NUMERICAL SIMULATION OF ROTATING STALL AND SURGE

ALLEVIATION IN AXIAL COMPRESSORS
A Thesis
Presented to
The Academic Faculty
by
Saeid Niazi
In Partial Fulfillment
of the Requirements for the Degree
Doctor of Philosophy in Aerospace Engineering
Georgia Institute of Technology
July 2000
ii
NUMERICAL SIMULATION OF ROTATING STALL AND SURGE
ALLEVIATION IN AXIAL COMPRESSORS
Approved:
______________________________
Lakshmi N. Sankar, Chairman
______________________________
J.V.R. Prasad
______________________________
Suresh Menon
______________________________
Stephen M. Ruffin
______________________________
Prasanna V. Kadaba
Date Approved _________________
iii
IN THE NAME OF GOD
Dedicated to my parents
Behjat Golbahar Haghighi and Sadrollah Niazi
for their most precious gift of selfless love
iv
ACKNOWLEDGEMENTS
Many people have touched my life not only from an academic point of view, but
both their friendship and spiritual support have been a source of encouragement and
strength to complete this work.
I would like to thank Dr. Lakshmi N. Sankar, my teacher and dissertation advisor,
for his support throughout the research period. His patience and kindness with his
detailed knowledge about this research topic played a key role in the development of this
work. I am really honored and grateful to have had the opportunity to work under him.
I would also like to thank Dr. J. V. R. Prasad for his helpful suggestions and
technical expertise in the area of flow control. Many thanks to Dr. S. Menon, Dr. S. M.
Ruffin, and Dr. P. V. Kadaba for their services as members of my reading committee and
their valuable comments.
The financial assistance given by Army Research Office under the Multidisciplinary
University Research Initiative (MURI) on Intelligent Turbine Engine is gratefully
acknowledged.
Thanks to my co-worker and my friend Alex Stein for his great efforts during code
development and his steadfast support throughout the research.
I would also like to thank the past and present members of CFD lab for their warm
friendship and support during my administrative responsibility in the lab. To: Justin
Russell, Mert Berkman, Ebru Usta, Mehmet Sahin, Guanpeng Xu, Zhong Yang, Yi Liu,
v
Zhijian Liu, Masayoshi Senga, and Gang Wang. Thanks also to my friend and roommate
Konstantin Ignatiev in the School of Materials and Science Engineering for providing me
assistance with computer services. I would like to thank Catherine Moseley Matos for
taking the time to read this dissertation and for her helpful suggestions. Thanks also to
all my friends whose advice and support made this achievement possible.
I would like to thank Dr. J. Jagoda, the graduate coordinator, and the staff of the
Aerospace Engineering Office: Revonda B. Mullis, Loretta Carroll, Terry M. Parrott,
Margaret A. Ojala, Carole W. Gaines, and Howard L. Simpson for being always ready to
help me.
I would like to thank all my sisters and brothers and their families for their great
support during my study. A special note of thanks to my brother and friend, Hamid and
his family, Hilda and sweet Sara, for their generous support during the pursuit of my
degree in Atlanta.
Finally, I owe my greatest gratitude to my parents, Behjat Golbahar Haghighi and
Sadrollah Niazi, to whom this work is dedicated. Their constant encouragement and
great unconditional love for myself and each of my sisters and brothers have been a
source of inspiration in my life.
