment center at
0, as indicated in Figure 2.4, we evaluate the angular momentum term in (2.12)
by the determinant
itan) _
io]
Rx V=|r O
Vo V.

The magnitude of the momentum about the rotor axis, i.e., the z axis is the
axial component, or z component, of the moment about the origin. Thus,

Ora") et
I Do ba wesc 1) a
V; Vu Ns

and the resulting scalar expression is

M, = Sa, Vut dm - fy Vyr dm (2.13)

Neglecting all forces other than those between the fluid and the rotor, we can
say that the magnitude of the moment of forces M, on the fluid in the control
volume equals the negative of the torque T applied to the rotor shaft by the
fluid. If we further assume constancy of the tangential component of the fluid
velocity V,, and of the radial position r overr
the area A, or A,, we can then
write

Turbomachine power is torque times the rotational speed N in radians

per second. Thus power P can be expressed as

P=m(Vy,Uy - Vy2 U2) (2.15)

 14 BASIC RELATIONS

Here we note that the blade speed U has been substituted for Nr. To obtain
the energy transfer E per unit mass corresponding to that in (2.7) through
_ (2.10), we simply divide (2.15) by the mass rate of flow m. Thus, the energy
transfer per unit mass from fluid to rotor, or vice versa, is given by

Ba VelU Vass (2.16)

The latter relation is the Euler turbine equation, but it is applied to all types
of turbomachines, including pumps and compressors.

2.5 Applications

Let us apply the above relations to a number of common turbomachines,

namely, the axial-flow impulse turbine, the axial-flow compressor,
the centri-
fugal pump, and the hydraulic turbine.

Ww Impulse Turbine

Flow in the impulse turbine is generally in the axial direction, and the blade
velocity is the same at the entrance and exit of the rotor. Figure 2.5 shows a
typical blade cross section and the corresponding velocity diagram. Steam or
hot gas leaves a nozzle with a velocity V, at a nozzle angle a, measured from
the tangential direction, and enters the region between the blades with rela-
tive velocity W,. Ideally, no pressure drop occurs in the blade passage, and
the relative velocity W, is equal in magnitude to W, . This is what is meant by
the term impulse turbine, also called a zero-reaction turbine. The absolute
velocity V, at the blade-passage exit is much reduced and is typically less
than half of V,. This energy, transferred from the fluid to the rotor, is found
from (2.16) by making the substitutions V,, = V, sin a,, Vy. =V, sin As ,
and UO = U>i= U; Thus,

MW,

a V :

NX U U
Figure 2.5 Velocity diagram for an impulse turbine.
 APPLICATIONS 15

E=U(V, sina, - V, sina,) (Qe lm)

The law of cosines applied to the two triangles in Figure 2.5 yields two equa-
tions, which when subtracted contain the right-hand side of (2.17). Substitu-
tion into (2.17) yields

| B= (V2 - v3 + Ww? - w?) (2.18)

Since W, = W,, it is clear that E is really the difference in kinetic energy of
the fluid, namely,

E = %4(V? - V3) (2.19)

Maximizing energy transfer means minimizing V, or requiring that V, has only
leeaxial component, i.e., a, = 0.
The result of (2.19) is also obtainable from (2.9), ifh, =h,. Equal en-
thalpy implies no change of termperature and pressure in the flow, which
agrees with the original assumption of zero reaction, i.e., no pressure drop in
the rotor. If the degree of reaction of a turbine stage, denoted by R, is defined
as the ratio of the enthalpy drop in the rotor to the enthalpy drop in the
stator plus that in the rotor, namely,

h, -h
Re ee (2.20)
he - hy

where h, is the enthalpy at the nozzle (stator) inlet, then we can write

Ve NE
WES Mesa fata tz (2.21)

as the energy balance for the entire stage, and R becomes

V2=Nile 7B
aap 2 CP)
V2 /2-V2/2+E
Substituting (2.18) into (2.22) yields
WwW? - Ww?
Re PG)
ee ae 2
(2.23)
Vie Veo Wit
which is generally applicable to axial-flow machines. Quite commonly, in the
analysis of multistage machines it is assumed that the fluid velocity V, leav-
ing the rotor is the same as that from the stage immediately upstream, i.e.,
V, = V,. The degree of reaction would then be expressed as
16 BASIC RELATIONS

Figure 2.6 Velocity diagram for an axial-flow compresser.

Ww? - w?
Riek Cceare sees (2.24)
v2 - v2 + w2 - w?
which we will utilize for axial-flow machines.
We have learned that the blade profile of the impulse turbine is designed
to make W, = W,. Clearly, (2.24) confirms the earlier assumption that R = 0.

Axial-Flow Compressor

An axial-flow compressor blade and velocity diagram are shown in Figure 2.6.
The fluid is deflected only slightly by the moving blade compared with the
turbine-blade deflection. Another difference is that the pressure rises in the
flow direction both in the stator and in the rotor. Pressure rise is related to
enthalpy rise, and the latter is dependent on the deflection of the fluid by the
moving blade. A relationship is obtained by eliminating E between (2.9) and
(2.18). This yields

h, -h, = 4(W? - W3) (2.25)

which is interpreted as in enthalpy rise associated with a loss of relative kinetic
energy in the blade passages. The associated pressure ratio is easily obtained
from the enthalpy rise through the use of a polytropic exponent n. Using the
perfect gas relation Ah = C,, AT, we obtain
By (n-1)/n
hy -hy =C,T, (7) -1 (2.26)

from which the pressure ratio, and hence the pressure rise, may be determined.
It is observed that pressure rise depends on the change of relative velocity,
which is directly related to the compressor blade shape, i.e., to the angle of
deflection of the fluid.
APPLICATIONS 17

Figure 2.7 Tangential components of relative velocity.

Energy transfer is also related to the deflection angle, since the applica-
tion of (2.3) gives

E=U(Vyy - Vu) = UW, - Wyo) (2.27)

Figure 2.7 shows that the difference in the tangential components of the rela-
tive velocity is proportional to the deflection angle @ =, - B,.
A typical compressor velocity diagram is constructed by making V, =
W, and V, = W,. Referring to Figure 2.6, it is seen that the triangles would
be symmetrical about the common altitude (V,). Such symmetry, whether in
a turbine or compressor diagram, results in R = 1/2, as determined from
(2.24). This condition is also terms a SO percent reaction. Physically this
means that 50 percent of the compression (or enthalpy rise) takes place in the
rotor of the compressor and 50 percent in the stator.
This degree of reaction is optimum for minimization of aerodynamic
drag losses of rotor and stator blades in both turbines and compressors.

Centrifugal Pump

The centrifugal pump is and has been an extremely important machine to

humans, and one would think it theoretically complex. However, it is ex-
tremely simple to analyze. It was discussed in Chapter 2 and is illustrated in
Figure 2.8. The inner and outer radii r, and r, define the inlet and outlet of
the control volume. Fluid, assumed incompressible, enters at station 1 with a
purely radial velocity V,, which implies that V,, = 0. The impeller imparts
angular momentum to the fluid so that it exits at station 2 with radial and
tangential velocity components. Note that U, > U, since r, > 1, and the
angular speed is constant. The energy transfer E, or the head H (gH = -E), is
calculated from (2.16) as
gH =-E= U,V, OID)
18 BASIC RELATIONS

—H ake
SANG
By =
FG

Figure 2.8 Centrifugal pump.

From Figure 2.9,

Viz = Un - Vinz tan By (2.29)

The meridional component V,,, is the volume flow rate Q divided by the
flow area 271, b, , and U, = Nr,. The head is thus expressed as

Pe Nr, [Nr, - (Q/2mr,

b, )tan By |
(2730)
g

where b, = impeller tip width.

An important performance curve, the head-capacity curve, for a centri-
fugal pump is constructed by plotting H as a function of Q. Equation (2.30)
expresses this relationship analytically and provides an ideal head-capacity
curve for comparison with actual curves. Since B, is usually about 65°, the
theoretical relation indicates decreasing head with increasing flow rate, a situa-
tion realized in practice. This equation indicates that H goes up as the square
of N, which also agrees with experience.
It is interesting to note that an actual pump impeller can be measured,
and the measurements used to predict expected flow rate. Figure 2.9 shows
that such a prediction can be easily made from a knowledge of B, , N, 1, , and
b, , since

Q = 2mr7b,N cot B, (2.31)

The enthalpy rise for a compressible fluid is determined from the thermo-
dynamic equation

Weds oe (2.32)
p
APPLICATIONS 19

ie ws
W, Vo Wo

: i
Figure 2.9 Velocity diagram for a centrifugal pump.

Equation (2.32), integrated for the ideal isentropic compression of a liquid

(for which the density is assumed constant), is

b= hy = Po :- Py (@333))

Substituting (2.33) into (2.9) gives

2 2
195) = 12) feWes
I= (2.34)
pg 22
The latter equation is useful in calculating the pressure rise across the pump im-
peller. Of course, the pressure can be raised further in the casing of the pump by
reducing V, in a passage of increasing cross-sectional area, i.e., a diffuser.

Hydraulic Turbine

The radial-flow hydraulic turbine, as depicted in Figure 2.10, the reverse of

the centrifugal pump. Stations 1 and 2 are reversed, and water enters at the
larger radius r, from a stator which controls the angle a, at which the water
enters the rotor inlet. Ideally, the absolute velocity V, at the exit is purely
radial so that the energy. transfer is simply

E=gH=U,V,, (2.35)
Since V,,, = Vin, tan a, and V,,, = Q/A,, we can write the turbine head as

H = Nr, ace tan a, (2336)

271, b,8

where b, = runner tip width. The flow rate Q is given, as in (2.31), by

Q = 2n13b,N cot By (2.37)

20 BASIC RELATIONS

Figure 2.10 Hydraulic turbine.

Further Examples

Similarities of other turbomachines to the four examples discussed above

should be noted. The axial-flow reaction turbines, which includes most steam
and gas turbines, are like the impulse turbine example given except that an
expansion of the fluid also occurs in the rotor. This means that an enthalpy
drop occurs there, and the degree of reaction R is greater than zero (typically
R = 4). It should also be noted that steam and gas turbines used to drive
large loads, such as electric generators, include many stages in series, frequently
with many rotors mounted on a single shaft. The energy transfer term for
each rotor (stage) must be added to obtain the total work done per unit mass
of steam or gas flowing. The turbine power is then obtained from the product
of the total specific work and the mass flow of fluid.
The axial-flow compressor example indicated calculations for a single
stage. Compressors usually involve many stages and the pressure ratios for
each must be multipled to obtain the overall pressure ratio of the machine.
In addition, the relations developed for the compressor stage would also
apply to axial-flow blowers, fans, and pumps. The difference is that the en-
thalpy rise is calculated from (2.33) for the approximately incompressible
flows usually assumed in these machines.
The centrifugal-pump example is applicable to centrifugal compressors,
blowers, and fans. The hydraulic turbine example has points of similarity
with the radial-flow gas turbine. No attempt is made, however, to cover all
turbomachines in this chapter. A detailed coverage will be given to each type
in subsequent chapters.
PROBLEMS 74)

References

Allen, T., and R.L. Ditsworth. 1972. Fluid Mechanics. McGraw-Hill, New York.
Jones, J.B., and G.A. Hawkins. 1960. Engineering Thermodynamics. John
Wiley & Sons, New York.

Problems

2.1 Construct the velocity diagram for an axial-flow gas turbine having a
degree of reaction of 0.25 and minimum leaving kinetic energy «v3 /2).

2.2 Derive the relationship between torque and speed for an axial-flow im-
pulse’ turbine.
2.3 Determine energy transfer E for an axial-flow turbine in terms of blade
speed U when the degree of reaction is 0.5 and the leaving kinetic energy is
minimal.

2.4 Repeat Problem 2.3 for the impulse turbine with minimal leaving kine-
tic energy.

2.5 Sketch the head-capacity curves for centrifugal pumps having B, be-
tween 0° and 90°, equal to 0° and less than 0°.
 3 Dimensionless Quantities

3.1 Introduction

In plotting the results of turbomachinery tests and in the analysis of per-

formance characteristics, it is useful to use dimensionless groups of variables.
Appropriate groups of variables are found by application of a dimensional
methodology (dimensional analysis), and it is known from the Buckingham pi
theorem that the dimensionless groups so formed have a functional relationship,
although the nature of the relationship is frequently unknown except by experi-
mentation. An important benefit of dimensional analysis is that the results of
model studies so analyzed and plotted may then be used to predict full-scale per-
formance. This is important to reduce the cost of the development of turbo-
machines. It is also useful to use in the analysis of data from full-sized machines,
when it is desired to predict the performance of other full-size machines of a
different size than those tested, or to operate under different conditions.

3.2 Turbomachine Variables

The important variables in turbomachine performance are shown in Table 3.1.

The Buckingham pi theorem applied to the four variables and two dimensions
indicates that two dimensionless groups can be formed. Of the several possible
groups that can be formed, the most useful combinations of variables are the
flow coefficient ¢defined by

=<, (3.1)

23
24 DIMENSIONLESS QUANTITIES

Table 3.1 Primary Turbomachine Variables

Variable Symbol Dimensions

Head or energy transfer gH (BE) L? /T?

Volume flow rate Q L3/T
Angular speed N WAL
Rotor diameter D 1L,

and the head. coefficient y defined as

peFp?
a G2)

where E = gH has been used in the analysis.

Three other groups are also used extensively by engineers. However, they
may be easily derived from ¢ and . Specific speed N, is formed in the follow-
ing way: a

Ss
Bi tat (gH)3/4
p34
NOS (3.3)

As @ and y are nondimensional flow rate and head, so N, is nondimensional

speed. In fact, if Q and H are unity, we observe that N, = N. The other parti-
cularly useful dimensionless groups are the specific diameter D, and the power
coefficient C,, as defined below:

ee Dany
st gil : Qi Co
1p
Cy ? ov im pN3D5 (3:5)

where power P = pQgH.

Recall that four variables are assumed to be of primary importance. If
additional variables are added to the list in Table 3.1, then a new dimension-
less group can be formed which will contain each, i.e., an additional group for
each new variable. For example, if kinematic viscosity v is added, we have the
Reynolds number Re, defined as

Re= = (3.6)

Another example is the inlet fluid temperature T,, or the inlet specific enthalpy
h,. Since the latter quantity contains the square of the acoustic speed in gases,
SIMILITUDE 25

we would expect the Mach number M to emerge as the appropriate dimension-

less group, namely,

M ENDS
ie (3.7)
Soul

This is an important variable in turbomachines involving high-speed flow of

gases.
If the number of variables is increased to seven by adding the inlet pres-
sure p,, the nature of the dimensionless groups is changed but not the number.
This is because pressure involves the force dimension, which is not present in
the others. Thus, the number of groups remains four. However, since density
p, can be introduced as a combination of p, and T, (from h, ), it appears in
the groups; for example, in combination with Q as the mass flow rate m. In
this form the flow coefficient can be written as

$= (3.8)
Vip, D?
Other forms of the head coefficient are the ratio of outlet pressure to
inlet pressure p,/p,, or the ratio of outlet to inlet temperature T, /T, . Clearly,
these ratios are equivalent, since head is proportional to enthalpy difference,
which in turnis proportional to temperature difference in gas-flow machines
and pressure difference in incompressible-flow machines. In gas-flow turbo-
machines either p,/p, or T,/T, could be used, since the two are related
through isentropic or polytropic process relations.
Efficiency n has many specialized definitions, but is, in general, output
power divided by input power. It, too, can be included in the list of variables,
and since it is already dimensionless, it is also included in the list of dimen-
sionless groups.

3.3 Similitude

Flow similarity occurs in turbomachines when geometric, kinematic, and dyna-

mic similarity exists between a model (i.e., a small-size turbomachine) and its
larger prototype. Thus, ratios of dimensions of corresponding parts are the
same throughout. Velocity triangles at corresponding points in the flow fields
are also similar, as are ratios of forces acting on the fluid elements. Similar
velocity triangles, for example, imply equal flow coefficients:

(0), "Sn
co CAR
-3.9
26 DIMENSIONLESS QUANTITIES

On the other hand, similar force triangles are equivalent to equal head coeffi-
cients:

(x>),
eH \ = ny a
(Ss
D2
(3.10)

The equality of dimensionless groups resulting from similitude has im-

portant practical consequences. It allows a most compact presentation of
graphical results. One example of this is seen from a consideration of head-
capacity curves for centrifugal pumps, which typically appear as shown in
Figure 3.1. A separate curve is needed for each shaft speed when plotting the
primary variables. On the other hand, if head coefficient is plotted against
flow coefficient, the curves collapse to a single curve, and a single relation-
ship exists between y and ¢, regardless of speed.
The so-called pump laws also follow from the similarity conditions ex-
pressed by (3.9) and (3.10). When it is desired to know how a given pump
will perform at another speed when its performance at one speed is known,
we simply cancel the D? factors in (3.9) and find

Seo
Cre
NG
22
Sot
(3.11)

which is a pump law; it implies that capacity Q varies directly with speed N.
In a similar manner, we see from (3.10) that the head H or pressure rise is
governed by

H, =eHy
2) 2 (3,12)
NUNS

Figure 3.1 Head-capacity curves.

