PRINCIPLES OF OPERATION AND PERFORMANCE

ESTIMATION OF CENTRIFUGAL COMPRESSORS

by
Dr. Meherwan P. Boyce
Chairman and CEO

Boyce Engineering International, Inc.

Houston, Texas

performance of a compressor and in its selection. Typical perfor

mance characteristics are shown in Figure 1 of different types of
compressors.

Meherwan P. Boyce is Chairman and

CEO of Boyce Engineering International,
Inc., in Houston, Texas. His past experience
incorporates many academic and profes
sional positions, including Professor of Me
chanical Engineering, Founder and first
Director of the Turbomachinery Laborato
ry. He was also responsible for founding the
Turbomachinery Symposium, which he
chaired for eight years.
His industrial positions include Manager
of Compressor and Turbine development at Curtiss Wright and
Manager of Aerodynamics Technology at Fairchild Hiller Corpora
tion.
Dr. Boyce has written more than J 00 significant publications and
technical reports and is the author of the Gas Turbine Engineering
Handbook and has contributed to oth(tt major handbooks. He has
been elected to membership in several honor societies such as Phi
Kappa Phi, Pi Tau Sigma, Sigma Xi, and Tau Beta Pi.
He is also a member of several professional socities such as
ASME, SAE, NSPE, HESS and ASEE. In 1985, Dr. Boyce was named
an ASME Fellow. Dr. Boyce was the 1974 recipient of the ASME
Herbert Allen Award for Excellence and the 1973 recipient of the
Ralph R Teetor Award of SAE.
Dr. Boyce pioneered a breakthrough in technology through the
development of a real time computer system which monitors, analyz
es, diagnoses, and prognisticates performance of major turboma
chinery. These systems are in use throughout the world.
Dr. Boyce received his Ph.D. in Mechanical Engineeringfrom the
University of Oklahoma.

30

I
If

10

1Q'

CAPACITY, SCFW

PO
C
A

EF'nctENCV

SPEED

1Q'

PDSITYE DISPLACEMENT
CENTRIFUGAL
AXIAL
PRESSURE

100 X
.OF' RATED F'LOW

Figure 1 . Performance Characteristics of Various Types of

Compressors.
Positive displacement compressors are generally suitable for
small flowrates while centrifugal and axial compressors are more
commonly applied for medium and large flow applications respec
tively . The advantages of centrifugal compressors are that they are
reliable, compact and robust, have better resistance to foreign
obj ect damage, and are less affected by performance degradation
due to fouling . Above all, as can be seen in Figure 1, they have a
wider operating domain when compared to other compressor
types.
Centrifugal compressors are most commonly applied in petro
chemical or process industries in the fl o wrates ranging from 1 000
to 1 00,000 ft3/min. Typical centrifugal compressor applications
are summarized in Table 1.

INTRODUCTION

A compressor is a fluid handling mechanical device capable of

efficiently transferring energy to the fluid medium so that it can be
delivered in large quantities at elevated pressure conditions. Com
pressors have numerous applications ranging from aircraft and
process industries to household appliances such as refrigerators
and air conditioners. There are numerous types of compressors,
each suitable for a particular application. Generally, compressors
can be categorized under two basic types-positive displacement
and dynamic . Positive displacement compressors include piston,
screw, vane, and lobe compressors. Axial and centrifugal com
pressor types are dynamic compressors as the required pressure
rise and flow is imparted to the fluid medium by transferring
kinetic energy to the process gas.
Flowrate, efficiency, and the pressure rise within the compres
sor are the three most important parameters used in defining the

BASIC COMPONENTS AND PRINCIPLES OF

OPERATION

A centrifugal compressor (Figure 2) consists of three basic

parts: ( 1) rotor assembly (impeller), (2) diffuser, and (3) scroll.
Inlet guide vanes direct the flow to the inducer at the right rotor
161

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

1 62

Table 1 . Application of Centrifugal Compressors

Industry or Application

Service of Process

Typical Gas Handled

Iron and Steel

Blast furnace

Combustion

Bessemer converter

Air

Off Gas

Blast furnace gas

Oxidation

Air

Cupola

Combustion

Air

Coke oven

Compressing

Coke oven gas

exhausting

Coke oven gas

For tools and machinery

Air

Copper and nickel

Air

Mining and Metallurgy

Power
Furnaces

diflu,.C

0
1
tl
e
u
i-1-fl!t----+--1------ - - ----.,-., -,-r---l

:;j

purification
Pelletizing (Iron Ore

Air

concentration)

Figure 3. Variation of Pressure and Velocity through the

Compressor.

Natural Gas
Production

Repressuring oil wells

Natural gas

Distribution

Transmission

Natural gas

Processing

Natural gasoline

Natural gas

separation
Refrigeration

Propane and Methane

Refrigeration
Various Processes

Chemical

Butane, propane,
ethylene, ammonia,
special refrigerants

Industrial and commercial

Air conditioning

Special refrigerants

inlet angle. The rotor increases kinetic energy of the medium as

well as its static and total pressure (Figure 3 ) .
Total conditions only vary when energy is put into the system,
thus only in the impeller. The gas leaves the rotor and enters the
diffuser at an angle determined by the rotor exit angle and the
rotational speed of the rotor. The primary purpose of the diffuser
RETURN CHANNEL
CASING

Ilot PELLER
SEAL

is to reduce the velocity of the gas and to efficiently increase the

static pressure. In the case of the centrifugal compressor, since the
flow direction is in the radially outward direction, the increase in
flow area in the downstream direction naturally diffuses the flow.
The compressed gas from the diffuser discharges into the rotor
downstream through return bend and return channel. The last
diffuser discharges the gas to a scroll or volute. The cross-section
of the scroll varies circumferentially. Since all the rotating blade
passages discharge the compressed gas into the scroll, the cross
section of the scroll increases in the direction of rotation of the
rotor. The increase in scroll cross-section is also designed to give
additional pressure rise. Since the gaseous medium passes though
a labyrinth of rotating and stationary flow passages, the entire gas
path has to be aerodynamically designed to compress the gas
efficiently and minimize losses.
DIMENSIONAL ANALYSIS OF A CENTRIFUGAL
COMPRESSOR

Turbomachines can be compared with each other by the use of

dimensional analysis. This type of analysis produces various types
of geometrically similar parameters. Dimensional analysis is a
procedure where variables representing a physical situation are
reduced into groups that are dimensionless. These dimensionless
groups can then be used to compare performance of various types
of machines with each other. Dimensional analysis as used in
turbomachines can be used for the following: ( 1 ) to compare data
from various types of machines, (2) a very useful technique in the
development of blade passages, blade profiles, and diffusers for
selection of various types of units based on maximum efficiency
and pressure head required, and 3) prediction of a prototype units
performance from test conducted on a smaller scale model or at
lower speeds and flows. S ome of the important nondimensional
parameters are :
Reynolds number
Specific Speed

Figure 2. Schematic of Centrifugal Compressor.

Re7,

=
=

pDV
ll

--

Nf Q
H%

DH3/4

(1)
(2)

Specific Diameter

Flow Coefficient

<P = _!!_

(4)

Head Coefficient

'P=_!!_
2 2

(5)

fQ

(3)

ND3
ND

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

where D, H, V, N are, respectively, diameter of the impeller, head,
velocity, and speed of the machine. The aforementioned are some

of the major dimensionless parameters. In many cases, for the flow

to remain dynamically similar, all the above parameters must

remain constant; however, this is not possible in a practical sense,

1 63

Basic Equations
Euler Turbine Equation
The Euler equation is a modification of the mathematical formu

lation of the law of conservation of momentum. It states that the

so one must make choices.

In selecting turbomachines, the choice of specific speed and

rate of change in linear momentum of a volume moving with the

suitable as seen in Figure 4

fluid. The velocity components in a generalized turbomachine are

specific diameter determines the type of compressor whichis most

fluid is equal to the surface forces and body forces acting on the

shown in Figure 6. The velocity vectors as shown are resolved into

three mutually perpendicular components: the axial component

'It aua dHclac'

relaa<l

(V), the tangential component(V) and the radial component(V,).

ca '1 aallaaa&

By examining each of these velocities, the following character

istics can be noted: the change in the magnitude of the axial

velocity gives rise to an axial force that is countered by a thrust

bearing and the change in radial velocity gives rise to a radial force

that is countered by the journal bearing. The tangential component

is the only component that causes a force which corresponds to a

change in angular momentum; the other two velocity components

have no effect on this force-except for what bearing friction may

arise. By applying the conservation of momentum principle, the

change in angular momentum obtained by the change in the

tangential velocity is equal to the summation of all the forces

applied on the rotor. This summation is the net torque of the rotor.
..

laacUlc apaa<l

c)

...

Figure 4. A Generalized Compressor Map.

It is obvious from this figure that high head and low flow require

A certain mass of fluid enters the turbomachine with an initial

velocity(V1) at a radius r1, and leaves with a tangential velocity

(V2) at a radius r2 Assuming that the mass flowrate through the

turbomachine remains unchanged,the torque exerted by the change
in angular velocity can be written as:

a positive displacement type unit, while a medium head and

m
T= -(r1V 1-r2V 2)
6
6
gc

medium fo
l w require a centrifugal type unit, and for high flow and

low head an axial flow type compressor is the best choice.