vi
TABLE OF CONTENTS
DEDICATIONS iii
ACKNOWLEDGEMENTS iv
TABLE OF CONTENTS vi
LIST OF TABLES ix
LIST OF FIGURES x
NOMENCLATURE xv
SUMMARY xix
I INTRODUCTION..................................................................................................... 1
1.1 AN OVERVIEW OF COMPRESSOR OPERATIONS ................................................. 2
1.2 COMPRESSOR STABILITY......................................................................................... 4
1.3 OBJECTIVES AND ORGANIZATION OF THE PRESENT WORK.......................... 6
II AN OVERVIEW OF COMPRESSOR INSTABILITY PHENOMENA........... 12
2.1 FUNDAMENTALS OF ROTATING STALL............................................................. 12
2.2 FUNDAMENTALS OF SURGE.................................................................................. 15
2.3 LITERATURE SURVEY OF STUDIES ON ROTATING STALL AND SURGE .... 17
2.3.1 EXPERIMENTAL STUDIES ON COMPRESSOR CONTROL..................................... 18
2.3.2 COMPUTATIONAL STUDIES OF COMPRESSOR PERFORMANCE
vii
AND CONTROL............................................................................................................. 23
III MATHEMATICAL AND NUMERICAL FORMULATION............................. 31
3.1 GOVERNING EQUATIONS....................................................................................... 31
3.2 DISCRETIZATION OF THE GOVERNING EQUATIONS....................................... 35
3.3 NUMERICAL SOLUTION OF THE DISCRETIZED EQUATIONS......................... 41
3.4 TURBULENCE MODELING...................................................................................... 44
3.5 INITIAL AND BOUNDARY CONDITIONS ............................................................. 47
3.5.1 INITIAL CONDITIONS .................................................................................................. 47
3.5.2 BOUNDARY CONDITIONS .......................................................................................... 48
IV CODE VALIDATION STUDIES .......................................................................... 58
4.1 DESCRIPTION OF THE CONFIGURATIONS STUDIED....................................... 59
4.1.1 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 67)............................................. 59
4.1.2 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 37)............................................. 61
4.2 PREVIOUSLY REPORTED STUDIES ON ROTOR 67 AND ROTOR 37
CONFIGURATIONS.................................................................................................. 62
4.3 PEAK EFFICIENCY RESULTS FOR ROTOR 67...................................................... 63
V RESULTS AT OFF-DESIGN CONDITIONS...................................................... 83
5.1 ROTOR 67 RESULTS.................................................................................................. 84
5.1.1 ONSET OF STALL.......................................................................................................... 84
5.1.2 ROTATING STALL AND MODIFIED SURGE SIMULATIONS................................. 88
viii
5.2 ROTOR 37 SIMULATIONS........................................................................................ 90
VI ACTIVE CONTROL STUDIES OF ROTOR 67 CONFIGURATION........... 104
6.1 OPEN-LOOP CONTROL STUDIES ......................................................................... 106
6.2 CLOSED-LOOP CONTROL STUDIES .................................................................... 110
VII CONCLUSIONS AND RECOMMENDATIONS.............................................. 125
7.1 CONCLUSIONS......................................................................................................... 126
7.2 RECOMMENDATIONS............................................................................................ 127
REFERENCES.............................................................................................................. 131
VITA ............................................................................................................................. 142
ix
LIST OF TABLES
Table 6.1 Summary of stall margin extension for the two control
schemes studied………………………………………………………...114
x
LIST OF FIGURES
Figure 1.1 Work input machinery classification, Reference [5]. ........................................ 8
Figure 1.2 Schematic diagram of changes in fluid properties and velocity through an
axial compressor stage, References [5], [11]. ................................................... 9
Figure 1.3 Typical compressor characteristic map for axial and centrifugal
compressors..................................................................................................... 10
Figure 1.4 Operational stability, matching the compressor and throttle characteristics. . 10
Figure 1.5 Effects of rotor RPM on compressor performance and stability. .................... 11
Figure 2.1 Rotating stall inception.................................................................................... 27