SIMILITUDE 27

re
BY) Nea A scot
y

ieee
P,
Figure 3.2 Compressor map.

i.e., head varies as the square of speed. Power is the product of Q and H, and
the third pump law states that

P tore iP2
1.2 (1133)
3 3
NU
Laws for scaling up or down, i.e., varying diameter D while keeping the speed
constant, follow in a similar manner after cancelling the factors containing N.
Thus we find’

Q, Q
D>
i
D3
De
SAEY
H, 1 H 2)
ee2
pes22 (Beis)
Di Ds

Pe oP
ae (3.16)
1 Be

Performance curves are frequently plotted from dimensionless or quasidimen-

sionless groups. Compressor maps, for example, are usually presented in graphs
of the form shown in Figure 3.2. The abscissa is determined from (3.8) by
dropping the diameter, since it is not a variable in the performance of a single
machine, and by using the same gas constant. Similarly, the parameter N/VT,
is a variation of the machine Mach number formed from (3.7) by eliminating
specific heat and the rotor diameter, which are both constants for a given map.
Besides their use in performance curves, dimensionless groups are useful
in design and machine selection. For example, specific speed is commonly
28 DIMENSIONLESS QUANTITIES

used to indicate the type of machine appropriate to a given service. Table 3.2
gives ranges of specific speeds corresponding to efficient operation of the turbo-
machines listed. Sizes of turbomachines required for a given service are also
determinable for correlations of groups such as those shown in Figure 3.3.
This correlation, developed by Csanady (1964), of the optimum specific
speeds of various machines as a function of specific diameter is useful in
determining an appropriate size for a given set of operating conditions. To
enter the diagram, called the Cordier diagram, a specific speed can be selected
from Table 3.2. The rotor diameter can be determined from the specific di-
ameter found from the Cordier diagram. A machine so selected or designed
would be expected to have high efficiency.

Figure 3.3 Cordier diagram. (From G.T. Csanady, Theory of Turbomachines.

Copyright © 1964 by McGraw-Hill, Inc. Used with the permission of McGraw-
Hill Book Company.)
PROBLEMS 29

Table 3.2 Specific Speeds

Turbomachine Specific speed range

Pelton wheel 0.03- 0.3

Francis turbine 0.3 - 2.0
Kaplan turbine 2A = SO)
Centrifugal pumps 0.2 - 2.5
Axial-flow pumps Cy at) SIGS)
Centrifugal compressors ODF =) 220
Axial-flow turbines 0.4 - 2.0
Axial-flow compressors 1.5 -20.0

Reference

Csanady, G.T. 1964. Theory of Turbomachines. McGraw-Hill, New York.

Problems

3.1 Derive expressions for specific speed, specific diameter, and power coef-
ficient by combining flow and head coefficient. Show that each is dimension-
less.

3.2 Determine the model-to-prototype diameter ratio for a water turbine

which will produce 30,000 hp at 100 rpm with a head of 50 ft, while the
model will produce 55 hp under a head of 15 ft.
3.3 Determine the speed and flow rate of the model turbine in Problem 3.2.

3.4 A centrifugal fan is to be compared with a larger, geometrically similar

fan. The smaller fan delivers 500 ft?/min of air at standard conditions with a
pressure rise (head) of 2 in. of water. The smaller fan runs at 1800 rpm, while
the larger one operates at 1400 rpm while producing the same head. Determine
the diameter ratio of the two fans and the flow rate of the larger.
3.5 Use the Cordier diagram to estimate the rotor diameter of a pump which,
while running at 1000 rpm at a head of 30 ft, will deliver 4500 gal/min.
Would you recommend a centrifugal or an axial-flow pump for this service?
&
 4 Centrifugal Pumps and Fans

4.1 Introduction

Rotors, known more commonly as impellers, of centrifugal pumps, blowers,

and fans are designed to transfer energy to a moving fluid that is considered
incompressible. Fans and blowers usually consist of a single impeller spinning
within an enclosure, known as the casing. Pumps, on the other hand, may be
designed to have several impellers mounted on the same shaft, and the fluid
discharging from one is conducted to the inlet of the neighboring rotor,
thus making the overall pressure rise of the pump the sum of the individual-
_ Stage pressure rises. The individual impellers are designed to look, in cross sec-
tion, somewhat like that shown in Figure 1.1.
An end view of the impeller is shown in Figure 4.1. The vanes shown
are curved backwards making the angles 8, and 6, with tangents to the circles
at radii r, and r,, respectively. Ideally, the relative velocity W, leaves the
vane at the outer edge of the impeller at the blade angle B,.
Figure 4.2 shows velocity diagrams at the inlet and outlet of the vane
passages. For the design-point operation the relative velocity W, is approxi-
mately aligned with the tangent to the vane surface at angle 6, . The absolute
velocity V, at the inlet is shown entering with no whirl. Thus V,, = 0 and
Vini = ¥,- The ideal or virtual head H;, which is the ideal energy transfer per
unit mass for perfect guidance by the vanes, is given by

He = WAV)
PA MAGS
i (4.1)
&
31
 oo»

32 yah CENTRIFUGAL PUMPS AND FANS —

Pressure face

Suction face

Circulatory
flow

Figure 4.1 Pump impeller.

Equation (4.1) was derived earlier as (2.28). The ideal head H,; is higher than
that found in practice. Reasons for this disparity and methods for correction
will be given in subsequent sections.

4.2 Impeller Flow

Figure 4.1 shows an impeller rotating in the clockwise direction. Fluid next
to the pressure face of the vane is forced to rotate at blade speed. Motion in a
purely circular path at radius r implies a net pressure force directed radially in
inward, so that the net pressure force A dp on a differential element of cross-
sectional area A balances the centrifugal force (9 A dr)N?r; thus the radial
pressure gradient is

dp _ pU?
dr ie (4.2)

Since the fluid does not follow the impeller as in solid body rotation, but in-
stead tends to remain stationary relative to the ground, a resultant outward
flow along the vane with an accompanying adverse pressure gradient occurs.
However, the magnitude of the pressure rise across the rotor is less than that
indicated by integration of (4.2); i.e., less than

p(U3 - Uj)
Rows Pals 5) (4.3)

A better estimate of the pressure rise is obtained from an equation formed

from (2.28) and (2.34):
 a.

IMPELLER FLOW 33

p(v2 - v2)
Po — Py = PUZVy» - (4.4)

Applying the law of trigonometry to the diagrams of Figure 4.2 yields the
relations

UZ + V3 - W3
U,V = ea 300 (4.5)

and
Dy

Combining (4.4), (4.5), and (4.6) results in the pressure rise expression

_ p(UZ - UZ + Wi - W3)
PD = A Gy

Although the static pressure at the inlet and outlet of the impeller is ex-
pected to be uniform across the opening between the vanes, pressures on the
two sides of a vane are expected to be different. As the fluid moves radially
outward, its angular momentum per unit mass V,,r is clearly increased. This
means that a moment of some force has been applied to the control volume
considered. The source of such a force is obviously a pressure difference be-
tween any two points on opposite sides of the control volume at the same
radial distance from the axis of rotation. The azimuthal force resulting from
this pressure difference is the so-called Coriolis force, of magnitude |2N x WI.
This force is applied to the impeller at the pressure face and the suction face
of the vane. Equation (4.7) applied between the inlet and some intermediate
radius less than r, implies that the greater pressure rise on the pressure face is
accompanied by a lower relative velocity W on that face. Conversely, a higher
relative velocity at the suction face is indicated. Figure 4.1 shows a circulatory
flow which is radially inward on the pressure face and radially outward on the
suction face, and this is superposed on the main flow, which is radially out-

DS
Figure 4.2 Velocity diagrams at inlet and outlet.
34 CENTRIFUGAL PUMPS AND FANS

ward. The difference in pressure rise on the two sides of the passage between
vanes implies a separation, or backflow, region near the outer end of the suc-
tion face. The latter implies a flow deflection away from the suction face near
the exit of the passage. The change in V,,, associated with this flow deflec-
tion is known as slip.
The ratio of the actual V,,, to the ideal V,,, is usually known as the slip
coefficient u,. Since the slip depends on the circulation, and the circulation is
clearly dependent on the geometry of the flow passage, a theoretical relation-
ship expressing u, as a function of the number of blades ng and exit angle B, is
not surprising. Shepherd (1956) has given such a relation:

TU, sinB
Decl = MN ao coe (4.8)
Vuo np

Equation (4.8) shows that for an infinite number of vanes, i.e., perfect guidance
of the fluid, u, becomes unity. The velocity diagram of Figure 4.2 must be
modified to agree with the corrected V,,,, and the angle B, must be taken as
the fluid angle rather than the vane angle. Equation (4.1) is used to compute
the actual energy transferred to the fluid from the impeller.

4.3 Efficiency

Flow in the impeller or casing passages is accompanied by frictional losses

which are proportional to the square of the flow velocity relative to the pas-
sage walls. All losses result in a conversion of mechanical energy into thermal
(internal) energy. Wall friction effects this transfer through direct dissipation
by viscous forces and by turbulence generation which culminates in viscous
dissipation within the small eddies. Secondary flow losses occur in regions of
flow separation, where circulation is maintained by the external flow, and in
curved flow passages, where it is maintained by centrifugal effects.
Outside the impeller, where no through flow occurs, the fluid is forced
to move tangentially and radially. This circulatory motion of unpumped fluid
results in an additonal (disk friction) loss. A different, but equally nonproduc-
tive, use of energy occurs because of a reverse flow (leakage) from the high-
pressure region near the impeller tip to the low-pressure region near the inlet.
The latter effect is the reason for the introduction of the volumetric efficiency
n,, defined as
a
Nes (4.9)
m+ my
PERFORMANCE CHARACTERISTICS 35

where re is the mass rate of leakage, and m is the mass rate of flow actually
discharged from the pump.
- Because of the loss of mechanical energy by the several mechanisms
mentioned above, the head H, i.e., the net mechanical energy added to the
fluid in the pump as determined by measurement, is less than the head com-
puted from (4.1). The losses are accounted for by defining hydraulic efficiency
Nyy aS

(4.10)

Usually, the practical performance parameter as determined by test is the over-

all pump efficiency n, defined by

es mgH
uf P (4.11)

where P is the power of the motor driving the pump as determined by dyna-
mometer test. The so-called total head H is determined from the steady-flow
energy equation after experimentally evaluating the mechanical energy terms
at the suction and discharge sides of the pump.
The mechanical efficiency n,, accounts for frictional losses occurring
between moving mechanical parts, which are typically bearings and seals, as
well as for disk friction, and is defined by

i (m+ m,) U,V

Nn (4.12)
P

Substitution of (4.9), (4.10), and (4.11) into (4.12) yields the simple relation-
ship

= ee (4.13)

4.4 Performance Characteristics

Characteristic curves for a given pump are determined by test, and they consist
primarily of a plot of head H as a function of volume rate of flow Q. A typical
characteristic curve is shown schematically in Figure 4.3. The theoretical head
from (4.1) is also shown in Figure 4.3. The actual curve is displaced downward
as a result of the losses of mechanical energy previously discussed. However,
(4.1) provides the engineer with the upper limit of performance which can be
36 CENTRIFUGAL PUMPS AND FANS

Actual
characteristics
External head

Figure 4.3 Pump characteristics.

achieved, since it does not account for losses. If the speed is increased, (4.1)
indicates that the curve will shift upward, and vice versa.
Expressing (4.1) in terms of Q, we have

H; = U, : (4.14)

Dividing (4.14) by the square of twice the tip speed ND, , we obtain

gH. D,Q cot B

See ter i5 (.:| (4.15)
N? D2 ™bND3

Equation (4.15) indicates a functional relationship between head coefficient

gH/N? D3 and flow coefficient Q/ND3, which is independent of speed. The
actual performance curves, when plotted nondimensionally, also show a func-
tional relationship which is independent of speed; i.e., data for different rotor
speeds will collapse into a single-head coefficient-flow coefficient curve.
We can predict the approximate value of head or flow rate resulting from
a change of speed if we assume that the operating state, i.e., the values of
head and flow coefficients, are the same before and after the change. Refer-
ring to Figure 4.3 and considering a change of speed from N, to Nj, the
operating state point on the characteristic plot moves from position 1 to
position 2. Since we are assuming similar flows,

H,
= Hy,
(4.16)
Ny Ny
and

Q; Q
Ni Ny (4.17)
PERFORMANCE CHARACTERISTICS 37

Equations (4.16) and (4.17) express the pump (or fan) laws. Manipulation of
these equations yields

H H
Go (4.18)

which states that H is proportional to Q?. The latter relation is approximately

that followed by the external system to which the pump is attached, assuming
that no changes have been made in it. Thus the path from 1 to 2 for a simple
change of speed is roughly the loci of similar states, and this fact makes the
pump (or fan) laws extremely useful.
As indicated by (4.11) overall efficiency varies with flow rate, and it is
required for the computation of brake power. Referring to Figure 4.4, which
is a typical variation, we see that efficiency varies with flow rate from zero at
no flow to a maximum value 7,,,, near the highest flow rate. The actual value
Of Nmax Varies from 70 to 90 percent, depending primarily upon the design
flow capacity. Machines handling large flows have higher maximum efficiencies,
since frictional head loss decreases proportionately with large flow area. On
the other hand, machines of high head and low flow, i.e., low specific speed,
tend to have lower efficiencies. High head is associated with large-diameter or
high-speed impellers, which increases disk-friction losses considerably, and
low flow implies higher proportional head loss associated with smaller flow
area. Characteristically, the latter-type machine is of the radial-flow design,
while the former is classified as a mixed-flow design. Although the choice of a
specific speed may be dictated by design requirements, it is worth noting that
test results show that centrifugal pumps with specific speeds between 0.7 and

=|const. YI
/NLE
LM MALL
YiN
aa
AN
Bin

Figure 4.4 Equal-efficiency contours for centrifugal pumps.

38 CENTRIFUGAL PUMPS AND FANS

1.0 seem to have the highest maximum efficiencies (for example, see Church,
1972).
The head-capacity (H-Q) curves can be altered at the high-flow end by
the occurrence of the phenomenon of cavitation. This process consists of the
formation and collapse of vapor bubbles, which occur when the fluid pressure
falls below the vapor pressure. Outward flow in the impeller passage, which is
accompanied by pressure rise, results in a collapse of the bubble. Acceleration
of fluid surrounding the bubble, which is required to fill the void left by the va-
por, results in losses and pressure waves which cause damage to solid-boundary
materials. Since the energy transfer per unit weight is reduced by the presence
of vapor, the head-capacity curve falls off at the flow corresponding to the
beginning of cavitation.
To avoid cavitation, the net positive suction head (NPSH), defined as
the atmospheric head plus the distance of liquid level above pump centerline
minus the friction head in suction piping minus the gauge vapor pressure, is
maintained above a certain critical value. A critical specific speed S, defined as
NQ?/2
© ~ (NSPHsA (4.19)

is used to determine the lowest safe value of NSPH. For single-suction water
pumps Shepherd (1956) gives S, = 3, and for double-suction pumps he gives
S, = 4. These form useful rules of thumb for the avoidance of cavitation by
designers and users of centrifugal pumps.

4.5 Design of Pumps

Requirements for a pump comprise the specification of head, capacity, and

speed. This section deals with the application of principles to the problem of
the determination of the basic dimensions of the impeller and casing. The
process outlined below would enable the engineer to carry out a preliminary
design to which the detailed mechanical design could be added, or to select a
suiltable pump from commercially available machines.
The impeller design can be started by computing the required specific
speed and using this value to determine efficiency from available test data
plotted in the form of 7 as a function of N, with Q as the parameter. Brake
power calculated from (4.11) is then used to determine shaft torque from
P
ie N (4.20)
DESIGN OF PUMPS 39

Referring to Figure 4.5, it is clear that a hub diameter D,, is obtained

by adding a small increment, say 0.5 in., to the shaft diameter which may be
calculated from shaft torque using a standard formula. Shroud diameter Dg is
then found from Dy, providing an inlet area sufficient to handle the capacity
Q plus the leakage flow Q,:

| bi sora)"
Dhy =Deee
TV 9
(4.21)
The inlet velocity can be taken as 10 ft/s or less, and the leakage can be as-
sumed to be 2 or 3 percent of the capacity.
The diameter D, at the leading edge of the vane can be taken as equal
to D). Then the width b, of the leading edge can be computed using the flow
Q + Q, and a radial velocity V, slightly higher than Vg, say 1.05V9. Thus
we have

er Sy
b 4.22
: 7™D,V;, ( )

Vane angle 8, can be taken as approximately equal to the fluid angle, and
thus it is easily determined from the velocity triangle shown in Figure 4.2.
The outlet vane angle B, is usually selected as 15 to 40°, which results in
a lower energy transfer and a reduced absolute exit velocity. Csanady (1964)
has derived a formula for the determination of the optimum (minimum-loss)
ratio of whirl velocity to tip speed:

Figure 4.5 Radial-flow impeller.