A detailed view is shown in Figure 5 of the centrifugal compres

sor section of Figure 4 and can be used as a reference for selection

The rate of change of energy transfer ft-lb/sec is the product of the

torque and the angular velocity

of centrifugal compressor units. Flow coefficients and pressure

m
Tro= -(r1V 1-r2roV 2)
6
6
gc

coefficients can be used to determine various offdesign character

istics. The Reynolds number affects the flow calculations as far as

(6)

(7)

the skin friction and velocity distribution are concerned. It must be

Thus,the total energy transfer is given by the following relationship:

formance or predicting performance based on tests performed on

(8)

remembered when using dimensional analysis in computing per

m
E= -(U1 Ve1-U2Va2)
gc

smaller scale units that, due to the fact that it is not physically

possible to keep all parameters constant, the variation of the final

results will depend on the scale up factor, the difference in the fluid

medium, the speed of the unit, and the pressure delivery. It is

therefore very important to understand the limit of the parameters

and thus the geometrical scale up while conducting a dimensional

where U1 and U2 are the linear velocity o f the rotor a t the respective
radii and(V1) and
are the tangential projections of the absolute

(V)

velocity. The previous relation for unit mass flow can be reduced

to the following:

study.

(9)
where H is the energy transfer per unit mass flow ft-lb/lbm,

commonly known as the head, is delivered. Understanding the

basics of this equation helps in visualizing practical turbomachin

ery operation.

As simple as it seems, this equation is the basis of nearly all

performance characteristics in turbomachines.

Based on the energy equation and using measured variables, the

specific work and head can be evaluated as:

C T1
P
H=h 2 -h1 "'C(T
T )=p 2- 1
T)

=--'
-:! =-----:--'-::!=---::300
=-o ---,e=-=oo':co:-.l
o.,o=------,so:----eo
COMPRESSOR

SPECIFIC

Figure 5. A Typical Centrifugal Compressor Performance Chart.

[(

P2
p1

)y

-1

( 1 0)

where h2 and h1 are inlet and exit gas enthalpies that are direct
functions of compressor inlet and exit temperatures and pressures.
For a thermally and callorically perfect gas, specific enthalpy is a

164

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

duce a higher pressure rise per stage than an axial compressor.

Also, the crosssectional area variation of the flow passage (be

tween the vanes) and the rotor inlet and exit flow angles are critical
to determine the work distribution among the rotor, diffuser, and

scroll in compressing the gas.

COMPRESSOR PERFORMANCE
Compressor performance curves consists primarily of a plot

showing the variation of compressor head or pressure ratio and

efficiency at various constant rpm conditions at different mass
flowrates such as the one shown in Figure 7. It can be seen from
Figure 7 that an increase in rotor speed increases the compressor

flowrate. At a particular rotor speed an increase in compressor

pressure ratio can be obtained by reducing the compressor mass

flowrate. This is to be expected because an increase in pressure at

compressor delivery can only be expected if there is a

resistance

to flow, otherwise, the compressor would behave more like a fan

or a blower. In a similar fashion, an increase in flowrate at constant

rpm can be obtained by a reduction in the resistance at the

Figure 6. Velocity Vectors in Compressor Rotor Flow.

compressor exit. An enhancement of compressor flowrate at

product of specific heat at constant pressure(C ) and temperature,

P

as shown in Equation 10.

The only difference between the above equations is that while

Equation (9) is based on mechanical principles, Equation(10) is

con
stant delivery pressure is also possible. This requirement, howev

er,can be met by not only changing exit throttle,but also increasing

the compressor rpm. Minimum and maximum permissible flow

rates at constant rpm are termed surge and choke limits. A line

based on thermal quantities and can be explained clearly in terms

tracing the stall points of all the constant rpm lines is called a

changeable. These two equations can be equated using the defini

various rpm is termed operating line. The compressor rotor should

turbomachine, Equation(10) is useful to quantify the efficiency of

the surge line, as is indicated in Figure 7, that is, when the

of the first law of thermodynamics that heat and work are inter

tion of efficiency. While Equation

(9) is used to design a

the design. Equations (9) and(10) can be combined together to

relate the running conditions and the compressor impeller geom

etry to the pressure rise within the compressor as:

"if=
I

1+

--

(UzVez -UIVei)
Cl2gc

]
r

surge
line. Similarly, locus of operating points of the compressor at

be designed such that at constant rpm, its efficiency peaks close to

compressor pressure ratio is maximum.

Compressor Efficiencies
The work in a compressor under ideal conditions occurs at

(11)

The compressor performance is thus clearly interrelated to the

compressor geometry.

constant entropy as shown in Figure 8 . The actual work done is

indicated by the dotted line. The isentropic efficiency of the

compressor can be written in terms of the total changes in enthalpy.

'1

de

Isentropic Work
Actual

Energy Transformation in Turbomachines

The compressor rotor transfers its kinetic energy to the gas

within the impeller. However, not all the energy of the gas gets

(h2, -hit)id
=
(h2., -hlt)act

(13)

This equation can be rewritten for a thermally and calorifically

perfect gas in terms of total pressure and temperature as follows:

converted into pressure(potential) energy within the impeller. As

explained in the introduction, the diffuser and volute or scroll aid

in transferring energy of high velocity gases into pressure. The

(14)

distribution of energy transfer within the compressor is dictated by

its design. Using basic trigonometry Equation(11) can be rewrit

ten as:

The process between 1 and 2' can be defined by the following

equation of state:

2
The first term in the above equation, 1/2gc(V /-V1 ), termed the
external effect represents the kinetic energy transferred to the gas

from the rotor. This term represents the virtual pressure rise
possible external to the rotor but within the compressor using the
diffuser and scroll. The second term,1/2gc(U / - U /),describes the
pressure rise possible within the rotor due to the

centrifugal effect

which is only possible in centrifugal compressors or radial in flow

turbines where the rotor entrance and exit are at different radii. The
2
third term, 1/2g(
c W /- W1 ), represents the internal diffusion effect
describing the pressure rise possible within the rotor, by reducing

the relative velocity within the rotor. Clearly, since a centri fugal

compressor takes advantage of the

centrifugal effect, it can pro-

Centrifugal compressor

Figure 7. Compressor Performance Curves.

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

1 65

increases approaches zero, and the value of the polytropic efficien

cy is higher than the corresponding adiabatic efficiency. The

Pz

relationship between adiabatic and polytropic efficiency is shown

in Figure 9, as the pressure ratio across the compressor increases.

Another characteristic of polytropic efficiency is that the polytrop

II

.c

c
w

..

ic efficiency of a multistage unit is equal to the stage efficiency, if

each stage has the same efficiency.

I
;

POLYTROPIC EfFICIENCY
15

10

8
s
y

liJ

Entropy (S)

85

15

75

85

80

8D

8D

70

75

70

Figure 8. Temperature Entropy Diagram of a Compressor.

p

=Canst

1.0

4.0

2.0

(15)

Where T] i s some polytropic process. The adiabatic efficiency can

then be represented by:

10.0

1.0

PRESSURE RAnD

Figure 9. Relationship between Adiabatic and Polytropic Efficien

cy Compressor.
Factors Affecting Compressor Performance
A number of factors can affect the compressor performance and

( 1 6)

these are described brie fly in the following paragraphs.

Gas Composition
Polytropic Efficiency

The performance characteristics of a compressor are not a

Polytropic efficiency is another concept of efficiency often used

in compressor evaluation. It is o ften referred to as small stage or

function of compressor geometry alone, but also depend on the

properties of the gas it is compressing.

infinitesimal stage efficiency. It is the true aerodynamic efficiency

;,
:w,
y 1

exclusive of the pressure-ratio effect. The efficiency is the same as

Utilizing real gas effects( PV= ZRT= A

efficiency:

compressibility factor, and C =

P

if the fluid is incompressible and identical with the hydraulic

T) where, Z is the

specific work or

adiabatic head required to compress the gaSiSgiven by:

( 1 7)

W= Z

T1

MW

Y_(

y- 1

y-1

-:y - 1 )
Pl

(20)

The specific work or energy required to compress the gas is a

which can be expanded assuming that

function of compressibility of the gaseous mixture, the equivalent

mixture molecular weight, and the inlet and exit pressure condi

( 1 8)

tions, as indicated in Equation(20). Hence, if the molecular weight

of the gas increases, for the same specific work, the pressure ratio
of the compressor also increases. Also, the mass flowrate increases

Neglecting second order terms, the following relationship is

obtained:

proportional to the molecular weight of the gas, volume flowrate

on the other hand remains constant.

The performance is shown in Figure 10 of an individual stage at

y- 1
-y

Tlpc= --n- 1
-n

a given speed for three level of gas molecular weight. The heavy
gas class includes gases such as propane,propylene and standard

( 1 9)

From this relationship, it is obvious that polytropic efficiency is

the limiting value of the isentropic efficiency as the pressure

ized refrigerant mixtures. Air, natural gases and nitrogen are

typical of the medium class gases. Hydrogen rich gases found in

hydrocarbon processing plants are representative of the light class

gases.

The following observations can be made with respect to the

curve for heavy gas as compared to lighter gas.

1 66

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

direction of rotation
POLYTROPIC
EFfiCIENCY

"

v., - v,

vm

-..

hout iGV(nonprewhirl)

pre

lrl

1 0. Effect of Performance

in ucer
vane

'

of Various

\t

Gases.

The flow at surge is higher.

The stage produces more head than that corresponding to

medium or lighter gas.

The right hand side of the curve near choke turns downward

The curve is flatter towards the surge point.