Figure 2.2 Transient response of system in rotating stall, Reference [14]. ..................... 27
Figure 2.3 Part-Span rotating stall with different stalled cells, Reference [12]................ 28
Figure 2.4 Full-Span rotating stall with different stalled cells, Reference [12]................ 28
Figure 2.5 Compressor map with the stalled flow characteristic, Reference [13]. ........... 29
Figure 2.6 Transient response of system in surge, Reference [14]. ................................. 29
Figure 2.7 Types of active and passive compressor control schemes............................... 30
Figure 3.1 Control volume and cell-vertex grid points..................................................... 54
Figure 3.2 Computation of inviscid flux terms on a cell face........................................... 54
Figure 3.3 Boundary conditions used in the axial compressor analysis. .......................... 55
Figure 3.4 Compressor outflow boundary conditions....................................................... 56
Figure 3.5 Periodic and zonal boundary conditions......................................................... 56
xi
Figure 3.6 Implementation of bleeding as an open-loop active control. ......................... 57
Figure 3.7 Implementation of bleeding as a closed-loop active control. ......................... 57
Figure 4.1 NASA Rotor 67 configuration, Reference [82].............................................. 70
Figure 4.2 Laser anemometer and aerodynamic locations for Rotor 67, Reference [90].70
Figure 4.3 Computational grid for Rotor 67. ................................................................... 71
Figure 4.4 Streamwise computational grid at midspan for Rotor 67. .............................. 71
Figure 4.5 Pitchwise computational grid for Rotor 67. .................................................. 72
Figure 4.6 Computational grid in the meridional plane for Rotor 67. ............................. 72
Figure 4.7 NASA Rotor 37 configuration, Reference [84]............................................. 73
Figure 4.8 Laser anemometer and aerodynamic locations for Rotor 37, Reference [90].73
Figure 4.9 Computational grid for Rotor 37. ................................................................... 74
Figure 4.10 Time history of mass flow rate fluctuations about mean flow at peak
efficiency conditions (Rotor 67). ................................................................. 74
Figure 4.11 Pressure and mass flow rate fluctuations at peak efficiency conditions
(Rotor 67). ..................................................................................................... 75
Figure 4.12 Comparison between experiments and present results of inlet axial velocity at
peak efficiency conditions (Rotor 67). .......................................................... 75
Figure 4.13 Comparison between experiments and present predictions of relative Mach
number contours at 30% span at peak efficiency (Rotor 67). ....................... 76
Figure 4.14 Comparison between experiments and present results of relative Mach
number at 90% span and 30% pitch at peak efficiency (Rotor 67). .............. 77
xii
Figure 4.15 Comparison between experiments and present results of relative Mach
number at 90% span and 50% pitch at peak efficiency (Rotor 67). ............. 77
Figure 4.16 Static pressure contours at 30% and 70% span at peak efficiency
(Rotor 67). ..................................................................................................... 78
Figure 4.17 Relative velocity profile at mid-span at peak efficiency (Rotor 67). ............ 79
Figure 4.18 Secondary flow in the circumferential direction at peak efficiency
(Rotor 67). ..................................................................................................... 79
Figure 4.19 Flow field near blade suction surface at peak efficiency (Rotor 67)............. 80
Figure 4.20 Flow field near blade pressure surface at peak efficiency (Rotor 67). .......... 81
Figure 4.21 Flow field at mid-passage at peak efficiency (Rotor 67). ............................. 82
Figure 5.1 Comparison of measured and computed characteristic performance map
(Rotor 67). ..................................................................................................... 93
Figure 5.2 Comparison of measured and computed adiabatic efficiency (Rotor 67). .... 93
Figure 5.3 Locations of the computational probes (Rotor 67 and Rotor 37). ................. 94
Figure 5.4 Pressure and mass flow rate fluctuations at the onset of stall, operating
point B (Rotor 67). ........................................................................................ 95
Figure 5.5 Time history of averaged pressure computed by probes upstream of the
compressor face at operating point B (Rotor 67). ........................................ 95
Figure 5.6 Time history of pressure deviation from its azimuthally averaged pressure
upstream of the compressor face at operating point B (Rotor 67). ............... 96