40 CENTRIFUGAL PUMPS AND FANS

Vio (= (a a) 1/2
(4.23)
U, eae

where ¢, and §, are diffuser and impeller loss coefficients, respectively. Using
§,/§, = 1/3 and B, = 25°, we arrive at a typical relation, V,, = 0.62U, .Using
the given value of head H with (4.10) and (4.13), we can determine a value of
U, and hence D,. If the radial component V,,, of the exit velocity is too high,
it can be reduced by choosing a lower value of 8, and repeating the calcula-
tion (see Figure 4.2). Vane width b, is easily determined from

_ QQ, (4.24)
ae m™D, Vin2

The widths b, and b, can be increased slightly to account for vane thickness
in a refined calculation.
The appropriate value of B and b at stations intermediate between inlet
and exit can be worked out using velocity triangles and assuming a smooth
variation in all quantities. The optimum number of vanes is determined from
a formula of Pfleiderer (see Church, 1972), namely,

6.5(D, +D,) Me
Np = - 4.25
BD, - D,) sin [G, + 8)/21
The fluid exits the impeller with tangential and radial components of
absolute velocity and is collected and conducted to the discharge of the pump
by the volute or scroll portion of the casing (see Figure 4.6). The volute is

Figure 4.6 Volute of a pump.

FANS 41

usually in the form of a channel of increasing cross-sectional area. The volute

begins at the tongue with no cross-sectional area and ends at the discharge
nozzle. At any angle ¢, measured from the tongue, the flow rate is (¢/360)Q.
The angular momentum of the exit flow, V,,, R, is conserved, so the distribu-
tion is approximately

Vr = constant (4.26)

Assuming a channel of constant width W, the outer wall of the volute is

located at R,, determined from the relation

OQ R3
360 5
If the channel width is variable, as in a channel of circular cross section, then
the governing relation should be

OQ = R3 W

The so-called discharge nozzle, which is really a diffuser, joins the volute
to the discharge flange of the pump. For water the nozzle is typically sized to
produce a discharge velocity of 25 ft/s. A radial diffuser may be added between
the impeller and the volute for high-pressure pumps. This may take the form
of a space of constant width without vanes, or it may include vanes forming
diverging passages aligned with the absolute velocity vector.

4.6 Fans

Fans produce very small pressure heads measured in inches of water pressure
differential, and of course are employed to move air or other gases. Acom-
pressor also handles gases, but with large enough pressure rises that significant
fluid density changes occur; i.e., if density is increased by 5 percent, then the
turbomachine may be called a compressor. 3
A centrifugal fan, as compared with a pump, requires a much smaller in-
crease in impeller blade speed, i.e., a smaller radius ratio R,/R,, as may be
inferred from (4.7). It requires a volute, of course, but no diffuser is needed
to enhance pressure rise. The flow passages between impeller vanes are quite
short, as indicated in Figure 4.7.
The analysis and design of the impeller proceeds as with the centrifugal
pump. The small changes of gas density are ignored, and the incompressible
equations are applied as with pumps. Performance curves are qualitatively the
42 CENTRIFUGAL PUMPS AND FANS

Discharge
—— >

Figure 4.7 Centrifugal fan.

same as for pumps, except that the units of head are customarily given in
inches of water, and those of capacity are typically in cubic feet per minute.
Other differences are that both total head and static (pressure) head are
usually shown on performance curves, and a fan static efficiency, based on
(4.11), is calculated using static head (P, - P,)/pg in place of total head H.
Similarity laws for pumps are applied and are known as fan laws; these are
represented by (4.16) and (4.18).

ft
HEAD,

0 500 1000 1500 2000 2500 3000 3500 4000 4500

Volume flow rate, gal/min

Figure 4.8 Head-capacity curve for a pump.

 PROBLEMS 43

References

Church, A.H. 1972. Centrifugal Pumps and Blowers. Krieger, Huntington, New
York.
b& Csanady, G.T. 1964. Theory of Turbomachines. McGraw-Hill, New York.
Shepherd, D.G. 1956. Principles of Turbomachinery. MacMillan, New York.

Problems

4.1 A centrifugal water pump has the characteristic curves shown in Figure
4.8. Using the 1160-rpm curve plot the characteristics (H versus Q) for a
geometrically similar pump having twice the speed and half the diameter (of
the rotor). Show calculations that were used to obtain the coordinates plotted.

4.2 Calculate the power required to drive the original pump at 1160 rpm at
a flow rate of 3100 gal/min. Also determine the specific speed (unitless).

4.3 Fora pump impeller with a diameter D, of 1.326 ft and axial width b, =
2 in. determine the velocity diagram at the exit of the rotor for the conditions
in Problem 4.2. Vane angle B, = 25°.
4.4 Determine the principal dimensions of a centrifugal pump which can
deliver 3100 gal/min of water at a 100 ft head. The speed is 1160 rpm.
 5 Centrifugal Compressors

§.1 Introduction

Although centrifugal compressors are slightly less efficient than axial-flow

compressors, they are easier to manufacture and are thus sometimes preferred.
For example, the vanes are typically radial and flat, except at the inlet, and can
be machined with relative ease. Thus, the centrifugal machine finds application
in ground-vehicle power plants, auxiliary power units, and other small units.
The parts of a centrifugal compressor are the same as those of a pump,
namely, the impeller, the diffuser, and the volute. The basic equations devel-
oped in Chapter 4 apply to compressors with the difference that density
does increase, and we must consider the thermodynamic equation of state of
a perfect gas in the detailed calculations.
The main difference in carrying out a compressor analysis, as opposed
to a pump analysis, is the appearance of an enthalpy term in place of the flow-
work or pressure-head term. It is convenient to use both total and static en-
thalpy, denoted by hy and h, respectively. Thus, energy transfer E is given by

as well as by

Ee, (5.2)
Equations like (5.1) and (5.2) used in the same analysis require care in
handling units, since the enthalpy difference in (5.1) may carry units such as
Btu/lb, whereas (5.2) would carry units of velocity squared. Suitable conver-

45
46 CENTRIFUGAL COMPRESSORS

Figure 5.1 Enthalpy-entropy diagram.

sion factors do not appear in the equations but must be applied in computa-
tions with them.
Since thermodynamic calculations are involved in compressor analysis
and design, the h-s diagram, such as that shown in Figure 5.1, becomes useful.
The state at the impeller inlet is indicated by point 1, and that at the impeller
outlet by point 2. The diffuser process is indicated between points 2 and 3.
The corresponding stagnation properties 01, 02, and 03 are also indicated in
Figure 5.1, since kinetic energies are usually considerable.
The expression for compressor efficiency appears to be somewhat dif-
ferent from that for pump efficiency, but, in reality, the principle of the defini-
tion is the same. Both definitions employ the ratio of the useful increase of
fluid energy divided by the actual energy input to the fluid. For the compres-
sion of a gas, the useful energy input is the work of an ideal, or isentropic,
compression to the actual final pressure P,. This is calculated from
Pee (Y-L)/Y
Ei =C,To1 722)= (5.3)

which evaluates the work of the isentropic process from state 01 to state i in
Figure 5.1. The compressor efficiency can be reduced to

Ty Tot
n —
(5.4)
i To3 = To4

which is the ratio of E; to E.

The compressor efficiency, an experimentally determined quantity, is
useful in predicting pressure ratios in new designs. Using (5.2), (5.3), and (5.4)
we can obtain the overall pressure ratio:
IMPELLER DESIGN 47
iF U, V. TWD)
evere f+ -2teaTe
re) (Se5))

Pou CyTo1

The velocity component V,,, is determined to be wU, for radial vanes (6, =
90°), as shown in Figure 5.2. Thus, the pressure ratio for a compressor stage
is expressed very simply as a function of tip speed U,, inlet temperature To, ,
slip factor u,, and compressor efficiency n,. Single stages can produce pres-
sure ratios as high as 6, which is sufficient to operate an efficient gas turbine
cycle, for example.

5.2 Impeller Design

The impeller is usually designed with a number of unshrouded blades, given

by the Pfleiderer equation [(4.25)] , to receive the axially directed fluid (V, =
Vini) and deliver the fluid with an almost radial relative velocity (8, ~ 90°).
Although the vanes can be flat and radial over most of their extent, they are
generally bent to conform to the direction of the relative velocity W, at the
inlet.
The angle B, varies over the leading edge, since V, remains constant
while U, (and r) varies. At the largest radius of the impeller, the relative ve-
locity W, and the corresponding relative Mach number M,, are highest. It is
easily shown for a fixed set of inlet operating conditions, i.e., N, m, Po,, and
To, , that the relative Mach number has its minimum where B, is approximately
32° (see Shepherd, 1956).

We

Rim of
ee impeller

Rotation

Figure 5.2 Velocity diagram at impeller exit.

48 CENTRIFUGAL COMPRESSORS

Figure 5.3 Velocity diagram at impeller inlet.

Referring to Figure 5.3, it is seen that a choice of relative Mach number

at the tip of the inlet vane allows the inlet design to proceed in the following
manner. The acoustic speed a, is calculated from the inlet termperature. Next,
W, at the tip is calculated from

Wit = Myay (5.6)

Then V, and U, are calculated from

V, = W, sin 32° (5.7)

and
U,,
= W, cos ae (5.8)

It is then possible to determine tip radius, since

eG
Tit = Ne (5.9)

The hub radius can also be calculated from the mass flow m; thus,

ee m- \ 12 .
(5.10)
iti (3,-38.)

Referring to Figure 5.3, the fluid angle at the hub is calculated from

-1 Vy
By, = tan ue i)

The radius r, at the impeller tip can be determined from (5.5) using a
value of V,,. (u,U, for radial vanes) determined from the exit-velocity diagram.
The slip factor u, is determined as for centrifugal pumps, and the ratio of V,,5
to U, is chosen to lie in the optimal range of 0.23 to 0.35 (see Ferguson,
1963). The value of Po, is, of course, specified in advance.
The calculation of U,, and hence r,, requires the selection of a value
for compressor efficiency n,. This choice will be based on experience, but a
conservative value will probably lie in the range 70 to 80 percent (see Shepherd,
1956). Since the efficiency has been shown to vary with Reynolds number
and Mach number, both based on U,, several iterations may be required.
 DIFFUSER DESIGN 49

The compressor efficiency n,, in addition to its use in (5.5), can be em-
ployed to estimate the impeller efficiency n,. The ratio x of impeller losses to
compressor losses

Mekist a= (S342)

can be estimated and lies between 0.25 and 0.50. The definition of impeller
efficiency

a i -To1
ny Cy)
To2 - tor
can be used to estimate T; (see Figure 5.1). The latter total temperature cor-
responds to the total pressure Po, , calculated from

nove (= a (S214)
Po. To,

The static pressure P, is then determined from

2 _
pene ee OM! 2) Y/(Y-1) (S215)
2
The static pressure P, from (5.15) and the static temperature T, deter-
mined from

Eo) NV;
LN reece res (5.16)
C, 2C,

are used to determine density p, at the impeller exit. Finally, the axial width
b, of the impeller passage at the periphery may be found from

m rl
fie epee
UI (aig
2 2m) Vina

5.3 Diffuser Design

A vaneless diffuser, or empty space, between the leading edges of diffuser vanes
and the impeller tip allows some equalization of velocity and a reduction of
the exit Mach number. The vaneless portion, which may have a width as large
as 20 percent of the impeller diameter, also effects a rise in static pressure. As
with the pump, angular momentum rV,, is conserved, and the fluid path is ap-
proximately a logarithmic spiral. Diffuser vanes are set with the diffuser axes
50 CENTRIFUGAL COMPRESSORS

tangent to the spiral paths and with an angle of divergence between them not
exceeding 12°.
Since the addition of a vaned portion in the diffusion system results in
a small-diameter casing, vanes are preferred in instances where size limitations
are imposed. On the other hand, a completely vaneless diffuser is more efficient.
If vanes are used, then their number should generally be less than the number
of impeller vanes to ensure uniformness of flow and high diffuser efficiency
in the range of flow coefficient V,,,/U, recommended in the previous section.
Diffuser efficiency np is usually defined as (see Figure 5.1)

Np =— (5.18)

This is experimentally determined (see Ferguson, 1963), but may be estimated

to lie between 65 and 75 percent for vane diffusers. The static pressure P,,
static temperature T,, cross-sectional area A,, and absolute velocity V, at
the diffuser outlet may be determined by solving (5.18) and the following
equations simultaneously:
Po3 L ie Y/(Y-1)
us
P, T; (By 1 USS))

me VAS
° P3
(5.20)
T,
T 03 = T +3 V3
AS,
+ —— (Spell )

Ty NG Gy 19) 0)
Hi =(23) (5.22)
Wy)
with P93 specified, the five unknowns P,,T3, A3, V3, and T;’ are easily deter-
mined by iteration. We can omit (5.19) if the static pressure P, is specified
instead.
A volute is designed by the same methods outlined in Chapter 4. The
volute functions to collect the diffuser’s discharge around the 360° periphery
and deliver it through a single nozzle to the connecting gas-piping system or
to the inlet of the next compressor stage.

5.4 Performance

Typical compressor characteristics are shown in Figure 5.4. Quantitatively,

their shape is similar to those of the centrifugal pump, but the sharp fall of
PERFORMANCE : 51

N= constant

|50
OU
1in€ —» Choke
line

Figure 5.4 Compressor map.

the constant-speed curves at higher mass flows is due to choking in some com-
ponent of the machine. At low flows operation is limited by the phenomenon
of surge. Thus, smooth operation occurs on the compressor map at some point
between the surge line and the choke line.
The phenomenon of choking is that associated with the attainment of a
Mach number of unity. In the stationary passages of the inlet or diffuser, the
Mach number is based on the absolute velocity V. Thus for a Mach number of
unity, the absolute velocity equals the acoustic speed a, calculated from

a = (yRT)1/2 (5.23)
The temperature at this point is calculated from the total temperature Tp
using the relation

Ty =T[1 + (y- 1) (0.5)M2] (5.24)

setting M = 1. Thus

T= 1,2} (5.25)
This Mach number is found near the cross section of minimum area, or throat
(A,), so that we can estimate the choking, or maximum, mass flow rate from
1/2
th = A,P, i (5.26)
52 CENTRIFUGAL COMPRESSORS

The pressure P, at the throat area may be estimated by assuming an isentropic

process from the inlet of the stationary component to the throat area. Thus

dis TED
t in T
in

The process of estimating choked flow rate in the impeller is the same except
that relative velocity is substituted for absolute velocity. When the relative
Mach number W/a is set equal to unity in the energy equation of the rotor,
namely,

fig (5.28)

we obtain

1) 21 + U2726 519
(5:29)
To, Vier

Using the isentropic relation between pressure and temperature and substitut-
ing into the continuity relation, the mass flow rate at the throat section of
the impeller is given by
3 1/2 2 (y+1)/2(Y-1)
th= A,Poy (= : = (+ a ) farsa (5.30)
01 Fe p+o
Thus, it is clear that mass flow for choking in stationary components, given
by (5.26), is independent of impeller speed, but that mass flow for choking in
the impeller, given by (5.30), actually increases with impeller oe This is
indicated schematically in Figure 5.4.
Referring to Figure 5.4 the point A represents a point of normal opera-
tion. An increase in flow resistance in the connected external flow system
results in a decrease in V,,, at the impeller exit and a corresponding increase
in V,,,, which results in an increased head or pressure increase. However, the
surge phenomenon results when a further increase in external resistance prod-
uces a decrease in impeller flow than tends to move the point beyond
C,
where stall at some point in the impeller leads to change of direction of W,
and an accompanying decrease in the head (or pressure rise) in the impeller
A temporary flow reversal in the impeller and the ensuing buildup to the ori-
ginal flow condition is known as surging. Surging continues cyclically until
the external resistance is removed. It is an unstable and dangerous condition
and must be avoided by éareful operational planning and system design.
PROBLEMS 53

References

Ferguson,
TB. 1963. The Centrifugal Compressor Stage. Butterworths, London.
Shepherd, D.G. 1956. Principles of Turbomachinery. Macmillan, New York.

Problems

5.1 Air enters a centrifugal compressor at 1 atm, 58°F, and V, = 328 ft/s.
At the impeller exit B, = 63.4°, V,,, = 394 ft/s, and U, = 1640 ft/s. The mass
flow rate is 5.5 lb/s, the mechanical efficiency is 95 percent, and the compres-
sor efficiency is 80 percent. Determine the ratio of total pressures at outlet
and inlet and the power required to drive the machine.

5.2 Design a single-stage centrifugal compressor which will handle 2.2 lb/s of
air at a pressure ratio of 4.2:1. Use radial blades with an appropriate inducer.
Assume d,/d; = 0.361 at the inlet eye and n, = 0.78. The machine is to operate
at 60,000 rpm and to supply air to the combustion chamber of a turboject
engine. The basic design parameters required are the following:

Hub diameter
. Shroud diameter at impeller inlet
Shroud diameter at impeller exit
. Impeller inlet vane angle
Vane width at impeller exit
Velocity triangle at impeller inlet (shroud diam.)
Rwmeaoge
Velocity triangle at impeller exit

The compressor is to have no inlet guide vanes and is to draw in ambient air
at 14.7 psia and 520°R.
 *

ee
—S~
< ——— ——
Tre

se

> ~
 6 Axial-Flow Pumps and Fans

6.1 Introduction

Axial-flow pumps and fans move liquids and gases without significant effect
on their density. They are like propellers in that power is supplied to produce
axial motion of the fluid, but they are different in that the fluid being moved
is enclosed by a casing. Like propellers, the vanes have small curvature and
cause little deflection of the relative velocity vector of a fluid particle as it
migrates through the moving passages.
Generally, the vanes have shapes, or profiles, like that of an airfoil: they
are thin, streamlined, and cambered (see Figure 6.1). The relative velocity W,
approaches the vane at an angle a (the angle of attack) to the chord line. The
exiting fluid with relative velocity W, has been deflected slightly, and the
change of momentum results in a lift force L perpendicular to the mean direc-
tion of W, and W,,i.e., perpendicular to a mean relative velocity W,,.
The lift force L is primarily responsible for the transfer of energy, and
the drag force D, which is directed parallel to W,,, , is strongly associated with
blade losses. Lift is maximized by setting the blades at high angle of attack,
but stall occurs if the angle is too high. Such characteristics of blades are
determined in a wind tunnel using a representative set of blades arranged in
series, known as a cascade. This kind of experimentation provides informa-
tion not only about optimum incidence, but also about optimum spacing for
maximum lift and minimum drag.