(approaches stonewall) more rapidly.

v .,

direction of rotation

Figure

v,

positive prewhirt

'

direction of rotation

---w-

11.
inducer

'

vane

dr

v,

negative prewhirt

Figure 1 1 . Inducer Inlet Velocity Diagram.

The main purpose, in most cases, of installing an IGV is to

decrease the relative Mach number at the inducer tip inlet because

the highest relative velocity occurs at the inducer inlet tip section.

When the relative velocity is close to the sonic velocity or greater

Inlet Conditions
Specific head developed by the compressor given in Equation

than it, a shock wave takes place at the inducer section, this gives
the shock loss and also chokes the inducer. Relative mach numbers

( 1 2) is the amount of energy transferred by the rotor to the gas and,

hence, is a function of rotor speed and geometry only and is not

rarely used since it also decreases the operating range. The effect

can get an idea of variation in pressure ratio and power with

curved blades is shown in Figure 12. The effect of inlet prewhirl

affected by inlet conditions. Using Equation ( 1 0) , however, one

ambient temperature variation.

above 0 . 7 increase the losses dramatically. Negative pre swirl is

of various guide vane settings on the ideal head of backward

is shown in Figure 1 3 with respect to a compressor efficiency for

high pressure radial bladed impellers.

Inlet Guide Vanes

The inlet guide vanes give a circumferential velocity component

to the fluid at the inducer inlet. This is called prewhirl. The inducer
inlet velocity diagrams are shown in Figure 1 1 with and without
inlet guide vanes( IGV), which may be installed directly in front

the inducer, or where an axial entry is not possible, located radially

IDEAL
HEAD

in an intake duct. A positive vane angle produces prewhirl in the

direction of the impeller rotation and a negative vane angle

produces prewhirl in the opposite direction. The disadvantage of

positive prewhirl is that a positive inlet whirl velocity reduces the

energy transfer. The Euler equation is defined by :

H=

gc

cu!ve1- uzvez)

(2 1 )

For non-prewhirl(without IGV) V.1 is equal t o zero. The Euler

work is thus equal to H = U2 V.12. Euler work is reduced by the

F'OR lloiPEll.ER WJTH BACKWARD LEANING BLADES.

Fl..OW

amount U1 V 11. in the case of positive prewhirl. Negative prewhirl

increases the energy transfer by the amount U1 V

1 . The positive

prewhirl, on the other hand, decreases the relative mach number at

the inducer inlet. This decrease in the relative Mach number is

important for units operating at or near Mach numbers of 0 . 7 .

Negative prewhirl o n the other hand increases the relative Mach

number. A relative mach number is defined by:

w '
M rei=
a!

(22)

where: M,.1is the relative mach number, W,1 is the relative velocity

at an inducer inlet and

1 is the sonic velocity at an inducer inlet

section based on the static temperature at that point.
oc

Figure

1 2. Swirl Effects on a Backward Curve Blade.

There are three kinds of prewhirl. These are as follows:

Free VortexType-this type is represented by r1 V1=constant

with respect to the inducer inlet radius. This prewhirl distribution

is shown in Figure 1 3 . In this type, V1 is a minimum at the inducer

inlet shroud radius. Therefore, it is not good to decrease the
relative Mach number in this manner.

Forced Vortex Type-This type is shown as V/r1=constant,

with respect to the inducer inlet radius. Prewhirl distribution with

this type of inducer is shown in Figure 1 4 . In this type, V1 is a

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

0.85

1 67

section, the blades bend toward the direction of rotation as shown

in Figure 1 6 . The inducer is an axial rotor and changes the flow

direction from the inlet flow angle to the axial direction. It has the

largest relative velocity in the impeller and, if not properly de

signed, can lead to choking conditions at its throat, as shown in

0.80

Figure 1 6 .

directi

otation

0.75

0.70 ._______.______._____.______,
0
10
20
30
40

.,.whirl angle, degrH

Figure 1 3. Effect of Prewhirl on Efficiency of High Pressure

Compressor.

leading edge

angle

trailing edge

fluid

-
----k-:;a, ---ll,.-..-,..-- u 1 !i

inducer leading

shroud radius

Inducer Inlet

Inducer Inlet
hub radlua

throat

Figure I6. Inducer Centrifugal Compressor.

There are three forms of inducer camber lines in the axial

Figure I4. Prewhirl Distribution Patterns.

direction. These are circular arc, parabolic arc, and elliptical arc.

maximum at the inducer inlet shroud radius, so this contributes to

decreasing the inlet relative Mach number.

Control Vortex Type This type is represented by V 1= Ar1

B/r1 where A and B are constants. This equation shows the first
type with A= 0, B 0 and the second type with B= 0, A 0 .

Euler work distributions a t a n impeller exit with respect t o the

impeller width are shown in Figure 1 5 . From Figure 1 5 , it is noted

that the prewhirl distribution should be made not only from the

relative Mach number at the inducer inlet shroud radius, but also
from Euler work distribution at the impeller exit. This is because

Circular arc camber lines are used in compressors with low

pressure ratios,while the elliptical arc produces good performance

at high pressure ratios where the flow has transonic mach numbers.

Because of choking conditions in the inducer, many compres

sors incorporate a splitter-blade design The flow pattern in such an

inducer section is shown in Figure

1 7 (a). This flow pattern

indicates a separation on the suction. Other designs include tan

dem inducers. In tandem inducers the inducer section is slightly

rotated as shown in Figure 17(b). This modification gives addi

tional kinetic energy to the boundary layer, which is otherwise
likely to separate.

uniform impeller exit flow conditions,considering both the impel

ler and diffuser losses, are important factors to obtain good

compressor performance.

separation region

shroud

nonprewhirl

hub

----- - ,--------- ------lorced = vortex prewhirl

..
oil

lree., vortex prewhlrl

hub

note: positive prewhlrl

Impeller exit width

shroyd

Figure I5. Euler Work Distribution at Impeller Exit.

Inducer
The function of an inducer is to increase the fluid's angular
momentum without increasing its radius of rotation. In an inducer

n
without tandem inducer

(b)

with tandem inducer

Figure I7. Impeller Channel Flow with a Splitter and Tandem

Inducer.

PROCEEDINGS O F THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

168

Effect of Blade Shape On Performance

The blade shapes of centrifugal compressors can be classified

based on rotor exit blade shape as backward, radial and forward

curved blades, as shown in Figure 1 8 . In order to study the effect

of blade shape on performance one resorts to velocity triangles.

The axial component of velocity represents the velocity vector

proportional to the volume flowrate. To begin with, the flowrate

andthe head produced for the three blade shapes are assumed to be

the same. By changing the volume flowrate for each of the three

blade shapes, that is, by increasing the axial velocity vector for the
three machines by the same amount one can determine the head

produced from each of the blade shapes considered using the Euler

Turbine equation. This exercise results in a plot as shown in Figure

1 9 . Solid lines indicate ideal characteristics and dotted lines real

I
I

zlb

zlb

Zlb-

====-_ twM=rt=Fi
l -:t:h0
. .

characteristics which take losses into account. The figure clearly

illustrates that in theory as the flowrate increases the head or

y'ft'

_....

y'!t'

y':t -

y' It'

zlb= relative meridional channel width

Zib

-.
-

.... -

y'lt'

......

relative blade spacng

y'lt'

pressure ratio of the compressor drops for backward-curved blades,

remains almost the same for radial exit compressors and increases

for forward-curved blades. Most impellers in the petrochemical

field are radial or backward curved.

.. . ...
l'lldll-

Figure 1 8. Velocity Triangles of Various Types of Impeller

Blading.
Figure 20. Velocity Profiles Through a Centrifugal Impeller.

forward curved blades at the rotor exit. The relative velocity of the

gas leaving the rotor does not exactly follow the exit rotor angle.
This phenomenon is due to circulation or slip which occurs in

various blade compartments. The effect of slip on the exit flow

angle is illustrated in Figure 2 1 .

Fr:
F;

=
=

centrifugal forces
Inertia lorces

fLDW

Figure 1 9. Head and Flowrate Characteristics of Various Impel

ler Blading.
Slip
The high velocity exiting from the forward bladed impeller

increases the losses in the diffuser section decreasing the efficien

cy in such a compressor to unacceptable levels. Blades should be

designed to eliminate large decelerations or accelerations of flow

in the impeller that lead to high losses and separation of the flow.

w = 0

Figure 2 1 . Forces and Flow Characteristics in a Centrifugal

Impeller.
To account for the flowdeviation(which is similar to the effect
accounted for by the deviation angle in axial flow machines), a

The velocity distributions measured in a typical centrifugal com

factor known as the slip factor is used, defined as,

types of rotors were introduced. These are radial, backward, and

velocity.

pressor are shown in Figure 20.

I n the previous section, the performance characteristics o f three

= V azac

JV82th,

where V ezact is the tangential component of the absolute exit

velocity andV ezthis the theoretical component of the absolute exit

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

To combine all these effects, consider a backward curved blade

Causes of Slip
The cause of the slip factor phenomenon that occurs within an

impeller is not known exactly. However,some general reasons can

be used to explain why the flow is changed, these are:

169

Coriolis Circulation-The pressure gradient between the

impeller. The exit velocity triangle for this impeller, with the

different slip phenomenon changes is shown in Figure 24. As this

show the actual conditions when the compressor is running may be

far removed from the design condition.

walls of two adjacent blades, Coriolis force and centrifugal force

causes circulation of the fluid within the rotor flow passages as

shown in Figure 2 2 . Because of this circulation,a velocity gradient

results at the impeller exit with a net change in the exit angle.