Figure 5.7 Time history of averaged axial velocity computed by probes upstream of the
xiii
compressor face at operating point B (Rotor 67). ......................................... 96
Figure 5.8 Time history of mass flow rate fluctuations at operating point C
(Rotor 67). ..................................................................................................... 97
Figure 5.9 Pressure and mass flow rate fluctuations at operating point C (Rotor 67). ... 97
Figure 5.10 Temporal growth of reversed flow at operating point C (Rotor 67). ............ 98
Figure 5.11 Instantaneous circumferential pressure fields at operating point C
(Rotor 67). ..................................................................................................... 99
Figure 5.12 Time history of averaged axial velocity and pressure computed by probes
upstream of the compressor face at operating point C (Rotor 67). ............. 100
Figure 5.13 Power spectral density of the azimuthally averaged pressure fluctuations
upstream of the compressor face at operating point C (Rotor 67). ............. 101
Figure 5.14 Time history of axial velocity deviation from its azimuthally averaged axial
velocity upstream of the compressor face at operating point C (Rotor 67). 101
Figure 5.15 Computed and measured characteristic performance map at 70% design
speed (Rotor 37). ........................................................................................ 102
Figure 5.16 Fluctuations of total pressure ratio versus mass flow rate at different
operating conditions (Rotor 37). ................................................................. 102
Figure 5.17 Time history of pressure deviation from its azimuthally averaged pressure
upstream of the compressor face at operating point C (Rotor 37). ............. 103
Figure 5.18 Power spectral density of pressure deviation from its azimuthally averaged
pressure fluctuations upstream of the compressor face at operating point C
(Rotor 37)................................................................................................... 103
xiv
Figure 6.1 Characteristic performance map with bleed control (Rotor 67). ................. 115
Figure 6.2 Velocity profile with/without open-loop control (Rotor 67). ...................... 116
Figure 6.3 Total pressure ratio and mass flow rate fluctuations with/without open-loop
control (Rotor 67). ...................................................................................... 117
Figure 6.4 Characteristic performance map with throttle characteristic for open-loop
control (Rotor 67). ...................................................................................... 118
Figure 6.5 Spanwise distribution of axial velocity at mid-passage near the leading edge
(Rotor 67). .................................................................................................. 119
Figure 6.6 Time history of axial velocity with/without open-loop control (Rotor 67). 120
Figure 6.7 Time history of axial velocity deviation from its azimuthally averaged axial
velocity with/without open-loop control (Rotor 67). ................................. 121
Figure 6.8 Time history of static pressure under open-loop control (Rotor 67). .......... 122
Figure 6.9 Upper and lower limit of pressures that trigger closed-loop control
(Rotor 67). .................................................................................................. 122
Figure 6.10 Total pressure ratio and mass flow rate fluctuations under closed-loop
control (Rotor 67). ..................................................................................... 123
Figure 6.11 Characteristic performance map with throttle characteristic for closed-loop
control (Rotor 67). ..................................................................................... 123
Figure 6.12 Time history of axial velocity under closed-loop control (Rotor 67)......... 124
Figure 6.13 Time history of axial velocity deviation from its azimuthally averaged axial
velocity upstream of the compressor face under closed-loop control
(Rotor 67)................................................................................................... 124
xv
xvi
NOMENCLATURE
A Jacobian matrix
A
b
bleed area
A
t
throttle area
a speed of sound
C
V
specific heat at constant volume
c
b1
, c
b2
constants in Spalart-Allmaras turbulence model
c
t1
, c
t2
, c
t3
, c
t4
constants in Spalart-Allmaras turbulence model
c
v1
constants in Spalart-Allmaras turbulence model
c
w1
, c
w2
, c
w3
constants in Spalart-Allmaras turbulence model
d distance to closest wall, used in Spalart-Allmaras turbulence model
E, F, G inviscid flux vectors
e internal energy per unit volume
f
t1
, f
t2
functions in Spalart-Allmaras turbulence model
f
v1
, f
v2
, f
w
functions in Spalart-Allmaras turbulence model
g, g
t
functions in Spalart-Allmaras turbulence model
I identity matrix
K
b
bleed valve constant
K
t
throttle valve constant
L
ref
characteristic length of the compressor, usually the rotor diameter
xvii
c m
.