55
56 AXIAL-FLOW PUMPS AND FANS

Figure 6.1 Blade profile. »

The axial-flow fan, pump, or compressor blade moves in a direction

opposite to that of the blade in an axial-flow turbine; the concave side comes
first. In Figure 6.2 blade motion is to the right. The tangential component
F,,, of the blade force can be obtained in terms of the angle B,, that W,,
makes with the axial direction. Referring to Figure 6.2, it is clear that

Pa L cos.
iyDsinp., “5 (6.1)

The rate of energy transfer Eis given by

* E=U(Locos Ba + Disin Be.) (6.2)

and the energy transfer per unit mass is expressed as

Pee yn) | (6.3)

in which U is the same at the inlet and exit planes, since the flow ideally con-
tains no radial components of velocity. An alternative expression for E is
obtained by multiplying (6.3) by mass flow rate m, given by

m = pV,S (6.4)

where V, is the axial component of absolute velocity, and S is the spacing

between two adjacent blades in a row. It is evident from (6.4) that m is mass
flow rate per blade per unit length of blade. The blade force equation (6.1),
and the blade power equation (6.2), should also be interpreted as force and
power, respectively, per blade per unit length of blade.
Equating (6.2) with (6.3) times (6.4) and nondimensionalizing, we obtain

cos? B, S
Cre) scan 7 (tan B, - tan B,) (6.5)

where, as is evident from Figure 6.3,

AV, = Vuo - Yur 2 Wut - Wyo = AW, (6.6)

 STAGE PRESSURE RISE 57

iG

W> ey.
Axial
direction
Bo

Figure 6.2 Blade motion.

In getting (6.5) we have defined the lift coefficient C,, as

CAS L =Hpwre
IE
(6.7)
where C is the chord, the length of a line drawn between the leading edge and
the trailing edge of the blade (see Figure 6.2). Also, in forming (6.5) we have
omitted the drag term that appears in (6.1) on the grounds that D << L. The
relationship expressed by (6.5) links the force produced by a blade with the
flow angle and the nondimensional spacing of blades, or pitch-chord ratio
S/C. For the ideal cascade, C, can be predicted from (6.5). The presence of
boundary layers and nonuniform flow deflection in actual cascades leads to
experimentally determined relationships between these variables.

6.2 Stage Pressure Rise

A single-stage fan or pump generally consists of inlet guide vanes, rotor vanes,
and diffuser vanes. The inlet guide vanes deflect the fluid so that it enters the
moving vanes at an angle with respect to the axial direction. This is indicated
by the angle velocity V, makes with the axial direction, as shown in Figure
6.3. The kinetic energy of the fluid is increased in the rotor, so that it leaves
with velocity V, > V,. It is decelerated in the stationary blades of the dif-
fuser and deflected back to the axial direction.
Pumps and fans frequently have few blades in the rotor. The solidity,
defined as C/S, ranges from 0.4 to 0.8 or higher. For the highest specific
speeds only two blades may be used. Increasing the number of vanes increases
guidance and thereby increases head, but friction losses also are increased.
The optimum soliditiy is determined by test.
38 AXIAL-FLOW PUMPS AND FANS

Figure 6.3 Velocity diagrams.

Let us consider a control volume, surrounding a single moving blade, of

width S and of unit height along the blade, as shown in Figure 6.4. Assuming
no change in the axial component of fluid velocity from inlet to outlet, we
can write the force equilibrium equation as

(P, -P,)S- Fp, =0 (6.8)

Expressing the axial component F,, of the blade force in terms of lift and
drag, we obtain for the pressure rise across the moving row

ee. D
(Pht) toe aoe Bm = wee Be (6.9)

In nondimensional form, (6.9) becomes

(P ~ Pi), C
= 5 (C, sin 8,, -Cp cos By) (6.10)
Yow?

a—Boundary of
/ control volume
~-@---~ Ui

ie X
Figure 6.4 Control volume for a cascade blade.
 STAGE PRESSURE RISE 59

Thus (6.10) allows prediction of pressure rise in terms of aerodynamic coef-

ficients and relative velocities. A similar relation can be obtained for the dif-
fuser section of the stage using the same method.
An alternate expression for pressure rise across the moving row is ob-
tained by obtaining the axial component F,, in terms of the tangential com-
ponent F,,,. Thus,

Fp,
= Fpy tan By - 9) (6.11)

Since it is small, the tangent of § may be approximated by 6, which represents

the ratio of drag to lift, or C,/C,. Using a trigonometric identity and noting
from Figure 6.3 that

um
tan B=
an B,, oe
7 ong
(6.12)
Vv a

we substitute in (6.8) with (6.11) and (6.12) to obtain

R - ¢6
(6.13)
rN Bele eres
in which the component Fy, is replaced by using the change in tangential
momentum flow, namely,

Fp, = PV_S(Vy2 - Vyz) = PU? SARO (6.14)

where Ag is the blade loading coefficient AV,,/U. A similar derivation may be
made for the diffuser blading to obtain the pressure rise in the stationary part
_of the stage. Addition of the two leads to the equation for overall stage pres-
sure rise:
ete R-¢65 , 1-R-¢6
Ap = pU #ig|
Ra « AaB at (6.15)

The above relation is useful in estimating stage efficiency, which is de-

fined as the ratio of pressure rise with blade drag accounted for to that with
frictionless blades. Elimination of drag means that the terms in (6.15) that in-
volve 6 vanish. Thus the stage efficiency 7 is given by

2AP “pa nat a Gate

Trae |OPOR GF 50 SR
The problem with using (6.16) is that the drag-lift ratio 6 accounts only for
the profile losses, and there are several other losses which affect the stage ef-
ficiency. These are considered in the next section.
60 AXIAL-FLOW PUMPS AND FANS

6.3 Losses

Boundary layers on the surfaces of blades, whether moving or stationary,

mark regions of high shear stress, and the resultant of the viscous forces prod-
uced at the surface of the blade is the drag force. In addition to resisting blade
movement, viscous forces retard fluid in the stationary passages and result in
total pressure losses. The thickness of the boundary layers on the blade sur-
faces deflects the main flow and thus changes the effective blade shape. The
effect of the increasing pressure in the flow direction slows down the fluid in
the boundary layer and promotes separation of the boundary layer from the
blade surfaces concomitantly creating regions of reversed flow. In the latter
case, the effective blade shape is drastically distorted and the flow direction is
severely modified.
Besides boundary-layer formation on the blades, layers are also formed
on the inner and outer surfaces of the annular-flow passage, the cylindrical
surfaces at the hub and tip radii. Since the flow actually takes place in the
rectangular passage bounded by blades on two sides and by walls of the annu-
lus on the other two, it is expected that losses will depend on the ratio of
blade spacing to blade height. Empirically, it has been found that the drag
produced by these surfaces is correctly reflected by the relation

C0102 = (6.17)
where h represents blade height and Cy is the increment to be added to the
previous drag coefficient to account for annulus losses.
The velocity variation due to boundary layers on the blades and walls of
the annulus, coupled with the curvature of the blade surfaces, results in an ad-
ditonal loss. Secondary currents are set up in a plane transverse to the flow, as
is indicated in Figure 6.5. Dissipation of the energy of these secondary currents
takes place in the blade passage and in the wakes behind the trailing edges via
vortices spawned by the interaction of neighboring secondary flow cells as
they leave the blade. Because the trailing vortices are similar to wing vortices,
it is expected that the corresponding drag is proportional to the square of the
lift coefficient. The recommended equation for drag coefficient is then
Ca 0.018C# (6.18)

The difference in pressure on the two sides of the moving blades results
in a leakage of fluid around the tip, i.e., through the narrow passage formed
between the blade tip and the casing. This loss is accounted for by the empiri-
cal formula
i
PUMP DESIGN 61

Figure 6.5 Secondary flow in blade passages.

C209
Die St
acyc3/2 (6.19)

where Cy is the tip clearance.

To obtain a more realistic value for the stage efficiency using (6.16),
we can artificially increase the blade drag force by an amount proportional
to the sum of Cy, Cy, and Cy. We then substitute for 6 in (6.16) using the
expression

5=
Chenery eer CH
Cy (6.20)

6.4 Pump Design

Axial flow pumps are used for specific speeds above approximately 3, with
centrifugal pumps occupying the range below 2 and mixed-flow pumps filling
the gap between the two. They are then machines of low head, high capacity,
and a single stage. They require several well-finished rotor vanes of airfoil sec-
tion, as shown in Figure 6.1.
As a starting point in the design of an axial-flow pump we can use that
part of a Cordier diagram (Csanady) for which N, > 3. The relationship be-
tween specific speed N, and specific diameter D, is given approximately by

D, = amas (6.21)
Ss

Calculating N, from specified values of N, Q, and H, we can arrive at D, from

(6.21). This value of specific diameter is used to compute a rotor-tip diameter
D, in the following manner:
D,Q'2
(6.22)
(en'4
By selecting a suitable hub-tip ratio, i.e., blade-root diameter D, divided
by blade-tip diameter D,, we are able to compute the hub diameter, which is
62 AXIAL-FLOW PUMPS AND FANS

a synomous expression for root diameter. Generally, axial-flow pumps have

hub-tip ratios in the range 0.3 to 0.7. The graphical display of the relationship
between D,/D,, N,, and solidity C/S, based on current practice, is given by
Stepanoff (1957) and may be approximated by the relation

Ce= 10 6.23
Ss (D,/D,)N3*
The solidity, calculated from (6.23), is based on a suitably chosen value of
hub-tip ratio and the required specific speed. It should lie in the range 0.4 to
1.1, and if the calculated solidity lies outside that range, a new choice of D,/D,
should be made.
The annular flow area and the required flow rate can now be used to
determine the axial velocity component V,. Thus, we have

Va= =4Q 6.24

4 n(D?
- D?) ae
The velocity diagram shown in Figure 6.6 shows the ease with which the re-
quired fluid angle 8, is determined. The absolute velocity V, is axially directed,
and V, is determined from (6.24). The mean blade speed U is determined
from the mean diameter D,, .Thus at the mean diameter we can write

_D.+D, re

and
ND
U= 5a ; (6.26)

where N is rotational speed in radians per second. Thus, the required fluid
angle 6, is given by
U
6, =t
=tan =1 =
(6.27)
i

U = Wu)

Figure 6.6 Velocity diagram at rotor inlet.

PUMP DESIGN 63

Figure 6.7 Velocity diagram at rotor outlet.

Similarly, the fluid angle B, at the rotor exit is determined by reference

to Figure 6.7. Assuming the same annular flow area, and hence the same axial
velocity V,, we know U and V, as before. With a known head and with no in-
let whirl, i.e., V,, = 0, we determine the exit whirl velocity V,,, from (6.3).
Thus we may write

N ot ce (6.28)

The exit fluid angle 8, is easily found through the use of the geometric relation

U-V
6, = tan"! ve (6.29)
a

The mean fluid angle B,, is determined from (6.12), where

W., +W

and W,,, and W,,, are defined by Figures 6.6 and 6.7.
The hydraulic efficiency n,,, appearing in (6.28), bears the same relation-
ship to overall efficiency n as was previously indicated in connection with
centrifugal pumps, Thus, (4.9) through (4.13) all apply to the present analysis.
The overall efficiency of the single-stage pump may be estimated from (6.16),
and hydraulic efficiency 7,, will be slightly higher, as indicated by (6.13). For
example, the latter is expected to lie in the range 0.85 to 0.90.
The stagger angle i.e., the angle between the chord line of the profile
and the axial direction, is determined from the required incidence, or angle of
attack a, necessary to produce the lift coefficient C, calculated from (6.5).
Generally, we will select an airfoil section for which cascade data are available.
NACA Report 460 is an example of such a source. Figure 6.8 shows schemati-
64 AXIAL-FLOW PUMPS AND FANS

Figure 6.8 Typical cascade results.

cally the sort of cascade results available in NACA reports and elsewhere. Cas-
cade results should also be checked to assure that the angle of attack chosen
does, in fact, produce the desired fluid deflection.
The blade may be twisted if the so-called free-vortex method is employed.
In this method the product of V,,, and D is kept constant, and this variation
of V,,, with radial position will result in a variation of 8, . The angle 6, varies
with U, and the blade may be twisted to provide proper guidance at the trail-
ing edge, as well as the correct incidence. Free-vortex design results in approxi-
mately uniform energy transfer at all radial positions. However, untwisted
blades may be used in the interest of economy of production.
The fluid leaving the rotor blades encounters a row of stationary vanes
(Figure 6.9). These serve to straighten the flow, i.e., to remove the whirl
component V,,,, and to increase the pressure. Referring to Figure 6.7 it is
seen that fluid enters the vanes at the angle a,, and leaves axially. The axial
component V, may be reduced by flaring the walls of the annulus by several
degrees. The angle formed by the camber line tangent at the leading edge and
the vector V, should vary from root to tip. It should be designed to provide

Figure 6.9 Rotor and stator vanes.

FAN DESIGN 65

a positive incidence over the operating range, down to 50 percent of design

flow rate.

6.5 Fan Design

The design of an axial-flow fan can proceed in a manner similar to that of the
axial-flow pump. Specific speed N, can be determined from specified values
of rotational speed N in rad/s, volume flow rate Q in ft3/s, and head H in
ft? /s?.. The Cordier curve relation (6.21) may then be used to determine speci-
fic diameter D,. Finally, blade tip diameter D, is found using (6.22). The root
diameter D, is then calculated from the hub-tip ratio DID, which is chosen
to lie in the usual range of 0.25 to 0.7.
The velocity diagrams, as depicted in Figures 6.6 and 6.7, are then con-
structed as described in the previous section using (6.24) through (6.30). Fan
efficiency, estimated from (6.16), would be expected to lie in the range 0.77
to 0.86.
The fluid angles 6, and 8, required by the velocity diagrams can then
be used with cascade data such as those depicted schematically in Figure 6.10
to select a suitable solidity C/S for the mean diameter. This is expected to be
less than or equal to unity. The actual chord c would be selected to provide
an aspect ratio, i.e., blade height to chord, of 3 or more.
The angle of attack a required to produce C, calculated from (6.5) is
determined from wind tunnel cascade data for the airfoil shape and solidity
used in the design. Figure 6.8 schematically shows data of this type. The stag-
ger angle, i.e., the angle with respect to the axial direction at the chord line, is
then the difference between 8, and a.

a|”
B =Bp

Be
Figure 6.10 Cascade data.
66 AXIAL-FLOW PUMPS AND FANS

The blade is twisted to accord with the angles determined from velocity
diagrams for the tip and root diameters. Here we use the free-vortex condition,
ie., V,,.D = constant, to establish velocity triangles at the extremities of the
blade.

Reference

Stepanoff, A.J. 1957. Centrifugal and Axial Flow Pumps. Wiley, New York.

Problems

6.1 An axial-flow fan is to be designed with a tip diameter of 9.5 in. and a
hub-tip ratio of 0.5. Assuming that the fan is driven by a 1.5 hp motor and
has an overall efficiency of 80 percent, determine the flow rate and desirable
speed. The fan discharges air into the room through an exit area of 78.54 in.2.

6.2 Construct velocity diagrams for the rotor inlet and outlet at the mean
diameter for the fan considered in Problem 6.1.
 , Axial-Flow Compressors

7.1 Introduction

Originally a very inefficient machine, the axial-flow compressor was not used
to compress air in the gas turbine power plant. However, the development of
the science of aerodynamics, which accompanied the development of high-
performance aircraft, made possible its present use of gas-turbine compressors.
Now a highly efficient machine, it must be studied and understood thoroughly
by engineers.
This machine resembles the axial-flow steam or gas turbine in general
appearance. Usually multistage, one observes rows of blades on a single shaft
with blade length varying monotonically as the shaft is traversed. The differ-
ence is, of course, that the blades are shorter at the outlet end of the com-
pressor, whereas the turbine receives gas or vapor on short blades and exhausts
it from long blades. A close look at the blades shows that the compressor blade
deflects the fluid through only a fraction of the angle that the turbine blade
does. This point is illustrated by Figure 7.1. Figure 7.1 also indicates that the
concave side of the blade moves ahead of the convex side; the reverse is true
of the turbine blade. Clearly, the fluid receives energy from the compressor
blade and gives up energy to the turbine blade.
Aerodynamic analysis must be carried out for compressor blades, since
flow in the boundary layer encounters an adverse pressure gradient, which
may lead to separation, stall, and the consequent surge phenomenon discussed

67
68 AXIAL-FLOW COMPRESSORS

Compressor Turbine
blade blade

Figure 7.1 Blade comparison.

in connection with centrifugal machines. To avoid separation the pressure rise

must be small for each stage, in contrast with the very large pressure drops
found in turbine stages. Typically, about one-half of the enthalpy rise occurs
in the rotor and one-half in the stator.
The approach to compressor stage design is the same as that used for
axial-flow pumps and fans, except that the compressibility (density change)
of the gas must be considered in the overall process of multistage machines.
Fortunately, an abundance of theory exists, and many blade shapes have been
tested in cascade tunnels, so that the designer has a large stock of data to
draw on in his or her considerations.