Boundary layer Development-The boundary layer that de

velops within an impeller passage causes the flowing fluid to

experience a smaller exit area as hown in Figure 2 3 . This is due

to small, if any, flow within the boundary-layer. For the fluid to

exit this smaller area, its velocity must increase. This gives a

higher relative exit velocity. Since the meridional velocity remains

constant, the increase in relative velocity must be accompanied

with a decrease in absolute velocity.

Leakage - Fluid flow from one side of a blade to the other side

is referred to as leakage. Leakage reduces the energy transfer from

impeller to fluid and decreases the exit velocity angle.

Number of Vanes-The higher the number of vanes, the lower

13"

13

13

is caused by Coriolis circulation

is caused by boundary-layer effects

is caused by the blade thickness

Figure 24. Effect on Exit Velocity Triangle by Various Parameters

Which Cause Slip.

the vane loading and the closer the fluid follows the vanes. With

higher vane loadings, the flow tends to group upon the pressure

surfaces and introduces a velocity gradient at the exit.

Vane Thickness-Due to manufacturing problems and phys

ical necessity, impeller vanes have some thickness. When the fluid
exits the impeller, the vanes no longer contain the flow and the

meridional velocity is immediately decreased, both the relative

and absolute velocities decrease, thus changing the exit angle of

the fluid.

Several empirical equations have been derived for the slip

factor. Some of the more widely used methods equations are

derived by Stodola and Stanitz.

Slip Factor Due to Stodola

The second Helmholtz law states that the vorticity of a friction

less fluid does not change with time. Hence, if the flow at the inlet

to an impeller is irrotational, the absolute flow must remain

irrotational throughout the impeller. As the impeller has an angular

velocity, the fluid must have an angular velocity, relative to the

impeller. This fluid motion is called the relative eddy. Thus, if

there were no flow through the impeller, the fluid in the impeller
channels would rotate with an angular velocity equal and opposite

to the impeller's angular velocity.

To approximate the flow, Stodola's theory assumes that the slip

is due to the relative eddy. The relative eddy is considered as a

rotation of a cy Iinder of fluid at the end of the blade passage(shown

as a shaded circle) at an angular velocity of about its own axis. The

Stodola slip factor is given by,

I..L

1- (1z

Coriolis circulation.

sin2

Wm2 Cot2
u2

(23)

Figure 22. Coriolis Circulation in a Centrifugal Impeller.

where 82 is the blade angle, Z the number of blades, Wm2 is the
meridional velocity and U2 the blade tip speed. Calculations using

this equation have been found to be generally lower than experi

mental values.

Stantiz Slip Factor

Stanitz calculated blade-to-blade solutions for eight impellers

and concluded that, for the range of conditions covered by the

solutions, U is a function of the number of blades (Z) and the blade

exit angle (82) is approximately the same whether the flow is

compressible or incompressible.

I..L

Figure 23. Boundary-Layer Development in a Centrifugallmpeller.

1- 0.631t (1Z

Wm2
u2

(24)

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

170

Stanitz' s solutions were for rt/4< <n/2 . This equation compares

2
well with experimental results for radial or near radial blades.

Diffusers
Most centrifugal compressors in service in petroleum or petro
chemical processing plants use vaneless diffusers. A vaneless
diffuser is generally a simple flow channel with parallel walls and
does not have any elements inside to guide the flow. The velocity
diagrams shown in Figure 25 at the eye and exit of an impeller
illustrate the trajectory a particle in the gas flow would take
through a vaneless diffuser at the design condition (compressor
rated point) .

=TIP SPEED
=ABSOLUTE VELOCITY OF GAS
=GAS VELOCITY RELATIVE TO BLADE
=TANGENTIAL COMPONENT OF'
ABSOWTE VELOCITY
=RADIAL COMPONENT OF
AVSOLUTE VELOCITY

U
VA
Vt
YT
V MD

...
v"
...
B
.
.....
...
ao

SURGE CQNQII!ON
INCREASED

DECREASED

"'
"'

UNCHANGED

...
Yn
..

Vaned diffusers are used to force the flow to take a shorter, more
efficient path through the diffuser. There are many styles of vaned
diffusers, with major differences in the types of vanes, vane angles
and contouring and vane spacing. Commonly used vane diffusers
employ wedge-shaped vanes (vane islands) or thin curved vanes.
The latter type of vane is illustrated in Figure 27. In high head
stages, there can be two to four stages of diffusion. These usually
are vane less spaces to decelerate the flow, followed by two or three
levels of vaned blades in order to prevent buildup of boundary
layer, thus causing separation and surging the unit. The flow
pattern in a vaned diffuser is indicated in Figure 27. The vaned
diffuser can increase the efficiency of a stage by two to four
percentage points, but the price for the efficiency gain is generally
a narrower operating span on the head-flow curve with respect to
both surge and stonewall. The effect of off-design flows is also
shown in Figure 27. Excessive positive incidence at the leading
edge of the diffuser vane occurs when
is too small at reduced
2
flow and this condition brings on stall. Conversely, as flow
increases beyond the rated point, excessive negative incidence can
cause stonewall. Despite its narrowing effect on the usable oper
ating range on the characteristic curve, the vaned diffuser has its
application in situations where efficiency is of utmost importance.
ex

....

ROTATION

---BASIC HfAQ EQUATION

H ra.v =
N

H "

Yn U2-Yr1 Ut

= POLYTROPIC EFfiCIENCY

Figure 25. Surge Effect of Exit Velocity Diagrams.

When the inlet flow to the impeller is reduced while the speed
is held constant, there is a decrease in V and
As
decreases,
,2
2
2
the length of the flow path spiral increases. The effect is shown in
Figure 26. If the flow path is extended enough, the flow momentum
at the diffuser walls is excessively dissipated by friction and stall .
With this greater loss, the diffuser becomes less efficient and
converts a proportionately smaller part of the velocity head to
pressure. As this condition progresses, the stage will eventually
stall. This could lead to surge.
oc:

a21S S UR GE-TENDEN CY

oc:

a2

VANELESS SPACE

at2

TOWARD STONEWALL

IS SMALL-TENDENCY TOWARD
SURGE-POSITIVE INCIDENCE ON
VANE

AT DESIGN VALUE
GOOD F"LDW PATH

Figure 2 7. Flow Trajectory in a Vaned Diffuser.

Although seldom used, movable diffuser vanes or vane islands can
be used to alleviate the shock losses at off-design conditions.
However, as the adjusting mechanisms required are quite compli
cated, they generally are applied only to single-stage machines.

Losses in Centrifugal Compressors

Proper evaluation and estimation of centrifugal compressor
performance requires knowledge of various types of losses within
the compressor. Losses are typically expressed in terms of a
reduction in total pressure. Losses are divided into two major
groups-rotor and stator losses.

Rotor Losses
There are various types of rotor losses and are described briefly
below.

Shock Loss- This is common in compressors in aircraft

applications and occur due to the presence of a compression shock
wave at rotor inlet. Since a shock wave induces a ram pressure rise
in a very small distance, it will be accompanied by a total pressure
loss. Normal shock are accompanied by a greater pressure drop
than an oblique shock. Hence, the compressor rotor inlet should be
wedge-like, so as to obtain a weak oblique shock. If the blades were
blunt, this would lead to a bow shock, which would cause the flow
to detach from the blade leading edge and the loss to be much
higher.

A - NORMAL CONDITION. GOOD FLOW ANGLE. RELATIVELY

SHORT FLOW PATH. NOT liiUCH FRICTION LOSS.
B- TENDENCY TOWARD SURGE. SHALLOW FLOW ANGLE. LONG
FLOW PATH. LARGE FRICTION LOSS. POSSIBIUTY THAT
PART OF THE FLOW FIELD WILL RE-ENTER THE IMPELLER.

Figure 26. Flow Trajectory in a Vaneless Diffuser.

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

Incidence Loss-At off design conditions, flow enters the

inducer at an incidence angle that is either positive or negative.
Fluid approaching a blade with incidence suffers an instantaneous
change of velocity at the blade so as to comply with the blade inlet
angle causing a total pressure loss. Incidence loss increases as the
incidence angle increase.

Disk Friction Loss- This is the loss due to the frictional

torque on the back surface of the rotor between the hub and the
casing. This loss is the same for a given size disk, whether it is used
for a centrifugal compressor or a radial inflow turbine. In many
cases, the losses in the seals, the bearings, and the gear box are also
lumped in with this loss, and the entire loss can be called an
external loss . In this loss, unless the gap is of the order of
magnitude of the boundary layer, the effect of the gap size is
negligible. A point of interest that should be indicated here is that
the disk friction in a housing is less than that on a free disk. This
is due the existence of a core vortex which rotates at half the
angular velocity .

Diffusion Blading Loss- This loss arises because of negative

velocity gradients in the boundary layer. This deceleration of the
flow increases the boundary layer and gives rise to a separation of
the flow. The adverse pressure gradient, which a compressor
normally works against, increases the chances of separation and
gives rise to a rather significant loss.

Exit Loss- This loss is typically calculated by assuming that

approximately one half of the kinetic energy leaving the vaned
diffuser is lost.