mass flow rate through the compressor
t m
.
throttle mass flow rate
b m
.
bleed mass flow rate
p pressure
q conserved flow variables
R, S, T viscous stresses and heat fluxes at a cell face
S vorticity magnitude in Spalart-Allmaras turbulence model
modified vorticity magnitude
T temperature
T also refers to matrix containing eigenvectors of the Jacobian
matrix
t time
U relative velocity normal to a cell face
u, v, w Cartesian velocity components
V
p
plenum volume
x, y, z Cartesian coordinates
∆q change in conserved flow variables from one time step to next
∆U trip point velocity in Spalart-Allmaras turbulence model
∆t time step
γ specific heat ratio
S
~
xviii
κ von Karman constant
Λ matrix containing eigenvalues of the Jacobian matrix
µ molecular viscosity
ν eddy viscosity
ν
t
turbulent viscosity
ν
~
working variable in Spalart-Allmaras turbulence model
ρ density
σ constant in Spalart-Allmaras model
τ
ij
viscous stress tensor
ω
t
trip point wall vorticity
Subscripts
0 stagnation quantity
b bleed
p plenum
t turbulence quantity
x, y, z, t Cartesian and time derivatives
∞ free stream quantity, upstream of the inlet
xix
Superscripts
n, n+1 two adjacent time level
Overbars
~ used to indicate Roe averages
→ used to indicate vectors
÷ time or spatial averages
xx
SUMMARY
Axial compression systems are widely used in many aerodynamic applications.
However, the operability of such systems is limited at low-mass flow rates by fluid
dynamic instabilities. These instabilities lead the compressor to rotating stall or surge. In
some instances, a combination of rotating stall and surge, called modified surge, has also
been observed. Experimental and computational methods are two approaches for
investigating these adverse aerodynamic phenomena. In this study, numerical
investigations have been performed to study these phenomena, and to develop control
strategies for alleviation of rotating stall and surge.
A three-dimensional unsteady Navier-Stokes analysis capable of modeling
multistage turbomachinery components has been developed. This method uses a finite
volume approach that is third order accurate in space, and first or second order in time.
The scheme is implicit in time, permitting the use of large time steps. A one-equation
Spalart-Allmaras model is used to model the effects of turbulence. The analysis is cast in
a very general form so that a variety of configurations -centrifugal compressors and
multistage compressors- may be analyzed with minor modifications to the analysis.
Calculations have been done both at design and off-design conditions for an axial
compressor tested at NASA Glenn Research Center. At off-design conditions the
calculations show that the tip leakage flow becomes strong, and its interaction with the
tip shock leads to compressor rotating stall and modified surge. Both global variations to
xxi
the mass flow rate, associated with surge, and azimuthal variations in flow conditions
indicative of rotating stall, were observed.
It is demonstrated that these adverse phenomena may be eliminated, and stable
operation restored, by the use of bleed valves located on the diffuser walls. Two types of
controls were examined: open-loop and closed-loop. In the open-loop case mass is
removed at a fixed, preset rate from the diffuser. In the closed-loop case, the rate of
bleed is linked to pressure fluctuations upstream of the compressor face. The bleed valve
is activated when the amplitude of pressure fluctuations sensed by the probes exceeds a
certain range. Calculations show that both types of bleeding eliminate both rotating stall
and modified surge, and suppress the precursor disturbances upstream of the compressor
face. It is observed that smaller amounts of compressed air need to be removed with the
closed-loop control, as compared to open-loop control.
xxii
1
1 CHAPTER I
INTRODUCTION
“My invention consists in a compressor or pump of the turbine type
operating by the motion of sets of movable blades or vanes between sets
of fixed blades, the movable blades being more widely spaced than in my
steam turbine, and constructed with curved surfaces on the delivery side,
and set at a suitable angle to the axis of rotation. The fixed blades may
have a similar configuration and be similarly arranged on the
containing casing at any suitable angle. Parsons 1901,” taken from
Reference [1]”
In 1853 the basic fundamentals of the operations of a multistage axial
compressor were first presented to the French Academie des Sciences
2-3
. Parsons built
and patented an axial flow compressor in 1901
1
. Since that time, compressors have
significantly evolved. There have been continuous improvements leading to increases in
efficiency, the pressure ratio per stage, and a decrease in weight.