7.2 Basic Theory

The theory utilized here is similar to that presented in Chapter 6 and will not
be redeveloped. The velocity diagram in Figure 7.2 shows that the blade de-
flects the relative velocity through the angle 6, - 6. The change in tangential
component AW,,, which is equal to AV,,, is proportional to the energy transfer
E. As in (6.3), we have

E = UAV, (7.1)

Figure 7.2 Velocity diagram fora compressor stage.

CASCADE TESTS 69

In terms of the axial component of velocity V, (7.1) becomes

E = UV, (tan 6, - tan B,) (2)

From (7.2) it is clear that if the through flow velocity V, remains con-
stand, the blade speed U increases with increasing radius, and the energy transfer
per unit mass E is to remain independent of radial position, then we must
vary the fluid angles B, and 8,, and hence, we must vary the blade angles.
This is, as previously discussed, the free-vortex condition, UAV,, = constant.
As before, with the axial-flow pump, we must have twisted blades in order to
achieve this equality of energy transfer along the blade. The variation of blade
angles then implies that 6,, will vary, and hence the degree of reaction R
varies. The latter has been defined by (2.20) and can vary between zero and
unity. It is found empirically and has been shown theoretically (see Shepherd,
1956) that a value of 0.5 is a near optimum for the degree of reaction produc-
ing maximum stage efficiency. Consequently, we find this value frequently
used for a design value at the mean diameter. Another design approach is to
use a value of 0.5 for R at all radial positions. Both bases for design are used,
as well as others not discussed here.

7.3 Cascade Tests

The theory discussed in the previous section relates energy transfer to fluid
angles, blade speed, and axial velocity through the velocity diagrams drawn at
the hub, mean, and tip radii. The development of a blade design requires
the use of wind tunnel results such as those shown in Figure 7.3 (Herrig et al.,
1957). Many such results are available to the designers and they are made for
very specific blade shapes. Thus the designer will generally specify the blade
shape for which results exist, and these proportions are given in the report
of the wind tunnel results. In addition, the tests are carried out for specific
values of solidity C/S and stagger angle. For example, Figure 7.3 gives results
for the NACA 68 (18): 10 airfoil shape, a solidity 0.75, and a fixed fluid
angle 8, of 60°. The stagger angle was varied to give a range of angles of attack
from 6° to 22°. From test data such as those of Figure 7.3, the designer may
select an angle of attack. A choice of 14° in Figure 7.3 corresponds to a value
of L/D of 62 and a fluid deflection of 20°. The designer knows that this
choice of blade profile, solidity, and stagger angle will produce 6, = 60° and
B, = 40°. Use of the theory begins at this point, and the designer will thus
construct a velocity diagram at the mean diameter with these values of
70 AXIAL-FLOW COMPRESSORS

32 0.06

28 0.05

2 0.04

8, deg Cp)
20) 0.03

16 0.02

l2 OOl

8 O

Angle of attack, deg

Figure 7.3 Cascade results for the NACA Profile 68 (18): 10. [Source:
Herrig et al. (1957).|

B, and B,. The designer also knows that the profile drag coefficient for this
design is expected to be 0.015.
A stage design may commence after the speed N, mass noe m, and
pressure ratio P,/P, (or energy transfer E) are specified. The given variables
N, Q, and H allow the calculation of specific speed, and a Cordier diagram can
be used to find the corresponding optimum specific diameter D,; for example,
we could use (6.21). The tip diameter can then be calculated from (6.22).
A trial-and-error procedure would then ensue, in which an assumed hub-tip
ratio would lead to a velocity diagram at the mean diameter. The axial velocity
component V, from this diagram must produce the required mass flow when
substitution is made in
th = pAV, (7.3)
where A is the annular area. We may illustrate the process diagramatically,
as in Figure 7.4.
PERFORMANCE 71

Check

Figure 7.4 Design process.

7.4 Performance

Prior to construction and testing of the prototype machine it is desirable to

determine estimated performance characteristics by means of calculation.
Normally, stage efficiency n, as well as multistage compressor efficiency are
based on total temperatures. Thus, referring to Figure 7.5, we have

Ne oe ou (7.4)
03 ~ *01
This is the same definition used in (5.4) for centrifugal compressors. In
Figure 7.5, state 1 denotes conditions at the rotor inlet and state 3 those at
the stator outlet. It has been shown, however, by Cohen et al. (1974) that the
incompressible definition (6.16) predicts the stage efficiency well, because
the rise of total temperature in the stage is sufficiently small. Cascade test
results may be used to determine values of lift and drag coefficients for the
blade profile, inlet Mach number, incidence, stagger, and solidity selected for
the compressor stage design, and (6.19) and (6.20) can then be used to esti-
72 AXIAL-FLOW COMPRESSORS

Figure 7.5 Enthalpy-entropy diagram.

mate stage efficiency. Noting that (7.4) can be rewritten in terms of the total
pressure ratio R,, we obtain

5 (14 AT ost) YICY-1) (7.5)

‘: Tor
as a convenient working formula. The stage total temperature rise AT gg is
related to energy transfer E by the relation

E=C, ATog (7.6)

The energy tranfer relation (7.2) is quite accurate for a single stage, because
the velocity profile V,(z) is nearly flat. The annular walls create a boundary
layer which causes peaking of V, near the mean radius. Thus (7.2) must be
multiplied by a factor A, called the work-done factor, and the resulting equa-
tion,

E = AUV, (tan B, - tan B,) (7a)

can be used for each stage with a constant value of A. The work-done factor
may be approximated from

A = 0.05(19.64 - In Ng) (7.8)

where N,, is the number of stages in the compressor.

The same basic relations, coupled with cascade data, can be used to
predict off-design performance. As ‘the flow rate through the compressor is
varied from the design value, the angle of incidence also varies, but the rotor
and stator fluid exit angles do not deviate appreciably from their design
REFERENCES 73

values. Thus it is possible to construct velocity diagrams for each off-design

flow rate, and from the indicated incidence to determine values of C, and Cp
from cascade test results.
The overall compressor pressure ratio can be determined from the pro-
duct of individual stage pressure ratios. Similarly, the overall total tempera-
ture rise is the sum of stage total temperature rises using the definitions of
(7.4) and (7.5) applied to the compressor as a whole. Thus,
M ae GED)i] aya
= 7.9
Me ATi lo; ey

where R, and AT in this equation denote the total pressure ratio and the
total temperature rise for the whole machine.
The compressor map for an axial-flow compressor will have the same
appearance as that shown schematically for the centrifugal compressor in
Figure 5.3. This plot of R, as a function of m with N as parameter shows
operational limits set by the phenomena of stalling at low flow rates and
choking at high flow rates. At low speeds, choking occurs in the rear stages
and stalling (due to high incidence) in the front stages, whereas the situation
is reversed at high rotor speeds. These phenomena can be predicted in advance
using indicators such as the critical Mach number M, based on inlet relative
velocity (usually M, * 0.7-0.8) to indicate the first appearance of sonic flow
in the blade passages, and the stalling incidence angle corresponding to the
maximum value of C, obtained in cascade tests.
Since temperatures increase in stages after the first, Mach numbers de-
crease. Thus the first stage will be the most likely site of shock losses. The
first stage may be designed for supersonic inlet velocities near the tips. The
leading edge of such blades will be sharp to accommodate attached oblique
shocks as discussed by Kerrebrock (1977). The blades are called transonic
in that they accommodate subsonic flow near the hub. Such a stage may be
desirable in aircraft compressors where the cross-sectional area is minimal.

References

Cohen, H., G.F.C. Rogers, and H.I.H. Saravanamuttoo. 1974. Gas Turbine
Theory. Longman, London.
Herrig, J., J.C.. Emery, and J.R. Erwin. 1957. Systematic Two-Dimensional
Cascade Tests of NACA 65-Series Compressor Blades at Low Speeds.
74 AXIAL-FLOW COMPRESSORS

NACA Tech. Note No. 3916, National Aeronautics and Space Admin-
istration, Washington, D.C.
Kerrebrock, J.L., 1977. Aircraft Engines and Gas Turbines, MIT Press, Cam-
bridge, Mass.
Shepherd, D.G. 1956. Principles of Turbomachinery, Macmillan, New York.

Problems

7.1 Use the cascade results of Figure 7.3 to determine the velocity triangles
(mean radius) and static pressure rise for a compressor stage having the fol-
lowing features: U = 1000 ft/s; ATp, = 54°F; B, = 60°; p = 0.00237 slug/
ft3.
7.2 Determine the mean radius, air angles, and blade length for the first
stage of a compressor having the following data: N = 150 rev/s; ATgg = 36°F;
m = 44 lb/s; Vi 4 Oo Nits. 590 ft/s; X = 0.96; R = 0.5 (mean radius);
Pou 1 atime, = S18cR.
 ; ef Gas Turbines

8.1 Introduction

The simplest gas turbine engine requires at least two major components besides
the turbine proper (see Figure 8.1). The gas (usually air) must be compressed
by a centrifugal or axial-flow compressor, and then it must be heated (usually
by burning a hydrocarbon fuel) in a combustor or heat exchanger. Gas is
delivered to the turbine inlet at an elevated pressure and temperature.
The ideal thermodynamic cycle associated with the simple gas turbine is
the Brayton cycle depicted in Figure 8.2. Process 1-2 is an isentropic compres-
sion, 2-3 an isobaric heating, and 3-4 an isentropic expansion. A more realis-
tic model of the gas processes would follow the dashed lines 1-2’ and 3+4’.
The latter processes reflect the compressor efficiency n, and turbine efficiency
n,- Besides component efficiencies n, and n,, the cycle thermal efficiency ny,
is very important. The latter efficiency is defined as

1 ae
pee We
(8.1)

where W, is turbine work per unit mass of gas, W, is compressor work, and
Q, is the heat added to the gas in process 2’-3.
Cycle thermal efficiency depends on the cycle pressure ratio P,/P, and
the cycle temperature ratio T,/T,. Usually, the design value of the peak
temperature T, is raised as high as possible, consistent with the required life

i
76 GAS TURBINES

Exhaust

Compressor

Figure 8.1 Gas-turbine power plant.

of the first-stage blades, and the cycle pressure ratio is chosen as that value
corresponding to maximum cycle thermal efficiency or maximum specific
output W, - W,. Thus determination of the optimum cycle pressure ratio is a
logical first step in the design ofa gas turbine, and it provides the turbomachine
designer with the thermodynamic state of the gas as it enters the first stage of
the turbine.
Another point that should be made is that the compressor and turbine
are interdependent. Usually mounted on the same shaft, the speed of one
must be the speed of the other. Furthermore, the pressure ratios and mass
flows are also roughly equal. The determination of a point of operation is a
process known as matching, and the two machines are matched when speeds,
mass flows, and pressure ratios are equal. Both compressor and turbine maps
are required to carry out the matching process.
In this chapter we shall consider primarily the design and performance
of axial-flow turbines, since this type is widely used in aircraft and stationary
power plants. Some space will be given to the simpler, radial-flow gas turbine
as well, for which a few applications have been found.

Figure 8.2. Thermodynamic cycle.

 BASIC THEORY 7

Val \y

Nozzle or stator blade

Figure 8.3 Velocity diagrams for a gas turbine.

8.2 Basic Theory

A gas turbine stage is shown schematically in Figure 8.3. A row of stationary

blades receives the gas in a nearly axial path and deflects it to a small nozzle
exit angle a of, say, 15°. The exit velocity V, is much larger than the inlet
velocity V,, and the axial component V,, at the inlet is typically that at the
exit, i.c., V,,. A near-isentropic expansion of the gas occurs in the nozzle
formed between the blades. The rotor blade, or bucket, turns the gas through
a large angle, say 75°. This large change of direction of the relative velocity
has a tangential component AW,, which gives rise to the large tangential force
on the turbine blade.
The key thermodynamic states within a single stage are indicated in
. Figure 8.4. The actual end states are 1 and 2 in the stator process and 2 and 3
in the rotor process. The corresponding total property states, i.e., 01,02, and
03, are found by constructing isentropic processes between the actual states
and the corresponding total pressure lines (isobars). The states 2’, 03’, and 3’
are those corresponding to an ideal (isentropic) expansion from the inlet to
the exit pressure for the stage. The stage efficiency n, can now be defined as

Tp,01 -T 03 (8.2)
a en re
" Toi ~To3
This is simply the ratio of actual work per unit mass of gas to ideal, or isen-
tropic, work between the same total pressures.
In order to solve for stage efficiency, it is necessary to establish the
velocity diagram. This is carried out from the known upstream conditions, V, ,
P,, and T,;the design conditions, U, E, and m; and the deflection angle B,
78 GAS TURBINES

Figure 8.4 Enthalpy-entropy diagram.

+ B,, determined from cascade tests. As was discussed in connection with

axial-flow compressors, a trial value of the hub-tip ratio produces a value of
UV paod m. Iteration leads to the required values of all three. The blade
speed is then used to find AW,, (equal to AV,,) from the Euler equation

AW, =— (8.3)

An arbitrary choice of a, is possible, but this angle is usually made small to

reduce the kinetic energy leaving the stages. (See Figures 8.5 and 8.6.)
In addition to the fluid deflected angle 6, + 6, , the cascade results supply
the loss coefficients for the stator and rotor blades, which are defined as

Poi - Poo
Vie
S ee (8.4)
Poo P, .

for the stator, and as

Vile PB02K -P 03R

Ree Oat tet Deer (8.5)
Pose ~ P3

Figure 8.5 Symmetrical diagram.

 BASIC THEORY 79

Figure 8.6 Construction of velocity diagrams.

for the rotor. These are useful in determining pressures P, and P,, as well as
in the calculation of stage efficiency n,.
We first modify (8.4) to read

_ Por /Po ~ Poa /Po

8.6
Papo ey
The ratio Pp, /P, is easily determined from the basic isentropic relation
2 :
ieee iene ie (8.7)
P, 2yRT,
where T, is obtained from the energy equation

T iy = 1) V2
oe a Z (8.8)
05 2yRT,
The stagnation temperature is constant in the nozzle, so that Tp, = Tg, , and
all inlet conditions are known at the outset. Using the experimental value of
Yg, (8.6) and (8.7) are solved for the ratio Py, /P,, and P, is found from the
known value of Pp, .The total pressure Py, is then calculated from the ratio
determined from (8.7).
The pressure P, is found in a similar manner using (8.5). Since the rotor
is moving, the relative total pressures Pp,p and Po3p are used. They are calcu-
lated from
P ~1)w2 |¥/(7-1)
ae? Gree ies (8.9)
P, 2yRT,
and

Ip = iywe |V/O-2)
JOSR 2 |efCees. (8.10)
P, 2yRT3
where T, is found from the energy relation
80 GAS TURBINES

BOE (8.11)

using T)3 determined from the energy transfer, i.e.,

E
To3 = Tos Sn (8.12)
p
We must use the empirical value of Yp in (8.5) to calculate Po. p/P3. Since
Poor is known from (8.9), we can calculate P3 directly.
Finally, to obtain the stage efficiency n, using (8.2), we must calculate
a value of Tj. Referring to Figure 8.4, it is evident that P5,/P3 is equal to
Po3/P3, and thus that (V,)?/T3 is equal to Vai Therefore, we write

(y- V3
fo fo (8.13)
YR13
where
ve' & Yil /Y ican

is used to determine T,. Thus the stage efficiency is easily determined from
the cascade loss coefficients Ys and Yp. These coefficients should include
not only the profile losses obtained by direct measurement in the cascade
tunnel, but also the effects of annulus, secondary, and tip-clearance losses, as
discussed by Cohen et al. (1974). These losses are expressed in terms of Cy in
(6.17), (6.18), and (6.19). It is easily shown that they are expressible as loss
coefficients in the form

Yor +
Vee us +
Ve = (Coe
wer = , +
Ger Ce) :
um wr
aa 8 WS

Se Bee i DBP AScogiay cos? a : )

for the stator, and in the form

Vet
vp Ye SC Cy Eee)
, wm we ’ ow wee C cos* By
(8.16)
S cos? B3 cos3 6,

for the rotor. If we denote the measured profile loss coefficients by Ymg and
Y mr for stator and rotor, respectively, then we can write

Vora Yivoit Vout Yo tier C1

and

YR oYipt ROR UR (8.18)

DESIGN 81

The corrected values of Yg and Yp should be used in (8.4) and (8.5) to obtain
P,,P,, and flee