Losses are complex phenomena and are a function of many

parameters such as inlet conditions, pressure ratios, blade angles,
flow , etc. The loss distribution is shown in Figure 28 in a typical
and centrifugal stage of pressure ratio below 2: 1 with backward
curved blades.

90

80

vaneless diffuser loss

Clearance Losses- When a fluid particle has a translatory

motion relative to a noninertial rotating coordinate system, it
experiences, besides other forces, a force known as the coriolis
force. A pressure difference exists between the driving and trailing
faces of an impeller blade, due mainly to Corio lis acceleration. The
shortest path, and generally that of least resistance, for the fluid to
flow and neutralize this pressure differential, is provided by the
clearance between the rotating impeller and the stationary casing.
In the case of the shrouded impellers, such a leakage from the
pressure side to the suction side of an impeller blade is not possible.
Instead, the existence of a pressure gradient in the c learance
between the casing and the impeller shrouds accounts for the
clearance loss. Clearance loss may be substantial. The leak flow
undergoes a large expansion and contraction due to temperature
variation across the clearance gap affecting the leaking flow and
the stream where it discharges.

17 1

vaned diffuser loss

exit loss
design
surge

Skin Friction Loss-Skin friction loss is defined as the loss

due to the shear forces on the impeller wall that are mostly due to
turbulent friction. This type of loss is usually determined by
considering the flow as an equivalent circular cross section with a
hydraulic diameter. The loss is then computed based on the well
known pipe flow pressure loss equations.

Stator Losses
Recirculating Loss- This loss occurs due to the back flow
into the impeller exit of compressor and is a direct function of the
air exit angle. As the flow through the compressor reduces, at the
same impeller speed, there is an increase in the absolute flow angle
at the exit of the impeller. Loss results due to the mixing of direct
and recirculating flow.

Wake Mixing Loss-This loss is due to the mixing of the low

velocity wake of the impeller blades with the main flow .

Vaneless Diffuser Loss- This loss i s the viscous flow loss

experienced within the vaneless diffuser.

Vaned Diffuser Loss- This takes into account the loss due to
blade incidence angle and the skin friction due to the vanes. Vaned
diffuser losses are based on the conical diffuser test results. They
are a function of the impeller blade loading and the vane less space
radius ratio.

choke

85

100
(%)

design flow

120

Figure 28. Losses in a Centrifugal Compressor.

COMPRESSOR SURGE

The phenomenon of surge, as it pertains to a centrifugal com

pressor and its connected systems, is an unstable condition result
ing in flow reversals and pressure fluctuations in the system. This
condition occurs when there is sufficient aerodynamic instability
within the compressor that the compressor is unable to produce
adequate pressure to deliver continuous flow to the downstream
system. The system and compressor then interact causing the surge
conditions with the large and sometimes violent flow oscillations
in the system. Surge, then, is an overall system phenomenon and
is not confined to the compressor only.
Surge is the result of an excessive increase in the resistance of
the system while the compressor is operating at a certain speed.
The added resistance reduces the flow to an unstable level. Alter
natively, if the resistance is unchanged, but the speed is reduced
appreciably, most systems will surge . Thus, surge occurrence
depends on the type of system and the shape of the resistance
curve.
The aerodynamic instability is brought about by flow reduction,
which causes stalling of one or more of the elements of a stage or
stages of the compressor. The stalling can occur at the inducer of
the impeller, in the radial portion of the impeller, in the diffuser,
or in the volute. A stall in one of these elements may not have
sufficient effect to cause the stage to be unstable. In fact, several
elements of a stage can stall without the entire stage stalling.
However, if the stalling is of sufficient strength, the stage will
become unstable, and this can lead to surge of the compressor.
A typical centrifugal multistage compressor performance map
used in the process industry is shown in Figure 29. The family of

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

172
1<40

550

500

450

2.50

"
Q..

.,;
"'
...
""
Q..

=i
0
"'
i5

100

2.25

400 2.00
350

120

2.75

1.75
0
;:::

0!

...
""
::l
"'
"'
...
""
Q..

;i

0
0::

1:

80

...

::t:
...
z
...
0
""
...
Q..

60
40
20
0

20

40

60

80

100

PERCENT INLET FLOW, Q, {ACFM)

SYSTEM CURVE (I)

SYSTEM CURVE (ll
SYSTEM CURVE

120

140

CONSTANT PRESSURE
PARTIALLY FRICTION-PARTIALLY CONSTANT
PREDOINANTLY FRICTION

Figure 29. Performance Map with System Curves and Surge

Cycles.
curves depicts the performance at various speeds, where N repre
sents rpm. The ordinate may be head "H," pressure ratio , discharge
pressure, or sometimes differential pressure. The abscissa, is
shown usually always as the inlet flow. It is important to under
stand that the inlet flow volume or capacity is based on a gas with
a particular molecular weight, specific heat ratio, and compress
ibility factor at a pressure and temperature corresponding to the
gas condition in the suction line to the compressor. If any of these
parameters is changed, the performance map is no longer exactly
valid.
The line on the left represents the surge limit. Operation to the
left ofthis line is unstable, resulting in unsatisfactory performance,
and is often harmful mechanically. Notice that the surge tl.ow
increases as the speed increases. In many multistage units, and
especially in units which are controlled by variable inlet guide
vanes, the surge line curves bend dramatically at higher flows and
speeds.
On the other side of the map, the capacity limit or overload line
is shown. Operation to the right of this line causes the head
producing capability of the machine to fall off very rapidly, and the
performance is difficult to predict. The area to the right of this line
is commonly known as "choke" or "stonewall." Operating the
machine in this region is usually harmless mechanically, although
a few impeller failures have been ascribed to prolonged operation
in stonewall.
The points labeled A, B, C, and D, and three typical system
operating curves which have been plotted. The terms frequently
used to define performance range is percent stability.
The rated "stable range" is generally taken as QA- Q6, where QA
is the design or rated point, and Q6 is the surge point along the
100% speed line. The percent stability is

fixed pressure drop, or pressure control in the particular system

external to the compressor through which the flow is being pumped.
However, it should be noted that, to follow any one of these system
curves, the speed must be changed. This, in turn, changes the flow.
With Systems 2 and 3, the head is also changed. These statements
are valid only if no changes within the system itself occur.
Changing the setting of a control valve, adding another piping
loop, or a changing catalyst level in a reactor which would modify
the system curve.
To a large extent, the frequency of the surge cycle varies
inversely with the volume of t he system. For example, if the piping
contains a check valve located near the compressor discharge
nozzle, the frequency will be correspondingly much higher than
that of a system with a large volume in the discharge upstream of
a check valve. The frequency can be as low as a few cycles per
minute or as high as 20 or more cycles per second. Generally
speaking , if the frequency is higher, the intensity of surge is lower.
The intensity or violence of surge tends to increase with increased
gas density, which is directly related to higher molecular weights
and pressures and lower temperatures. Higher differential pressure
generally increases the intensity.
ROTATING STALL

Rotating stall (propagating stall) consists of large stall zones

covering several blade passages, and propagates in the direction of
the rotor and at some fraction of rotor speed. The number of stall
zones and the propagating rates vary considerably. Rotating stall
can and does occur in centrifugal compressors.
The propagation mechanism can be described by considering
the blade row to be a cascade of blades (say, an inducer), as shown
in Figure 30. A t1ow perturbation causes blade two to reach a
stalled condition before the other blades. This stalled blade does
not produce a sufficient pressure rise to maintain the flow around
it, and an effective flow blockage or a zone of reduced flow
develops. This retarded flow diverts the flow around it so that the
angle of attack increases on blade three and decreases on blade one.
The stall propagates downward relative to the blade row at a rate
about half the blade speed; the diverted flow stalls the blades below
the retarded-flow zone and unstalls the blades above it. The
retarded flow or stall zone moves from the pressure side to the
suction side of each blade in the opposite direction of rotor

DIRECTION
Of
ROTATION

RETARDED
FLOW

(25)
DIRECTION
OF
PROPAGATION

expressed as a percentage.
The three representative system curves need little explanation.
The shape of these curves is governed by the amount of friction,

Figure 30. Rotating Stall in a Centrifugal Compressor Inducer.

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

rotation, and i t may cover several blade passages. The relative
speed of propagation has been observed from compressor tests to
be less than the rotor speed (40 to 75 percent of rotor speed) .
Observed from an absolute frame of reference, the stall zones
appear to be moving in the direction of rotor rotation. This
phenomenon can lead to inefficient performance and excitation of
the resonant frequency of the inducer, thus leading to failure of that
section. Rotating stall is accompanied sometimes by a pulsating
sound and pressure pulsations that can be noted at both the inlet and
exit sections of the impellers.

173

Centrifugal compressors for industrial applications have rela

tively low pressure ratios per stage. This condition is necessary so
that the compressors can have a wide operating range while stress
levels are kept at a minimum. Because of the low pressure ratios for
each stage, a single machine may have a number of stages in one
"barrel" to achieve the desired overall pressure ratio. Some of the
many configurations are shown in Figure 3 3 . S ome factors to be
considered when selecting a configuration to meet plant needs are :

Surge Detection and Control

Surge detection devices are those that attempt to avoid stall and
surge by the measurement of some compressor parameters which
meets or exceed the limit, of stable operation. A typical pressure
oriented anti- surge control system is shown in Figure 3 1 . The
pressure transmitter monitors the pressure and controls a device
which might open a blow-off valve. A temperature sensing device
corrects the readings for the effect of flow and speed for the effect
of temperature. A typical flow oriented surge control system is
shown in Figure 3 2 . These are two very simple surge control
systems. More complex surge control systems are needed for
complex configurations. With the advent of more computing
power, surge lines can be adjusted automatically in the field to take
care of operational changes.

l:, ,t

)??
I

,..-----r--,.- to process

pressure
transmitter

::;:::;: -pressure

alr ln
_-f--'

transmlt1er

,.

blowoH valve

""

surge
control
speed transmitter

to
atmosphere

Figure 31. Pressure Oriented Anti-Surge Control System.

llow measuring device

-=:-----...... to process
,--------...,I

L - - - - - _ ..,

_ _

- - -

- - - - - -

Back-to-hack impellers allow for a balanced rotor thrust and

minimize overloading the thrust bearings.

Single inlet or single discharge reduces external piping prob

lems.

Balance planes that are easily accessible in the field can

appreciably reduce field-balancing time.

blowoff valve

Balance piston with no external leakage will greatly reduce

wear on the thrust bearings.

surge

control

1
I
I
I
I

0--

air in

Intercooling between stages can considerably reduce the

power consumed.

Cold inlet or hot discharge at the middle of the case reduces

oil-seal and lubrication problems.

flow
transmitter

Figure 33. Various Configurations of Centrifugal Compressors.

to atmosphere
or
to air inlet

Hot and cold sections of the case that are adjacent to each other

Figure 32. Flow Oriented Anti-Surge Control System.

will reduce thermal gradients, and thus reduce case distortion.

Compressor Configuration

Horizontally split casings are easier to open for inspection

than vertically split ones, reducing maintenance time.

To properly design a centrifugal compressor, one must know the

operating conditions-the type of gas, its pressure, temperature,
and molecular weight. One must also know the corrosive proper
ties of the gas so that proper metallurgical selection can be made.
Gas fluctuations due to process instabilities must be pinpointed so
that the compressor can operate without surging.

Overhung rotors present an easier alignment problem because

shaft end alignment is necessary only at the coupling between the
compressor and driver.

Smaller, high-pressure compressors that do the same j ob will

reduce foundation problems, but will have greatly reduced opera
tional range.

1 74

PROCEEDINGS OF THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

Impeller Fabrication
Centrifugal-compressor impellers are either shrouded or un
shrouded. Open, shrouded impellers that are mainly used in single
stage applications are made by investment casting techniques or by
three-dimensional milling. Such impellers are used, in most cases,
for the high-pressure-ratio stages. The shrouded impeller is com
monly used in the process compressor because of its low pressure
ratio stages. The low tip stresses in this application make it a
feasible design. Several fabrication techniques are shown in Fig
ure 34. The most common type of construction is seen in (A) and
(B) where the blades are fill e t-welded to the hub and shroud. In (B),
the welds are full penetration. The disadvantage in this type of
construction is the obstruction of the aerodynamic passage . In (C) ,
the blades are partially machined with the covers and then butt
welded down the middle . For backward lean-angled blades, this
technique has not been very successful, and there has been difficul
ty in achieving a smooth contour around the leading edge. A slot
w elding technique is illustrated in (D), and is used where
blade-passage height is too small (or the backward lean-angle too
high) to permit convention,al fillet welding. In (E) , an electron
beam technique is still in its infancy, and work needs to be done to
perfect it. Its major disadvantage is that electron-beam welds
should preferably be stressed in tension but, for the configuration
of (E) , they are in shear. The configuration of (G) through (J) use
rivets. Where rivet heads protrude into the passage, aerodynamic
performance is reduced. With today ' s technology in the area of
welding, riveted impellers should be relegated to the scrap heap of
history .

A. fillet weld

B. full
penetration
fillet weld

C. butt weld

D. slot weld

E. electron
beam weld

F. machined

G. riveted

H. riveted

1. riveted

J. riveted

Figure 34. Several Fabrication Techniques for Centrifugal

Impellers.
aterials for fabricating these impellers are usually low-alloy
steels, such as AISI 4 1 40 or AISI 4340. AISI 4 1 40 is satisfactory
for most applications; AISI 4340 is used for larger impellers
requiring higher strengths. For corrosive gases, AISI 4 1 0 stainless
steel (about 12 percent chromium) is used. Monel K-500 is em
ployed in halogen gas atmospheres and oxygen compressors be
cause of its resistance to sparking . Titanium impellers have been
applied to chlorine service. Aluminum-alloy impellers have been
used in great numbers, especially at lower temperatures (below

300F) . With new developments in aluminum alloys, this range is

increasing. Aluminum and titanium are sometimes selected be
cause of their low density . This low density can cause a shift in the
critical speed of the rotor which may be advantageous.

BIBLIOGRAPHY

Alford, J. S . , "Protecting Turbomachinery from S elf-Excited Ro

tor Whirl," Journal of Engineering for Power, ASME Transac
tion, pp. 3 3 3-344 (October 1 96 5 ) .
Anderson, R. J . , Riter, W . K . , and Dilding, D . M . , " A n Investiga
tion of the Effect of Blade Curvature on Centrifugal-Impeller
Performance," NACA TN NO. 1 3 1 3 ( 1 947).
Balje, 0 . E . , "A Contribution to the Problem of D esigning Radial
Turbomachines," ASME Trans. , 74, p. 45 1 ( 1 95 2 ) .
Balje, 0 . E . , "A Study on Design Criteria a n d Matching of
Turbomachines Part 1 3, " Journal of Engineering for Power,
ASME Trans. , 84, S eries A, p. 1 0 3 ( 1 962) .
Balje, 0. E . , "A Study of Reynold Number Effects in Turboma
chinery ," Journal of Engineering for Power, A S ME Trans. 86,
Series A, p. 227 ( 1 964) .
Balje, 0 . E . , "Loss and Flow Path S tudies on C entrifugal Compres
sors-Part I & II," ASME Paper No. 70-GT- 1 2-A and 70-GT1 2 . B ( 1 970) .
Bammert, K. and Rautenberg, M . , "On the Energy Transfer in
Centrifugal Compressors," ASME Paper No. 74-GT- 1 2 1 ( 1 974).
Batman, J., "Design and Development of a Family of Natural Gas
Compressors for a 3000 HP Gas Turbine Engine, " ASME Paper
No. 72-GT- 1 0 ( 1 970) .
Benser, W. A. and Moser, J. J . , "An Investigation of Back Flow
Phenomena in Centrifugal Compressors, " NACA Report No.
806 ( 1 94 5 ) .
Betz, A . , Introduction to the Theory of Flow Machines, N e w York,
New York: Pergamon Press ( 1 966) .
Bhinder, F. S. and Ingham, D . R . , "The Effect of Inducer Shape on
the Performance of High Pressure Ratio Centrifugal Compres
sors," ASME Paper No. 74-GT- 1 2 2 ( 1 974) .
Boyce, M . P . , "A Practical Three-Dimensional Flow Visualization
Approach to the Complex Flow Characteristics in a Centrifugal
Impeller," ASME Paper No. 66-GT-8 3 ( 1 966).
Boyce, M . P. "Design of Compressor Blade Suitable for Transonic
Axial-Flow Compressors," ASME Paper No. 76-GT-47 ( 1 97 6 ) .
Boyce, M . P . , "How t o Achieve Online Availability o f Centrifugal
Compressors," Chemical Engineering, pp. 1 1 5 - 1 1 7 (June 5 ,
1978).
Boyce, M . P . , "Centrifugal Compressors," Process Equipment
S eries, 3 ( 1 98 1 ) .
Boyce, M . P . , Gas Turbine Engineering Handbook, Houston,
Texas : Gulf Publishing ( 1 98 7 ) .
Boyce, M . P. and Bale, Y . S . , "Diffusion Loss in a Mixed Flow
Compressor, Paper No. 72906 1 , Intersociety Energy Conver
sion Compressor Impeller," NRCC ME-220, Ottawa, Canada,
(July 1 966) .
Boyce, M . P. and Bale, Y. S . , "A New Method for the Calculations
of Blade Loadings in a Radial Flow Compressor, " ASME Paper
No. 7 1 -GT-60 ( 1 97 1 ) .
Boyce, M . P. and Desai, A . R . , "Clearance Loss i n a Centrifugal
Impeller," Paper No. 7 3 9 1 26 . Proc. of the 8th Intersociety
Energy Conversion Engineering Conference, p. 638 ( 1 973 ) .
Boyce, M . P., Schiller, R . N . M . , and Desai, A . R . , "Study of
Casing Treatment Effects in Axial Flow Compressors," ASME
Paper No. 74-GT-8 9 ( 1 974) .
Boyce, M. P. and Hanawa, D . A . , "Parametric S tudy of a Gas
Turbine," J. Eng. Power, (December 1 9 7 5 ) .
Boyce, M . P . , Morgan, E . , a n d White, G . , " Simulation of Rotor
Dynamics of High Speed Rotating Machinery," Proceedings of