8.3 Design

As indicated earlier in this chapter, a thermodynamic analysis allows the choice

of a suitable cycle pressure ratio P,/P, (see Figure 8.2). Metallurigcal con-
siderations govern the choice of the turbine inlet temperature T,. Using an
assumed turbine efficiency n,, the turbine work W, for the cycle is computed
using the relation
P, OEY)
W,= ConsT3 1- (72) (8.19)

The stage work, or energy transfer E, is related to the turbine work through

at
E=— (8.20)

where N, is the number of stages. The mass flow rate m is related to the tur-
bine power P through

i
° P
(8.21)

We have already seen how the designer can construct a stage velocity
diagram from a knowledge of m, E, and N. We assume a specified rotational
speed N, and we have seen that E can be obtained from (8.20), providing that
a value is assumed for the number of stages N,. An approximate formula for
estimating N, initially was given by Vincent (1950) as

ee W, sin? a
3 (8.22)
@ V2 (1 - k2 sin? a)

where V,, is the assumed axial velocity, a is the nozzle angle, and ky is a blade
friction coetficient, typically having a value between 0.9 and 0.95. The stage
formula (8.22) was derived by assuming that all stage velocity diagrams are
symmetrical (see Figure 8.6) and have the same nozzle angle a and axial ve-
locity V,. The final design may utilize velocity diagrams different from this,
but this assumption is useful in the early stages of the design process.
The velocity diagram is constructed for each of the N, stages in the
manner previously discussed in connection with Figure 8.3. The pressures P,
82 GAS TURBINES

and P, and temperatures T, and T, are calculated, as previously discussed,

for the mean diameter. From these values density, flow area, and blade height
are calculated at sections 1,2, and 3.
Several parameters of the design can be checked to be sure they are in
acceptable ranges. The blade loading coefficient ,, defined as

Vp = 4 (8.23)

should lie between 1.5 and 5. The flow coefficient ¢, defined as

= —
Na
(8.24)
? U

would be in the range 0.4 to 1.0. The degree of reaction R, defined as

T, -T
Recess (8.25)
di eats ¢3

should have a value between 0.4 and 0.6 at the mean diameter. The Mach
numbers M, and M,, based on V, and V3, respectively, should be subsonic
and as low as possible. The nozzle exit Mach number M, , based on V, should
be subsonic or slightly supersonic, but not greater than 1.2. The correspond-
ing relative Mach number Mp,,,, based on W., should be less than the critical
Mach number, which would normally lie in the range 0.7 to 0.8. Adjustments
to the velocity diagram can be made, if necessary, to satisfy the above condi-
tions.
The free-vortex condition can be utilized to obtain velocity diagrams at
the root and tip of the blade. This is frequently done, since it yields a constant
value of energy transfer and of axial velocity, as has been pointed out pre-
viously. Twisting of blades, which is required to achieve the free-vortex con-
dition, is often avoided to reduce manufacturing expense. If a condition other
than the free-vortex condition is assumed, then the basic equations must be
utilized to determine the variation of V,, i.e., to find V, at the root and tip
of the blade. Then the velocity triangles at root, mean, and tip locations can
be constructed, and the energy transfer and mass flow rate can be determined.
The blade angles and dimensions to produce the velocity triangles must
be sought. The stator gas angles a, anda, differ from the stator blade angles
dp, and ag,, as shown in Figure 8.7. These differences in angles may be ex-
pressed as incidence i, defined by

i=, - 4p; (8.26)

 DESIGN 33

Figure 8.7 Blade angles and gas angles.

and by deviation 6,, defined by

5p = dp, - 4, (8.27)
For the moving blades we define incidence by

i= eee} (8.28)
and deviation by

55 = Bps - B3 (8.29)

The profile loss coefficient Y, obtained from a cascade test, is usually

plotted as a function of incidence, as illustrated in Figure 8.8. Curve A repre-
sents data from impulse blading, while curve B depicts reaction-blading re-
sults. The incidence corresponding to the lowest value of Y would generally
be chosen for the design condition. In the case of curve B, we observe a con-
siderable range of values of i in which Y is low. A design incidence near the
middle of the low-value range should result in good off-design performance.
The selection of a design incidence allows us to calculate the blade inlet angle
from (8.26) and (8.28) using gas angles from the velocity diagrams.
Deviation can be calculated from an empirical relation given by Horlock
(1973):

ie ad eg (8,30)
84 GAS TURBINES

acs B

Figure 8.8 Blade loss coefficient from cascade tests.

where the camber @ is estimated by

0 = ap, + Gpo (8.31)

for the stator, and by

6 = Bay + Bas | (8.32)

for the rotor. The constant m depends on stagger angle y, (see Figure 8.7).
For the circular-arc camber line, m varies from 0.21 at yg = 0° to 0.17 at yg =
60°. For the parabolic-arc camber line, m is 0.12 at 0° stagger angle and is
0.06 at 60° stagger angle. Thus, knowing i and 5p we are able to determine
the blade angles for stator and rotor. )
Typically, the blade will be laid out in the form of an airfoil along a
circular arc or parabolic camber line using a chord c of from one-half to one-
fourth the blade length. A typical (maximum) thickness-to-chord ratio is 0.1,
and this maximum thickness is located at the 40 percent chord position. The
optimum spacing between blades at the mean diameter may be determined
from the relations given by Shepherd (1956), i.e.,

tana, - tana,
= 2.5 cos? a, (8.33)
ala COS Ay

for the stator, and from

c tan B, - tan B;
— = 2.5 cos? (8.34)
S Bs cos B,

for the rotor. Generally, the viaue of C/S will lie between 1.0 and 1.8.
The aerodynamic design is thus complete. However, calculation of cen-
trifugal and bending stress must also be carried out to ensure safe operation
with available materials. Stress calculations are treated by Cohen et al. (1974).
 RADIAL-FLOW TURBINE 85

Figure 8.9 Velocity diagrams for radial-flow gas turbines.

8.4 Radial-Flow Turbine

A typical radial-flow turbine is constructed like a centrifugal compressor with

radial vanes. The difference is that gas flows from the outer radius to the inner
radius, rather than the reverse. If the vanes are purely radial, except at the
trailing edges, then the ideal velocity diagrams would appear as in Figure 8.9.
With the exit velocity V, axially directed, the energy transfer is simply Lae
The thermodynamic processes are the same as occur in an axial-flow
stage. Hence, we can use the enthalpy-entropy diagram of Figure 8.4. In lieu
of the diffuser vanes of the centrifugal compressor, we find nozzle vanes sur-
rounding the rotor. The expansion from state 1 to state 2 occurs in these
nozzles. Losses in the stator are typically expressed in terms of the enthalpy
rise due to friction, i.e.,

:
_ 2(hy= - hy)
(8.35)
2

where §, is the loss coefficient. In the rotor the loss coefficient ¢, is given by

aTea ie AW fe
The turbine efficiency is typically defined as

Hour bs: (8.37)

ie ho; -h3
Starting with the velocity diagrams and assumed loss coefficients, we can easily
determine intermediate and exit pressures from h, and hy. A knowledge of P;
leads to h, and 7.
Dixon (1975) presents the variation of turbine efficiency with specific
speed and with nozzle angle. In this work it is shown that the maximum effi-
86 GAS TURBINES

ciency occurs at a specific speed of 0.64, exit hub-tip ratio D,3/D,3 = 0.4,
D,3/D, = 0.7, and a, = 74°. Shepherd (1956) also presents turbine efficiency
as a function of Mach number and Reynolds number. Shepherd states that
efficiencies in the range 0.75 to 0.85 should be expected for small inward-
flow radial turbines.
Because of the uncomplicated geometry and the simple expression for
energy transfer, preliminary design calculations are extremely easy for this
machine. An estimate of turbine efficiency can be made from the literature
(e.g., see Shepherd, 1956).

References

Cohen, H., G.F.C. Rogers, and H.I.H. Saravanamuttoo. 1974. Gas Turbine
Theory. Longman, London.
Dixon, S.L. 1975. Fluid Mechanics and Thermodynamics of Turbomachinery.
Pergamon, Oxford.
Horlock, J.H. 1973. Axial Flow Turbines. Krieger, Huntington, New York.
Shepherd, D.G. 1956. Principles of Turbomachinery. Macmillan, New York.
Vincent, E.T. 1950. The Theory and Design of Gas Turbines and Jet Engines.
McGraw-Hill, New York.

Problems

8.1 Estimate the number of stages needed for a 50 percent reaction turbine
operating at standard sea level in a basic open-cycle gas-turbine plant. Turbine
inlet temperature is to be 1660°R. Assume cold-air properties throughout,
nozzle angle = 20°, axial component of velocity = 420 ft/s, minimum leaving
kinetic energy, n, = 0.90, and blade friction coefficient Kp = 0.90. Shaft speed
is 10,000 rpm and shaft power is 9000 hp. Cycle pressure ratio is 8.3.
8.2 Draw a mean velocity diagram for the turbine described in Problem 8.1.

8.3 Estimate hub and tip diameters for the last stage of the turbine of Prob-
lem 8.1.

8.4 A gas-turbine stage is designed for which To, =1100K, Po, =4 bar, V55
519 m/s, U = 340 m/s, V,. = Vaz = 272 m/s, a, = 10°, Tp, = 955 K, andB, =
20.5° (see Figure 8.6). The loss coefficients, calculated from (8.17) and (8.18),
are Y, = 0.0688 and Yp = 0.152. Determine stage efficiency.

8.5 A radial-flow gas turbine with flat radial vanes is to run at 24,200 rpm.
Gas is to enter the rotor at a radius of 6 in. and exit at a mean radius of 3 in.
Exhaust gases are to leave at 14.7 psia and 700°F and are to have a relative
Mach number of 0.75. Calculate the mass flow rate to produce 100 hp, and
estimate the blade height at the exit.
 Steam Turbines

9.1 Introduction

Steam turbines, like gas turbines, are predominantly axial-flow units. They are
used extensively in power plants to drive electric generators, as are gas turbines,
but are usually much larger than gas turbines. In addition, the large units
typically use much higher pressures in the first stages and lower pressures in
their later stages than do the large gas turbines.
Steam-turbine calculations are different than gas-turbine calculations in
that tables of steam properties are substituted for simple gas relations. The
basic features of the axial-fiow turbine are the same for both steam and gas
turbines; however, the steam can contain droplets of liquid water, making the
flow a two-phase flow, and thus a further complication is added.
The simplest ideal cycle in which the steam turbine performs a function
is the Rankine cycle. This cycle is pictured on the temperature-entropy dia-
gram of Figure 9.1. The line drawn from point 1 to point 2 represents the
ideal expansion of steam in a turbine from a superheated state to a wet state.
The actual process, i.e., that with an entropy increase, is shown as a dashed
line 1-2’. The wet steam is exhausted from the turbine into a condenser where
it is condensed to a saturated liquid at point 3. A pump and boiler then act
upon the water to raise its pressure in process 3-4 and heat it in process 4-1.
The boiler, or steam generator, delivers the superheated steam to the first stage
of the turbine in state 1.
87
88 STEAM TURBINES

Saturated steam curve

Figure 9.1 Temperature-entropy diagram for a steam power cycle.

The thermal efficiency of the power-plant cycle is calculated by divid-

ing the net work, i.e., the turbine work minus the pump work, by the heat
supplied in the boiler. Although the cycles used in modern power plants in-
volve reheating the steam at one or more points during the expansion and
other complexities, the turbine designer can enhance overall cycle efficiency
by simply increasing the turbine work realized during the expansion 1-2’.
As in previous chapters, the individual stage design will be emphasized
in the present chapter. After the decisions regarding the cycle thermodynamics
are made, and the inlet and exhaust conditions are thus established, the tur-
bine designer must divide the expansion process into smaller stage pressure or
enthalpy drops and then proceed to design individual stages.
Both impulse and reaction axial-flow stages will be considered, since the
former does find application in the early stages of large reaction turbines, and
the impulse stage is used in small turbines. The Ljungstrom turbine, which
utilizes a radial flow of steam, will not be treated here. The latter is analyzed
by Shepherd (1956).

9.2 Impulse Turbines

If the turbine stage is to be an impulse stage, the entire pressure drop must
occur in the nozzles. The purpose of the moving blade is to reduce the kinetic
energy of the steam and transfer this energy to work done on the moving
blades. The resulting energy transfer may be evaluated using (2.19):
2_y2
Ng V3
Loe aaaeT (9.1)
IMPULSE TURBINES 89

Figure 9.2 Velocity diagram for an impulse turbine.

where the subscripts 2 and 3 refer to rotor inlet and exit, respectively. Typi-
cally, the velocity diagram is that shown in Figure 9.2. Note that the relative
velocity W, is equal in magnitude to W,. Thus in the ideal impulse turbine
the relative velocity changes direction from B, to B,, but the magnitude holds
constant. However, the absolute velocity leaving the rotor is much reduced
from the velocity V, exiting from the nozzles.
Since the boundary layers which form on the blade surfaces actually slow
the steam in the passage between the blades, velocity coefficients K, and Kp are
usually defined for the stator and rotor, respectively, by the following relations:

Ko = s (9 2)
: \ 2

and

Kp (9.3)
= W> ;

where the thermodynamic states 1, 2, 2’, and 3 are indicated in Figure 9.3.
These coefficients can be estimated from values given in Table 9.1. The values
in Table 9.1 were calculated from empirical data presented by Horlock (1973),
although the small effect of the Reynolds number is neglected. The quantity
AB denotes the deflection of fluid by the stator or rotor blades, and H/c,
denotes the blade aspect ratio, i.e., blade height H over axial chord c,.
The Euler turbine equation (2.16) applied to the steam turbine becomes
either

E=U(V,,. - Vy3) (9.4)

or

E = U(Wy, - Wy3) (9.5)

90 STEAM TURBINES

Figure 9.3. Expansion process in an impulse turbine.

Table 9.1 Velocity Coefficients Kg and Kp

AB(deg)

20 40 60 90 100

1 Oe) WL Wee Ou 393

3 O98) 1019S.) 0:97. 70.96 095
co ORO OFF ORG aos 0.97

Applying the latter form and expressing components in terms of the steam
angles B, and 6, relative to the moving blades, we have

E=U(W, sin B, + W3 sin B3) (9.6)

Using (9.3) we have

E = UW, (sin 8, + Kp sin B3) OD)

Using the relation, derived from Figure 9.2, that

W, sin
B, = V, cosa-U (9.8)

(9.7) becomes

Kp sin B3
E=U(V, cosa-U) (14 (9.9)
sin B,

If we then define a blade efficiency np, by

“ BS
Baa (9.10)
V3
IMPULSE TURBINES 91

which is a reasonable measure of the effectiveness of impulse-turbine blading,

we find that

_2U U Kp sinB3
Tp
= —|\cos a= ies Cea
Vv2 Vv2 sin
B,

By differentiation with respect to U/V, we are able to find the optimum ratio
U/V.,, for maximum blade efficiency:

Ua COsia
(9712)
Ve opt 2

Since the nozzle angle a usually falls between 15° and 25°, we expect the
optimum ratio of blade speed to nozzle exit speed to lie between 0.483 and
0.453.
The actual value of np,,,, depends on the values of Kx, B,, and B;,
but it is usually quite high. It varies, however, from zero at zero blade speed
to zero at U = V,, as shown in Figure 9.4.
Torque as a function of blade speed is also shown in Figure 9.4. Torque,
denoted by T in Figure 9.4, varies in a straight line from its maximum at zero
blade speed to zero at U = V,. This easily shown by noting that torque is
power over rotational speed. Thus

ea]
T= = mE (9.13)

and

y=Nl (9.14)

U/Vp
Figure 9.4 Efficiency and torque for an impulse turbine.
92 STEAM TURBINES

fu fu
eae
Sa a
Nozzle vA Swe

Figure 9.5 Curtis stage.

Substitution of (9.9) and (9.14) into (9.13) yields

mD ND Kp sinB
T=" (v,cosa- ~) (+ E (9.15)
sin B,

in which T is a linear function of N. The design torque Tp is the torque at

maximum efficiency and is obtained by substitution of (9.12) in to (9.15).
This gives
mD Kp sin 3
Ip= es (V, cos a) (aF ee) (9.16)

A velocity-compounded, or Curtis, stage is sometimes used when the

blade speed is lower than the optimum, and it is desirable to extract more
kinetic energy from the steam than is possible with a single row of moving
blades. An additional row of moving blades is added to the same turbine wheel,
and a fixed row of blades is interposed between the moving rows, as is shown
in Figure 9.5. Thus the steam leaves the fixed row having a velocity V, only
slightly less than V, and then passes over the second row of moving blades,
leaving with velocity V;, which ideally is axial. Additional moving rows may
also be added if required to reduce the absolute exit velocity to its minimum
(axial) value. An analysis of blade efficiency, similar to that carried out above,
shows that
U cos a
= (9.17)
Veopt 4

for the stage depicted in Figure 9.5 with Kp = Kg = 1. Thus (9.17) indicates
one-half the blade speed needed for maximum efficiency with only one row
REACTION TURBINES 93

of moving blades and thus illustrates that velocity compounding implies lower
biade speeds for the same nozzle exit velocity.

9.3 Reaction Turbines

For a large steam turbine, the designer could choose to use pressure-compound-
ing, i.e., use a row of nozzles after each row of moving impulse blades. A more
effective procedure, however, is to allow the expansion to proceed in the
moving blades as well as in the nozzles. This results in a higher optimum blade
speed, a higher maximum stage efficiency than for the impulse stage, and a
broader range of blade speeds corresponding to high efficiency.
Referring to the velocity diagram in Figure 9.2 again, we see that a
reaction turbine diagram would be similar, but would have W, considerably
longer than W.. This follows from the steady-flow energy equation (2.9)
and the Euler turbine equation (2.18). Eliminating the energy transfer results
in the form

Nimes:
W> W3 (9.18)

Thus a pressure drop in the moving blades is accompanied by aloss of enthalpy

and a gain in relative kinetic Ww? /2. Figure 9.6 illustrates this increase of rela-
tive velocity, and from it we should also observe that the blade would not
have the symmetry that would be expected in an impulse blade.
The velocity coefficients from Table 9.1 can still be used in the analy-
sis of this type of turbine if we approximate the enthalpy rise associated with
blade friction in the same manner used previously in the nozzles of the im-
pulse turbine. Thus for the stator of the reaction turbine we can write

v3 (V5?
ho, = h, ey + (9.19)
2

Figure 9.6 Velocity diagram for a reaction turbine.

94 STEAM TURBINES

Figure 9.7 Thermodynamic processes in reaction turbines.

where the states 2 and 2’ are those depicted in Figure 9.7. Using (9.2) to sub-
stitute for V, , we have

SCRE AOD:
hy - hy = 5 (9.20)

Since V, is the ideal velocity resulting from an isentropic expansion, it is easily

found, and then (9.20) can be used to find h,, the actual entalpy leaving the
nozzle.
The same method applied to the rotor, but using relative velocities in
place of absolute velocities, leads to

| _ (W5)? - K3(W3)? (9225)

h3 -h3 = 5

Thus, knowing pressures P, , P,, and P3, we can easily determine V, and W;,
and we can then use (9.20) and (9.21) to estimate the actual enthalpies h,
and h,. The velocities V, and W, are also easily obtained from the velocity
coefficient.
With the velocities V, and W, known, a velocity diagram, such as that
shown in Figure 9.6, is easily constructed for a given nozzle angle a and blade
speed U. The axial and tangential components of V, are easily calcualted and
used to find W,. Assuming that a prior knowledge of A is required to select
K,, we thus know 83, and we can determine W,,, from W3, and finally V,,,
and V,. Thus all the velocities and angles may be determined using the velocity
coefficient K, to determine the enthalpy rise due to friction.
Energy transfer can be found with the help of Figure 9.6. The Euler
turbine equation becomes

E=U(V, cosa+ W3 sin B3 - U) (9.22)

PROBLEMS 95

Since the stage efficiency can be defined as in (8.2), and since

Eh, hg, (9.23)

efficiency is a maximum when E is a maximum. Since the symmetrical velocity
diagram is commonly used, we will, for convenience, assume it here, i.e., we
assume that V, = W, and V, = W,. For this case, (9.22) becomes
E=U(2V, cosa-U) (9.24)

Differentiation with respect to U/V, shows that maximum efficiency cores-

ponds to

=cosa (9.25)
V2 opt

which is twice the optimum blade speed found for the impulse turbine. As
mentioned previously the resulting stage efficiency is higher and the curve
flatter than for the impulse turbine, implying better off-design performance
as well. The range of speed ratios for high efficiency is roughly 0.7 < U/V, <
ey

9.4 Design

The design of steam turbines will proceed along the same lines previously
outlined in Chapter 8 for gas turbines. Of course, steam tables or Mollier
charts must be applied in lieu of perfect-gas relations for the determination of
properties, but otherwise the same methods apply. For a detailed account of
the details of steam-turbine construction and design, the reader is referred to
a book by Church (1950).

References

Church, E.F. 1950. Steam Turbines. McGraw-Hill, New York.

Horlock, J.H. 1973. Axial Flow Turbines. Krieger, Huntington, New York.
Shepherd, D.G. 1956. Principles of Turbomachinery. Macmillan, New York.

Problems

9.1 Construct a velocity diagram for a single-stage impulse turbine having a

nozzle angle a = 20°, a mean radius r,, = 10 in., a blade height H of 2 in.,
a speed N = 7200 rpm, an aspect ratio H/C, = 2, and equal relative steam
angles (6) = B3). At the nozzle exit, the enthalpy hj = 1183.5 Btu/lb and
pressure is 14.0 psia. V7 = 1256.6 ft/s.
96 STEAM TURBINES

9.2 Determine the power produced by the turbine in Problem 9.1, assuming
full admission through all nozzles in the annulus.

9.3 Assuming that turbine power is varied by progressively blocking nozzles,

i.e., through the use of partial admission, show that the steam consumption
of the turbine in Problem 9.1 is a linear function of turbine power. This
straight line is the Willans line. Plot the Willans line using lb/hr units for steam
flow and kilowatts for turbine power.
 1Q Hydraulic Turbines

10.1 Introduction

The oldest form of power-producing turbine is that utilizing the motive power
of water. Flowing rivers, streams, and waterfalls have had their energy extracted
by vanes or buckets fixed to the circumference of rotating wheels. Although
these waterwheels were designed and built as late as the nineteenth century,
and some of them were fairly efficient, they eventually gave way to more
powerful machines requiring reservoirs.
Modern installations utilize a reservoir of water, which is usually water
collected from a flowing river. The level of the reservoir next to the hydraulic
power station is maintained at a nearly constant elevation by controlling in-
flow from other reservoirs further upstream. The water flows into the hydrau-
lic turbine through a large pipe, known as a penstock. It leaves the turbine
through a diverging duct, known as a draft tube, and enters a downstream
reservoir, known as a tail race. The available head, which ranges from several
feet to several thousand feet in existing plants, is the vertical distance between
the free surfaces of the water in the reservoir and the tail race. A schematic
diagram is shown in Figure 10.1.
The designer is generally presented with an available flow rate Q, based
on runoff records, and an available head H. The turbine speed N will also be
given, since it wili usually be required to drive a generator at a prescribed rate.

a7
98 HYDRAULIC TURBINES

Reservoir
level
vie
Turbine

Draft tube

Tailrace

Figure 10.1 Hydraulic power plant arrangement.

The choice of the type of turbine will then follow naturally after a calculation
of specific speed, since we can observe in Table 3.2 that the ranges of specific
speeds corresponding to peak efficiency are quite different for Pelton, Francis,
and Kaplan Turbines. Thus the Pelton, or tangential-flow, turbine is most
efficient for N, < 0.3, the Francis, or radial-flow, turbine is best for N, be-
tween 0.3 and 2.0, and the Kaplan, or axial-flow, turbine is best for Nea2u:

10.2 Pelton Wheel

The Pelton-type turbine is illustrated in Figure 1.3. For this turbine the pen-
stock ends in a nozzle which creates a high-speed waterjet. The latter impinges
on vanes in the form of hemispheres or half-ellipsoids. The force on the vanes,
created by deflection of the water through just less than 180°, drives the
wheel around against the resistance of a load.
As indicated in Figure 10.2, the absolute velocity V, of the jet is the
arithmetic sum of its relative velocity W, and the vane speed U. Leaving the
vane at some small angle 8, , the tangential component of the absolute velocity
is given by

Vio
= W, cos, -U (10.1)

Thus the energy transfer can be expressed as

4._______________.

Vane TB Vv,
We

Figure 10.2 Vane velocity diagram for a Pelton wheel.

PELTON WHEEL 99

E = U(V, + W, cos, - U) (10.2)

Since ideally

W,=W,=V, =U (10.3)
we have

E = U(1 + cos By (V, - U) (10.4)

If a velocity coefficient K is defined to account for the retardation of W,,
then (10.4) becomes

E = U(1 +K cos 8B, (V, -U) (10.5)

The hydraulic efficiency is the ratio of energy transfer to available
energy gH. The latter is almost entirely converted to jet kinetic energy, i.c.,
gHC = V?/2, so that
ny = 5 (10.6)

Differentiation with respect to U/V, leads to the result that maximum hy-
draulic efficiency corresponds to

zl (10.7)
Vi opt 2

The shaft power P of the Pelton turbine may be estimated from the
foregoing relations if leakage and mechanical losses are also considered. The
latter losses are usually small, e.g., 3 to 5 percent, and are accounted for in
the volumetric and mechanical efficiencies n, and n,,, as with pumps. Thus
we can write

Pe Oy wae ues)
or, in terms of volume flow rate and head,

P= Nm NyNyPQsH (10.9)
in which the product of three efficiencies is usually denoted by n, the over-
all efficiency. It should be noted that the flow rate also depends on the head,
ee

o=(7)arccasiy"” (10.10)
100 HYDRAULIC TURBINES

V2

Figure 10.3. Velocity diagram for a Francis turbine.

where C is a coefficient which accounts for head loss due to friction in the
nozzle, control valve, and penstock, and d is the jet diameter. Of course,
several jets may be used around the periphery of the wheel, in which case the
flows would be additive.
The ratio of wheel diameter D to jet diameter d varies from 6 to 25.
The buckets are positioned close together to avoid spillage, the number of
buckets varying from 20 to 30 per wheel. Such designs result in overall
efficiencies of 80 to 90 percent.

10.3 Francis Turbine

The Francis turbine is a radial-flow reaction machine, much like a centrifugal

pump with the flow direction reversed. Such a turbine is shown schematically
in Figure 1.4. Flow from the penstock enters a spiral casing, which distributes
water to the adjustable guide vanes located around the turbine wheel. The
water leaves the guide vanes at angle a and speed V,, as depicted in Figure
10.3. Ideally, the vanes are designed so that the exit velocity V, has no whirl
component, i.e., it is axial. Thus energy transfer is calculated from

E=U,V, cosa, (10.11)

The shaft power P will again be calculated from (10.9) with hydraulic
efficiency defined by

e. U, Vi COs a,
Ny Sener (O12)

and flow rate calculated from

CbrD
= (10.13)
~ (2gH)'2
FRANCIS TURBINE 101

where b denotes axial width at the inlet and C denotes a velocity coefficient
which depends on frictional resistance along the entire flow path from
reservoir to tail race. Typically, the coefficient C is in the range 0.6 to 0.7.
A Cordier diagram, such as that appearing in Figure 3.3, can be used to
determine a suitable wheel diameter D. This is obtained from the specific
diameter after the Cordier diagram is entered for a known specific speed. The
desired flow rate is assured by determining the appropriate axial width b, of
the runner from (10.13).
Figure 10.4 illustrates differences in vane design arising from differences
of the specific speeds of Francis runners (rotors). Low specific speeds imply
low flow rates, which result in a smaller axial width b, and angle a, , as well
as a smaller exit diameter at the base of the rotor. Water leaves the high-
specific-speed vane with a large axial component, in contrast to the large
radial component present at the vane exit of a low-specific-speed runner. The
inlet vane angle 6, is also increased and can be as much as 135°. The lower
edge of the high-N, vane must be twisted to provide axial discharge velocities

are Lea

_— Yeap
(a)

Figure 10.4 Francis turbine runner designs: (a) low specific speed; (b) high
specific speed.
102 HYDRAULIC TURBINES

at varying blade speed. The rotor exit diameter is as large as, or larger than,
the mean inlet diameter, as shown in Figure 10.4, to accommodate the higher
flow rate associated with larger values of N,.
As indicated in Figure 1.3, the connection between the turbine casing
and the tail race, called a draft tube, is installed to conserve the available head
between these two levels. The draft tube accomplishes a diffusion as well,
and the exit velocity from the system is thus reduced, improving overall
efficiency.

10.4 Kaplan Turbine

The axial-flow hydraulic turbine is the third type of important hydraulic

turbine. It is used primarily for low-head applications, i.e., for H < 150 ft.
Such a turbine is shown in Figure 1.8. As is seen in this schematic representa-
tion, adjustable guide vanes are located around the inside of a volute casing,
as with the Francis turbine. Also, the vanes of the propeller-type runner cor-
respond to the lower part of the high-N, Francis turbine vanes, where axial
flow predominates. If the airfoil-shaped vanes are adjustable, the turbine is a
Kaplan type; otherwise it is called a propeller type.
The same principles previously applied to axial-flow steam and gas tur-
bines apply to this machine, except that the fluid is treated as incompressible.
Uniform axial velocity and the free-vortex variation of whirl velocity can be
applied in the analysis of these blades as well. Whirl is imparted to the water
in the inlet guide vanes, and it is deflected through a small angle by the blade.
The exit velocity has no whirl component (see Figure 10.5) ideally, so that, as
with the Francis turbine, we have

E=UV,, (10.14)

Figure 10.5 Velocity diagram for a Kaplan turbine.

CAVITATION 103

BeUVuy
NH (10.15)
gH

The flow rate can be expressed in terms of a hub diameter D,, a tip diameter
D,, and a coefficient C,, which accounts for head loss due to friction. Thus
we may write

Cz - (D? - D2)C, (2gH)¥2 (10.16)

According to Kadambi and Prasad (1977), the coefficient C, can vary from
0.35 to 0.75. Power to the turbine shaft is calculated from (10.9), as before.

10.5 Cavitation

As previously discussed in Chapter 4, the occurrence of cavitation is avoided

in pumps, and also in hydraulic turbines, by using a safe value of the net
positive suction head, NPSH. This is defined by
NPSH
= Haim + Hy - H vap (10.17)
where all terms in (10.17) are expressed in feet of liquid flowing, and the sub-
scripts refer to atmospheric, total, and vapor pressures. H;, is the total (static
plus dynamic) head, referred to the centerline of the turbine or pump. If the
turbine has a draft tube, it is negative. Thus NPSH is a measure of the differ-
ence between the absolute static pressure at the turbine blade and the vapor
pressure of the water in the turbine. If the NPSH falls below a certain critical
value, the lowest static pressure in the turbine can be equal to the vapor pres-
sure of the water, and cavitation can likewise occur.
An important ratio, known as Thoma’s cavitation parameter og, is de-
fined by

gaueae
H
(10.18)
Critical values of o, corresponding to incipient cavitation, have been correlated
with specific speed N,, defined as NP1/2/H*/4. For the Francis turbine Shep-
herd (1956) gives the critical Thoma parameter correlation as
N g2

o= 0.625(™*) (10.19)

in which specific speed is calculated using rpm, hp, and ft units for N, P, and
H, respectively. Similarly, the critical value of o for Kaplan turbines is given as
104 HYDRAULIC TURBINES

he PNY?
o = 0.28
+ —— |—— (10.20)
7.5 \100

Values of o calculated from flow conditions should always exceed values of o

calculated from (10.19) and (10.20) in order to assure avoidance of cavitation.

References

Kadambi, V. and M. Prasad. 1977. An Introduction to Energy Conversion.

Vol. III, Turbomachinery. Wiley, New York.
Shepherd, D.G. 1956. Principles of Turbomachinery. Macmillan, New York.

Problems

10.1 Develop an expression for design torque (maximum efficiency) for the
Pelton turbine in terms of wheel diameter and jet characteristics.

10.2 For the turbine of Problem 10.1, H = 100 ft of water, By =0,K =1,
jet diameter = 6 in., C = 0.94, n = 0.9, and N = 120 rpm. Find the following
for maximum efficiency:
a. Flow rate
b. Power
c. Torque
d. Wheel diameter

10.3 A hydraulic turbine is to be designed to produce 36,600 hp at 93.7

rpm under a 54 ft head. A model of the turbine is to be designed to produce
50 hp at 554 rpm. If model efficiency is assumed to be 88 percent, find
a. Diameter of the model
b. Diameter of the turbine (prototype)
c. Type of turbine
 11 Wind Turbines

11.1 Introduction

The natural motion of atmospheric air provides us with an excellent source of

available energy for conversion to useful work. The moving rotor of a wind
turbine arranged with a horizontal or vertical axis provides the means of energy
conversion.
Wind turbines have been utilized by humans for hundreds of years for
pumping water, grinding grain, and even generating electricity, but a renewed
interest in the ancient device had been observed in recent years. The variability
of the winds and the availability of cheap electrical energy has delayed the
development of efficient, scientifically designed machines.
Csanady (1964) classifies the wind turbine as an extended turbomachine;
the absence of a casing results in the rotor’s effect extending to distant points
in the fluid. It requires no nozzles to accelerate fluid, since the fluid is already
moving. Although it lacks these conventional components, it can be analyzed
as a turbomachine by applying momentum and energy principles to fluid
passing through the rotor.
If fluid passes through the rotor in a direction parallel to the axis of the
rotor, it can be called a horizontal-axis wind turbine. If the direction is nor-
mal to the axis, it is a vertical-axis machine. The horizontal-axis turbine may
be compared with the axial-flow, or propeller, type of hydraulic turbine, but

105
106 WIND TURBINES

Rotor

Figure 11.1 Horizontal-axis wind turbine.

the vertical-axis type differs from the radial-flow turbine in that the flow is
across the rotor rather than radially inward. Figures 11.1 and 11.2 show typi-
cal arrangements for these two types of wind turbines.
In what follows an attempt will be made to discuss briefly the aerody-
namic design of the basic horizontal and vertical types of wind turbines. There
are many considerations which are not discussed in this text, such as speed
control, energy storage, and structural design. Generally, the wind machine
must be designed to operate continuously at constant speed with various
wind and weather conditions and to drive a generator or pump to supply
immediate and future energy needs. Schemes for meeting these and other
design requirements are discussed in books by De Renzo (1979), Bossel (1977),
and Cheremisinoff (1978).