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

the First Int. Conf. Centrifugal Compressor Techno I . , pp. E6-32,
Madras, India ( 1 978) .
Bullock, R. 0. and Finger, H. B . , " S urging in Centrifugal and
Axial-Flow Compressors," SAE Quar. Trans. 6, p. 220 ( 1952).
Cameron, J . A . and Danowski, F . M . " S ome Metallurgical Consid
erations in Centrifugal Compressors, " Proceedings of the Sec
ond Turbomachinery Symposium, The Turbomachinery
Laboratory, The Texas A&M University System, College Sta
tion, Texas, pp. 1 1 6- 1 2 8 ( 1 97 3 ) .
Campbell, K . and Talbert, J. E . , " S ome Advantages and Limita
tions of Centrifugal and Axial Compressors, " SAE Trans. , 53, p .
607 ( 1 945 ) .
Cheshire, L. J . , "Centrifugal Compressors for Aircraft Gas Tur
bines," Proc. L. Mech. E . , 1 53, 1 96 5 ; reprinted by the ASME,
"Lectures on the Development of the British Gas-Turbine Jet
Unit" ( 1 967) .
Clapp, A. M . , "Fundamentals of Lubrications Relating to Opera
tion and Maintenance of Turbomachinery, " Proceedings of the
First Turbomachinery Symposium, The Turbomachinery Labo
ratory, The Texas A&M University System, College Station,
Texas, pp. 66-74 ( 1 972 ) .
Concorida, C . and Dowell, M. F . , " Analytical Design o f Centrif
ugal Air Compressors, " J. App. Mechs . , 1 3, p. A-27 1 ( 1 946).
Coppage, J . E . , et al. , "Study of Supersonic Radial Compressors
for Refrigeration and Pressurization System," W ADC Techni
cal Report 5 5-257, Astia Document No. AD 1 1 0467 ( 1 95 6 ) .
Curle, N . and Skan, S . W . , "Approximate Method for Predicting
Separation Properties of Laminar Boundary Layers," Aero Wuart,
G, 257 (August 1 957).
Davis, H. M . , Centrifugal Compressor Operation and Mainte
nance, Proceedings of the First Turbomachinery Symposium,
The Turbomachinery Laboratory, The Texas A&M University
System, College Station, Texas, pp. 1 0-25 ( 1 972) .
Dean, R. C . , "The Fluid Dynamic Design of Advanced Centrifugal
Compressors," Von Karman Institute, Lecture Notes No. 50,
( 1 972 ) .
Delamey, R. A . and Kavanagh, P . , "Transonic Flow Analysis i n
Axial Flow Turbomachinery Cascades by a Time-Dependent
Method of Characteristics," ASME Paper No. 75 -GT-8 ( 1 97 5 ) .
Eckardt, D . , "Instantaneous Measurements in the Jet-Wake Dis
charge Flow of a Centrifugal Compressor Impeller," ASME
Paper No. 74-GT-90 ( 1 974) .
Eckert, B . , "Axial and Radialkompressoren, " Springer ( 1 9 5 3 ) .
Edminister, W. C., A pplied Hydrocarbon Thermodynamics, Hous
ton, Texas : Gulf Publishing ( 1 98 3 ) .
Ehrich, F. F . , "Identification and Avoidance o f Instabilities and
Self- Excited Vibrations in Rotating Machinery," Adopted from
ASME Paper 72-DE-2 1 . General Electric Co. , Aircraft Engine
Group, Group Engineering Division, ( 1 972) .
Ellis, G. D . and Stanitz, J . D . , "Two Dimensional Compressible
Flow in Centrifugal Compressors with Logarithmic-Spiral
Blades, NACA TN No. 2255 ( 1 95 1 ) .
Ellis, G. 0 . and Stanitz, J. D . , " Comparison o f Two and Three
Dimensional Potential-Flow S olutions in a Rotating Impeller
Passage," NACA TN No. 2806 ( 1 9 5 2 ) .
Ferguson, T . B . , "The Centrifugal Compressor Stage," Butter
worth and Co., Ltd . , London ( 1 9 63).
Flaherty, R . , "A Method for Estimating Turbulent Boundary
Layers and Heat Transfer in an Arbitrary Pressure Gradient,"
United Aircraft Research Laboratories Report UAR-G5 1 (Au
gust 1 96 8 ) .
Fowler, H. S . , " A n Investigation o f the Flow Processes in a
Centrifugal Compressor Impeller," NRCC ME-220, Ottawa,
Canada (July 1 966) .
Fowler, H. S . , "Experiments on the Flow Processes in S imple
Rotating Channels," NRCC ME-229, Ottawa, Canada ( 1 969).

175

Ginsberg, A . , Ritter, W . K., and Polasics , J., "Effects of Perfor

mance of Changing the Division of Work Between Increase of
Angular Velocity and Increase of Radius of Rotation in an
Impeller," NACA TN No. 1 2 1 6 ( 1 947 ) .
Gregory, A. T . , "Elastic Fluid Compressor, " United States Patent
Office, 3305 1 65 (February 2 1 , 1 967) .
Gresh, M . T . , "Compressor Performance," Butterworth and Hei
nemann ( 1 99 1 ) .
Gunter, E. J . , " Dynamic Stability o f Rotor Bearing Systems, "
N A S A Sp- 1 1 3 ( 1 966) .
Gunter, E. J . , Jr. , "Rotor Bearing Stability, " Proceedings of the
First Turbomachinery Symposium, The Turbomachinery Labo
ratory, The Texas A&M University S ystem, College Station,
Texas ( 1 972) .
Haag, A. C . , "The Influences of Oil Film Journal Bearings on the
Stability of Rotating Machines," J. of Appl. Mech. Trans.
ASME, 68, p. 2 1 1 ( 1 94 6 ) .
Hamrick, J . T . , Ginsbert, A . , a n d Osborn, W . M . , "Method o f
Analysis for Compressible Flow Through Mixed-Flow Centrif
ugal Impellers of Arbitrary Design," NACA Report No. 1 08 2
( 1 952).
Hansen, A . G. (Editor) , " Symposium on Fully S eparated Flows,"
ASME Fluids Energy Division Conference, Philadelphia, Penn
sylvania ( 1 964) .
Hayes, W. D . , "The Three Dimensional Boundary Layer" NA
VORD Report No. 1 3 1 3 , Washington, D . C . , 1 05 1 .
Herbage, B . S . , "High Speed Jourrfal and Thrust Bearing Design,"
Proceedings of the First Turbomachinery Symposium, The
Turbomachinery Laboratory, The Texas A&M University Sys
tem, College Station, Texas, pp. 55-6 1 ( 1 972).
Housmann, J . G . , "Turbomachinery Specifications," Proceedings
of the First Turbomachinery Symposium, The Turbomachinery
Laboratory, The Texas A&M University System, College Sta
tion, Texas, pp. 77-78, ( 1 97 2 ) .
Jackson, C . J . , "Cold and Hot Alignment Techniques for Turboma
chinery, " Proceedings of the Second Turbomachinery Sympo
sium, The Turbomachinery Laboratory, The Texas A&M
University System, College Station, Texas, pp. 1 -7 ( 1 97 3 ) .
Kasai, T . , " O n the Exit Velocity and Slip Coefficient o f Flow a t the
Outlet of the Centrifugal Pump Impeller," Memoirs of the
Faculty of Engineering, Kyushu Imperial University, Japan, 8,
( 1 ) ( 1 93 6) .
Katsanis, T . , "Use f Arbitrary Quasi-Orthogonals for Calculating
Flow Distribution in the Meridional Plane of a Turbomachine, "
N A S A TND-2546 ( 1 9 64 ) .
Kearton, W . F . , "The Influence o f the Number o f Impeller Blades
on the Pressure Generated in a Centrifugal Compressor and of
its General Performance," Proc. I . Mech. E., 1 24, p . 48 1 ( 1 9 3 3 ) .
Klassen, H. A . , "Effect of Inducer Inlet and Diffuser Throat Areas
on Performance of a Low Pressure Ratio Sweptback Centrifugal
Compressor," NASA TM X-3 1 4 8 , Lewis Research Center (Jan
uary 1 9 7 5 ) .
Kovats, A . , "Design a n d Performance of Centrifugal and Axial
Flow Pumps and Compressors, " New York, New York: The
Macmillan Co. ( 1 964).
Kramer, J . J . , Osborn, W . M., and Hamrick, J . T . , "Design and Test
of Mixed-Flow and Centrifugal Impellers, " Journal of Engi
neering for Power, ASME Trans. , S eries A . , 82, p. 127 ( 1 960) .
Kuethe, A . M . and S chetzer, J . D . , Foundations of Aerodynamics,
New York, New York: John Wiley & Sons ( 1 950).
Lapina, R. P . , Estimating Centrifugal Compressor Performance,
Houston, Texas : Gulf Publishing ( 1 9 8 1 ) .
Lazarkiewicz, S . and Troskolanski, A . T . , "Impeller Pumps,"
Pergamon Press ( 1 96 5 ) .
Lesiecki, G . , " Evaluation of Liquid Film Seals: Associated Sys
tems and Process Considerations," Proceedings of the Sixth