11.2 Actuator Theory

If one imagines the turbine rotor to be replaced by a disk of area A which can
extract energy from the stream, as shown in Figure 11.3, then a useful ana-
lytical result follows through application of the basic equations. The theory
has been developed by Glauert (1959) for propellers and by Betz (1966) for
windmills. As indicated in Figure 11.3, the actuator causes a divergence of
streamlines and a deceleration of the fluid from an upstream speed V to a
ACTUATOR THEORY 107

downstream speed V - v,. At the plane of the actuator, the speed is V - v.

The mass flow rate is the product of density p, velocity at the actuator V - v,
and cross-sectional area A of the stream tube. Because ambient pressure pre-
vails far from the actuator, and the axial momentum difference is simply v, ,
we can express the drag by

D = p(V -v)Av, (11.1)

Since the total pressure upstream pg is given by

P}
Po = Poo+ PE Citi)

and that downstream by

p(V -v,)?
————————— (11.3)
2

we can see that the drop in total pressure across the actuator is simply

pv 2
Po - Po = PW ~~ (11.4)

Figure 11.2 Vertical-axis wind turbine.

108 WIND TURBINES

Vv vay V-y,
—_ SEE phen —_~
|
|
|

Disk of
area A

Figure 11.3 Actuator disk.

Since the static pressure drop at the actuator is equal to the total pressure
drop, we have

Po - Po =P -P (11.5)

Thus the pressure difference on the two sides of the actuator is given by
; V r= vy

The drag force on the actuator can then be expressed as

D=(p-p’)A CLE)

or

D = pAy, (11.8)
V = Vy

Equating (11.1) and (11.8) yields

Viucaee ¥ (11.9)

Substituting (11.9) into (11.8) results in

D = 2pAVv(V - v) (11.10)

If the efficiency 7 of the wind turbine is defined by

n - DV
12 CG ea
ACTUATOR THEORY 109

where T is the torque and 22 is rotational speed, then the efficiency compares
the actual shaft power produced to the “thrust power” available from the
stream. The actual retardation of the stream is less than DV, and it is derived
as the rate of decrease of kinetic energy:

v2 (V-vy, )
KEKES o(V OWA
va(® ; Le
(11.12)

In simplest form (11.12) reduces to D(V - v), which is somewhat less than
DV. Since this energy ideally is transferred to the rotor, we have

TQ = D(V -v) (lilies)

The efficiency expression can then be rendered as

n sy (11.14 )
Vv
7
To determine the efficiency corresponding to maximum power, (11.14)
is used to eliminate v in (11.10). The torque-speed product is replaced by
power P; thus

P=TQ Cit215)

and (10.11) becomes

P = 2pAV3n7(1 - 7) (11.16)

Differentiation of (11.16) yields

dP
= 2pAV3[2n(1 - n) - 7] (11.17)
dn
Setting the derivative equal to zero gives one the optimum value of 7, namely,

or (11.18)

Putting (11.18) into (11.16) gives the maximum wind turbine power as

8 CIIETS)
Pax i 97 p AN?

Equation (10.19) is a well-known result and is frequently used to estimate the

upper limit of power attainable from a wind turbine of area A in a wind of
speed V and density p.
110 WIND TURBINES

Actuator theory provides a useful model for the design of horizontal-

axis machines. It must be extended to include the tangential component of
velocity associated with energy transfer. This is accomplished in the next sec-
tion, where the theory is related to the aerodynamic design of a rotor.

11.3. Horizontal-Axis Machines

Since the rotor imparts angular momentum to the air as it passes between the
blades, a change in whirl velocity AV,, must be indicated by the theory. The
change AV,, is expected to be proportional to the blade speed U, which is the
product of rotational speed Q and radial position r. This may be expressed as

AV,, = 2a'Qr (11.20)

where the coefficient a’ accounts for the difference between the speeds of
blade and fluid.
The coefficient a’ is used by Wilson and Lissaman (1974), who also
define the coefficient a by the expression for wind speed-u at the actuator as

u=(1-a)V Cir2
The actuator theory has shown that a = 1/3 for maximum power, since a is
identical with v/V in (11.18). Wilson and Lissaman (1974) show that the tur-
bine delivers power for 0 <a< 0.5. The theoretical value is helpful in that a =
1/3 can be chosen as a first approximation in design calculations.
Blade

(l-a)V

Figure 11.4 Velocity diagram for a horizontal-axis turbine.

HORIZONTAL-AXIS MACHINES lil

Referring to Figure 11.4, we see that the velocity triangle is easily con-
structed from a knowledge of radial position r, blade rotational speed Q,
wind speed V, and the coefficient a. Normally, the designer would be given
values for wind speed V and rotor speed Q. The solution of Glauert for the
optimum actuator disk given by Wilson and Lissaman (1974) is useful in ob-
taining approximate values of the parameter a at each radial position along
the rotor blade. Glauert’s solution is

(oe (ale
pe clamese | hCaD) (11.22)
from which the dimensionless group rQ/V can be determined for any value of
the parameter a. The relation can be used to make tables or graphs for practi-
cal use. Figure 11.5, based on (11.22), is such a plot. It is clear from Figure
11.5 that a © 1/3 over most of a practical blade. The axial component of the
absolute velocity at the rotor is then approximately two-thirds of the wind
speed.
The Glauert theory also relates the parameter a’, defined in (11.20),
to the parameter a by the relation
lis 3a
,
Cir 223)
4a-1

Calculation of a’ completes the required operations to ensure the construc-

tion of a velocity diagram such as that shown in Figure 11.4 at any radial
position r along the blade length.

.34

52S)

s2

sell|e

Factor
a

.28 $$ —________1__ 41___ ——— =

0 | ee 3 4 S) 6 7

Local Speed Ratio, rQ/V

Figure 11.5 Dependence of parameter a on the local speed ratio.

112 WIND TURBINES

To determine the blade pitch angle @ at any radial position r, we first

calculate the angle ¢ from the relation
6)
oO CEE ae) Senet
Next, we must consult a reference such as Abbott and von Doenhoff (1959)
for the determination of a suitable angle of attack a. Airfoil lift curves (a
plot of Cy, versus a) indicate the stalling value of a. The angle of attack for
design purposes will be less than this value, but near it, to get the largest
possible force from the wind. Selection of this value of a for design also pro-
vides us with lift and drag coefficients at that section. Calculation of the blade
pitch angle can be made easily as well.
Bossel (1977) shows data which indicate that the power coefficient,
defined by
2P
So mPa (11.25)
depends upon blade number and the tip-speed ratio. An attempt to make a
rough correlation of the tip-speed ratio corresponding to maximum C,, as a
function of blade number has been made, and the results are shown in Figure
11.6. The 24-blade American windmill, having a solidity of very nearly unity
and a C, of about 0.15, requires a tip-speed ratio of less than unity. A two-
bladed, high-speed wind turbine, having a C, slightly less than 0.5, on the
other hand, has a tip-speed ratio of 6 or more. The need for simplicity, econo-
my, and efficiency would indicate a choice of the higher tip speeds with
correspondingly fewer blades.
A preliminary design of a rotor blade could be effected using the fol-
lowing approximation for the blade chord C
_ 4nr sin?
nach ese (11.26)

which follows from the Glauert theory presented by Wilson and Lissaman
(1974). Using (11.26), a value of the chord of the airfoil section for the tur-
bine blade can be determined at each radial position in terms of the airfoil
lift coefficient and the velocity diagram angle ¢ applicable at that position r.
The number of blades nz would have to have been previously determined
using the guidance of correlations such as those presented in Figure 11.6.
Lift and drag coefficients are utilized to compute torque for the design
obtained. The tangential component of the blade force per unit length is
given by
 HORIZONTAL-AXIS MACHINES 113

24 '
i)
\
{
x
'
\
\
\
\
‘
\

Number
Blades,
of
MB

Tip Speed Ratio, R&2/V

Figure 11.6 Variation of blade number with tip speed.

pWr72C(Cy, sin ¢ - Cp cos ¢)

ERs 5 Cieza)

Torque on the rotor is computed by integrating F, from the hub radius to

the tip radius. Thus

T=, Jet, Fyr dr (11.28)

which can be evaluated numerically. The turbine power follows from

P=TQ (Cine29))

As pointed out by Wilson and Lissaman (1974), the values of a, a’, ¢,

a, Cy, and Cp must be refined before the final calculation of blade force are
made. Since torque and: thrust can be calculated from momentum or blade-
element theory, internal consistency requires that
a’ feNpRC(Cy, sin @ - Cp cos ¢)
11.30
1+a’ Amr sin 26 ( )
and
a NpC(Cy cos $+ Cp sin ¢)
(1134)
l-a i (87r sin? ¢)

be satisfied. An iterative procedure can be carried out using the initially chosen
values of a and a’ to start the calculation of ¢, a, Cy, Cp, and new values of a
114 WIND TURBINES

and a’ from (11.30) and (11.31). After convergence is obtained the set of
values may be used to calculate blade force and torque.

11.4 Vertical-Axis Machines

As noted by Blackwell (1974), the vertical-axis turbine of the type shown in

Figure 11.2 operates equally well at any wind orientation. The airfoil-shaped
blades follow a circular path, as indicated in Figure 11.7, and this motion re-
sults in a continually changing angle of attack a.
It is necessary that the force on the blade have acomponentin the direc-
tion of its motion to sustain the rotation of the rotor. The velocity triangles
for a blade in the four quadrants are shown in Figure 11.7. A lift vector is
sketched on each diagram perpendicular to the relative velocity vector. The
diagrams clearly indicate that there exists a tangential component of the lift
vector L in all four quadrants. Thus a favorable torque is produced regard-
less of blade position. Wilson and Lissaman (1974) show that the torque per
blade can be estimated by the expression

T =pmCRV?(1 - a)? sin? 6 (11.32)

where a is given by

(l-a)V Ww

Figure 11.7 Velocity diagram for a vertical-axis turbine.

PROBLEMS 115

NgCRAlsin 6
a= ——_______
IRV (11.33)

and R is the maximum radius of the blade with respect to the axis of rotation, as
shown in Figure 11.2. The airfoil chord is denoted by C, and the angular posi-
tion @ is as shown in Figure 11.7. The power is obtained from (11.29).
Blackwell (1974) finds that maximum efficiency is obtained when the
tip-speed ratio RQ/V is around 6. The velocity diagrams of Figure 11.7 indi-
cate a tip-speed ratio of about 4. Blackwell finds that vertical-axis turbines of
this type fail to deliver power for RQ/V <3.
Two other types of vertical-axis machines which have achieved fame are
the Savonius (S-shaped blade) and the Madaras (cylinders), For a discussion
of these and other types the reader is referred to a book by Simmons (1975).

References

Abbott, I.H., and A.E. von Doenhoff. 1959. Theory of Wing Sections. Dover,
New York.
Betz, A. 1966. Introduction to the Theory of Flow Machines. Pergamon,
Oxford.
Blackwell, B.F. 1974. The Vertical-Axis Wind Turbine, “How it Works”. Sandia
Albuquerque Laboratories Report No., SLA-74-0160, Albuquerque,
New Mexico.
Bossel, V. 1977. Energie vom Wind. Deutsche Gesellschaft fur Sonnenenergie,
Munich.
Cheremisinoff, N.P. 1978. Fundamentals of Wind Energy. Ann Arbor Science,
Ann Arbor, Michigan.
Csanady, G.T. 1964. Theory of Turbomachines. McGraw-Hill, New York.
De Renzo, D.J. 1979. Wind Power. Noyes Data Corporation, Park Ridge,
California.
Glauert, H. 1959. The Elements of Aerofoil and Airscrew Theory. Cambridge
Press, Cambridge, Massachusetts.
Simmons, D.M. 1975. Wind Power. Noyes Data Corporation, Park Ridge,
California.
Wilson, R.E., and P.B.S. Lissaman. 1974. Applied Aerodynamics of Wind
Power Machines. Oregon State University. Report No. NSF-RA-N-74-113.

Problems

11.1 The Boeing MOD-2 wind turbine has a blade diameter of 300 ft and
reaches a rated power of 2.5 MW at 27.5 mph. What is its power coefficient
under these conditions?
116 WIND TURBINES

11.2 For a horizontal-axis wind turbine operating at a = 1/3, show that the
entropy increases through the actuator disk is given by

Me ERO)2)
9 Iss)

Sketch the flow process from upstream, through the actuator disk, and into
the wake on a T-S diagram.

11.3. For a vertical-axis wind turbine, momentum conservation requires that

the wake of the turbine be at some angle a to the free-stream velocity. Find
ain terms of the other parameters for a vertical-axis machine.
 Index

Actuator, 106 [Coefficient]

velocity, 89
Blade Compressor, axial-flow, 16
compressor, 67
turbine, 67 Deviation, 83
Diameter
Camber, 84 mean, 62
Casdade, 63, 69 shroud, 39
Cavitation specific, 24
parameter, 103 Diffuser, 49
pump, 38 Dimensional analysis, 23
turbine, 102
Choking, 73 Efficiency
Coefficient blade, 90
blade-loading, 82 compressor, 46, 71
drag, 57 diffuser, 50
flow, 23, 25, 82 hydraulic, 35, 99, 100
Glauert, 111 impeller, 49
head, 24 mechanical, 35, 99
itty) overall, 99, 35
loss, 78, 85 Stage, 9d le 7
power, 24, 112 thermal, 75
slip, 34 turbine, 85

117
118 INDEX

[Efficiency | Performance, 27
volumetric, 34, 99 Polytropic, 16
wind-turbine, 108 Power
Energy balance, 11 brake, 38
Energy transfer, 12, 14 turbine, 109
Enthalpy turbomachine, 13
static, 11 Pressure rise
total, 12 compressor, 47, 71
Euler, 14 pump, 33,57
Pump
Fan axial-flow, 61
axial, 65 centrifugal, 17
centrifugal, 41 laws, 26, 37
laws, 37 similitude, 26
Force, blade, 56, 113
Ratio
Head hub-tip, 62 .
cavitation, 38 pitch-chord, 57
ideal, 18, 32 tip-speed, 114
total, 35 Reaction, 15, 82, 93
Reynolds number, 24
Impeller
compressor, 47 Slip, 34
pump, 38 Solidity, 58, 62, 69, 84
Incidence, 83 Speed
acoustic, 51
Loss specific, 24, 28, 61
boundary-layer, 59 Stage
leakage, 60 Curtis, 92
pump, 34 number, 81
secondary-flow, 60 Surge, 51

Mach number, 24 Temperature; 51

Mass flow Torque
chocking, 51 pump, 39
inlet, 10 turbine, 91, 114
Matchine, 76 turbomachine, 13
Momentum, 12 Turbine
gas, 14, 83
Nozzle, 91, 14 hydraulic, 19, 97
INDEX 119

[Turbine] Volute, 40
steam, 87 Vortex, 64
wind, 105
Work, 75
Velocity, 9 Work-done, 72
Volume flow, 17
 ]
oo0728—ee05

“IQ ier
nd
,
about the textbook... ae aes x

Turbomachinery: Basic Theory and Applications is a concise, easy-to-read, under- ©

graduate textbook on the basic principles of turbomachine theory and the appli-
cation of these principles to specific devices. Undergraduate-level mechanical en-
gineering students who have had introductory courses in fluid mechanics an
thermodynamics will be quite successful in their use of this book.

Based on the courses Professor Logan has taught, the material in this textbook is d
arranged sequentially in order to provide a clear, unified exposition of the topic, = 9
Beginning with the first principles of physics—conservation of mass, momentum, ~
and energy—the text proceeds to centrifugal pumps. More complex machines are
then considered, concluding with the basic types of hydraulic turbines. Each
chapter addresses the questions of how the ee of turbomachine theory _—
may be applied in design and how they may be used to predict the performance
of the turbomachine under consideration. littsarathve to accompany the
text and a solutions manual is available. :
i
i
4

The importance of turbomachines in our lives cannot be overemphasized and, as

such, good mechanical engineering education depends upon a student’s ability tod: #
understand the principles and to be able to make use of these principles in design. ’
Turbomachinery: Basic Theory and Applications is the best textbook available a
for instructing students in this important area. ¥
i al ) ' =
about the author... ae E

EARL LOGAN, JR. is a member of the faculty of the College of Engineering and
Applied Science at Arizona State University, Tempe, Arizona. He received the ;
B.S. degree (1949) and M.S. degree (1958) from Texas A&M University and the :
Ph.D. degree from Purdue University (1961). Dr. Logan’s research focuses on
turbulent boundary layers and he has published various technical papers on fluid |
mechanics, heat transfer, and flow measurement. He is a member of the Ameri- ,
can Society of Mechanical Engineers, American Society for Engineering Educa-
tion, and the American Physical Society. |

Printed in the United States of America Pema _—_ISBN:0—8247—1509—8 m q

marcel dekker, INC. iNew yode—lrereel —___ aver

ys