176

PROCEEDINGS O F THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM

Turbomachinery Symposium, The Turbomachinery Laboratory,

The Texas A&M University System, College Station, Texas, pp.
1 45 - 1 4 8 ( 1 977).
Lewis, R . A . , Mechanical Contact Shaft Seal," Proceedings of the
Sixth Turbomachinery Symposium, The Turbomachinery Labo
ratory, The Texas A&M University System, College Station,
Texas, 1 977 .
Lown, H. and Wiesner, F. J . , Jr., Prediction of Choking Flow in
Centrifugal Impeller," Transaction of ASME Journal of Basic
Engineering, (March 1 959) .
Mechanical Engineers Handbook, "Centrifugal and Axial Fans, "
6th Edition, p . 1 4-66 t o 1 4-79.
Michel, D . J., Mizisin, J., and Prian, V . D., Effect of Changing
Passage Configuration on Internal-flow Characteristics of a 48inch Centrifugal Compressor," NACA TN No. 2706 ( 1 95 2 ) .
Mikolaj czak, A . A . , Weingold, H . D . , and Nikkanen, J . P . , "Flow
Through Cascades of S lotted Compressor Blades, " Journal of
Engineering for Power, ASME Trans. , p . 5 7 (January 1 970) .
Moody, L.F., "Friction Factors," ASME Trans. , p . 672, (Novem
ber 1 944) .
Newkirk, B. L . , " Shaft Whipping," General Electric Review, 2 7,
p. 1 69 ( 1 924) .
N uell , W . V. D . , " The Maximum Delivery Pressure of Single Stage
Radial Supercharges for Aircraft Engines," NACA TN No. 949
( 1 940) .
Owczarek, J. A . , Fundamentals of Gas Dynamics," Philadelphia,
Pennsylvania: International Textbook Company ( 1 96 8 ) .
Peck, J . F . , "Investigations Concerning Flow Conditions in a
Centrifugal Pump, and the Effect of Blade Loading of Head
Slip," Proc. I . Mech. E., 1 64, ( 1 ) , p . 1 ( 1 95 1 ) .
Petermann, H . , Untersuchungen a m Zentsipetalrad fur Kreiselver
dichter Forshung, 1 7 ( 1 952).
Pfleiderer, C . , "Kreisel pumpen, " Springer, F 1 5 5 .
Prian, V . D . and Michel, D . J . , "An Analysis o f Flow i n Rotating
Passage of Large Radial Inlet Centrifugal Compressor at Top
Speed of 700 Feet Persecond, " NACA TN No. 2584 ( 1 95 1 ) .
Reshotko, E . and Tucker, M . , "Approximate Calculation o f the
Compressible Turbulent Boundary Layer with Heat Transfer
and Arbitrary Pressure Gradient," NACA Tn 4 1 5 4 ( 1 95 4 ) .
Ribary, F . , "The Friction Losses of Steam Turbine Discs," Brown
Boveri Review, 2 1 , p. 1 20 ( 1 934).
Ritter, W . K . and Johnson I. A . , "Performance Effect of Fully
Shrouded a Centrifugal Supercharger Impeller," NACA ARR
E5H23 or E 1 8 .
Rodgers, C . , "Influence o f Impeller and Diffuser Characteristics
and Matching on Radial Compressor Performance," SAE Re
print 268B (January 1 96 1 ) .
Rodgers, C . , "Typical Performance Characteristics o f Gas Turbine
Radial Compressors," Journal of Engineering for Power, ASME
Trans. , p. 1 6 1 (April 1 964) .
Rodgers, C. and Sapiro, L . , "Design Considerations for High
Pressure Ratio Centrifugal Compressors, " ASME Paper No. 72GT-9 1 ( 1 972) .
Sawyer 's Gas Turbine Engineering Handbook- Vol. I, II, III.
S apiro, L., "Preliminary Staging Selection for Gas Turbine Driven
Centrifugal Gas Compressors, " ASME Paper No. 7 3-GT-3 1
( 1 97 3 ) .
Schlichting, H . , " Application o f Boundary-Layer Theory in Tur
bomachinery, " ASME Trans. 81 , p. 543 ( 1 95 9 ) .
S chlichting, H . , Boundary Layer Theory, N e w York, N e w York:
McGraw-Hill Book Co. ( 1 968 ) .
Schmieden, C . , "Uber den Widerstand einer i n einer Flussigkeit
Rotierenden S cheibe," ZAMM 8, p. 460 ( 1 92 8 ) .
S c hultz-Gruno w , F . , " D e r Eirbungswiderstand Rotierender
Scheiben in Gehansen" ZAMM 15, p . 1 9 1 ( 1 935) .
Senoo, Y. and Nakase, Y . , "An Analysis of Flow Through a Mixed

Flow Impeller," ASME Paper No. 7 1 -GT-2 ( 1 97 1 ) .

Senoo, Y . , e t al., "Viscous Effects o n S lip Factor o f Centrifugal
Blowers, " ASME publication 73-GT- 5 6 ( 1 9 7 3 ) .
Sheets, H . E . , "The Flow Through Centrifugal Compressors and
Pumps," ASME Trans. 72, p. 1 009 ( 1 950).
Shepherd, D.G., Principles of Turbomachinery, New York, New
York: The MacMillan Company, ( 1 95 7 ) .
Shouman, A . R. and Anderson, J . R . , "The U s e of Compressor Inlet
Prewhirl for the Control of Small Gas Turbines," Journal of
Engineering for Power, Trans. ASME Ser. A . , 86, pp. 1 3 6- 1 40
( 1 964) .
Sovran, G. and Klomp, E. D . , Fluid Mechanics of Internal Flow,
New York, New York: Elsevier Publishing Co. ( 1 967) .
Speer, I. E . , " Design and Development o f a Broad Range High
Efficiency Centrifugal Compressor for a Small Gas Turbine
Compressor Unit, ASME Paper No. 52-SA- 1 4 ( 1 952).
Stahler, A . F . , "Transonic Flow Problems in Centrifugal Compres
sors," SAE Reprint 2 6 8 C (January 1 9 6 1 ) .
Stanitz, J . D . , "Two Dimensional Compressible Flow i n Coni cal
Mixed Flow Compressors, " NACA TN No. 1 744 ( 1 948).
Stanitz, J . D . , "Two Dimensional Compressible Flow in Turboma
chines with Conic Flow Surfaces," NACA Report No. 935
( 1 949) .
Stanitz, J. D . , " S ome Theoretical Aerodynamic Investigation of
Impellers in Radial and Mixed Flow Centrifugal Compressors,"
ASME Trans. 74, p . 472 ( 1 95 2 ) .
S tanitz, J . D . and Ellis, G . 0 . , "Two Dimensional Compressible
Flow In Centrifugal C ompressors with Straight Blades," NACA
Report No. 954 ( 1 950).
Stanitz, , J . D . and Ellis, G . 0 . , "Two Dimensional Flow on General
Surfaces of Revolution in Turbomachines," NACA TN 265 4 .
Stanitz, J . D . and Brian, V . D . , "A Rapid Approximate Method for
Determining Velocity Distribution on Impeller Blades of Cen
trifugal Compressors, " NACA TN No. 242 1 ( 1 95 1 ) .
Stepanoff, A . J . , Turboblowers, New York, New York: Wiley
( 1 955 ) .
Stepanoff, A . J . , Centrifugal and Axial Flow Pumps, New York,
New York: John Wiley & S ons, Inc . , ( 1 96 8 ) .
Stockman, N . 0 . and Kramer, J . L . , "Method for Design o f Pump
Impellers Using a High Speed Digital Computer," NASA TND
1 562.
Stodola, A., " S team and Gas Turbines, " New York, New York:
Mc-Graw-Hill Book Co . , ( 1 92 7 ) .
Thomson, W . T . , Mechanical Vibrations, S econd Edition, Engle
wood Cliffs, New Jersey : Prentice-Hall, Inc . ( 1 9 6 1 ) .
Vavra, M . H . , Aerothermodynamics and Flow i n Turbomachines,
New York, New York: John Wiley & S ons, Inc . ( 1 960) .
W atabe, K . , "Effects of Clearances and Grooves on Fluid Friction
of Rotation Discs," JSME Bulletin, 8, (29) , p. 55 (February
1 965) .
Watabe, K . , JSME Bulletin, 7, (November 26, 1 966).
Watabe, K . , "Experiments on Fluid Friction of Rotating Disc with
Blades, " Japan Society of Mechanical Engineers Bulletin, 5,
( 1 7) .
Wiesner, F . J . , Jr. , "Practical S tage Performance Correlations for
Centrifugal Compressors," ASME Paper No. 60-Hyd- 1 7 ( 1 960) .
Wilcock, D. E. S . and Booser, E. R . , "Bearing Design and Appli
cation," New York, New York: McGraw-Hill ( 1 9 6 1 ) .
Wislicensu, G. F . , "Fluid Mechanics o f Turbomachinery," New
York, New York: Mc-Graw-Hill Book C o . , Inc . ( 1 947).
Woodhouse, H., "Inlet Conditions of Centrifugal Compressors for
Aircraft Engine Superchargers and Gas Turbines," J . lust. Aeron.
S c . , 15, p. 403 ( 1 94 8 ) .
Wu, C . H . , " A General Theory o f Three Dimensional Flow i n
S ubsonic and Supersonic Turbomachines of Axial, Radial and
Mixed-Flow Type, " NACA TN 2604, ( 1 9 5 2 ) .

TUTORIAL ON PRINCIPLES OF OPERATION AND PERFORMANCE ESTIMATION OF CENTRIFUGAL COMPRESSORS

Yuan, S . W . , Foundations o f Fluid Mechanics, N e w Delhi, India:
Prentice Hall ( 1 969) .
York, R. E. and Woodard, H. S . , "Supersonic Compressor Cas
cades-An Analysis of the Entrance Region Flow Field Con
taining Detached Shock Waves," ASME Paper No . 75-GT-3 3
( 1 97 5 ) .

177

178

PROCEEDINGS O F THE TWENTY-SECOND TURBOMACHINERY SYMPOSIUM