Compressors have a wide variety of applications. They are a primary component in
turbojet engines used in aerospace propulsion, industrial gas turbines that generate
power, and processors in chemical industry to pressurize gas or fluids. Compressors can
vary in size from a few feet to tens of feet in diameter. In turbomachinery applications,
safe and efficient operation of the compression system is imperative. To run a
compressor as efficiently as possible, and to prevent damage, flow instabilities such as
rotating stall and surge must be avoided, or dealt with soon after their inception.
2
Considerable interest exists in the jet propulsion community in understanding and
controlling flow instabilities.
1.1 AN OVERVIEW OF COMPRESSOR OPERATIONS
The basic purpose of a compressor is to increase the total pressure of the working
fluid using shaft work. Depending on their type, compressors increase the pressure in
different ways. They can be divided into four general groups: rotary, reciprocating,
centrifugal and axial. In rotary and reciprocating compressors, shaft work is used to
reduce the volume of gas and increase the gas pressure. In axial and centrifugal
compressors, also known as turbo-compressors, the fluid is first accelerated through
moving blades. In the next step, the high kinetic energy of the fluid is converted into
pressure by decelerating the gas in stator blade passages or in a diffuser.
In centrifugal compressors, the flow leaves the compressor in a direction
perpendicular to the rotation axis. In axial compressors, flow enters and leaves the
compressor in the axial direction. Because an axial compressor does not benefit from
the increase in radius that occurs in a centrifugal compressor, the pressure rise obtained
from a single axial stage is lower. However, compared to centrifugal compressors, axial
compressors can handle a higher mass flow rate for the same frontal area. This is one of
the reasons axial compressors have been used more in aircraft jet engines, where frontal
area plays an important role. Another advantage of axial compressors is that multi-
staging is much easier, and does not need the complex return channels required in
multiple centrifugal stages. As Ferguson
4
points out, a turbo-compressor may also be
3
called as blower, fan, booster, turbo-charger or exhauster, and the distinctions between
these are vague. Generally speaking, fans are the first stage of the compression system
in jet engines, and are low- pressure compressors. Blowers may be thought of medium
pressure compressors.
A variety of turbo-machines and their ranges of utilization in terms of basic non-
dimensional parameters are shown in Figure 1.1, taken from Reference [5]. The
horizontal axis represents the flow coefficient, which is a non-dimensional volume flow
rate. The vertical axis shows the head coefficient, which is a dimensionless measure of
the total enthalpy change through the stage, and roughly equals the work input per unit
mass flow. The literature on compressors is vast. References [5]-[9] give a basic
introduction to turbomachinery, and more advanced topics on compressors may be
found in References [10]-[12].
The compressors considered in this study are from the family of axial compressors.
An axial compressor, shown in Figure 1.2, consists of a row of rotor blades followed by
a row of stator blades. The working fluid passes through these blades without
significant change in radius. Energy is transferred to the fluid by changing its swirl, or
tangential velocity, through the stage. A schematic diagram of the changes in velocity
and fluid properties through an axial compressor stage is shown in Figure 1.2. It shows
how pressure rises through the rotor and stator passages.
Early axial compressors had entirely subsonic flow. Since modern applications
require compression systems with higher-pressure ratios and mass flow rates, designers
4
have permitted supersonic flow, particularly near the leading edge tip where the highest
total velocity occurs. Today, most high performance compression stages are transonic,
where regions of subsonic and supersonic flow both exist in the blade passages. A
transonic compression system is now one of the main components of high-bypass ratio
engines. Large fans with inlet relative Mach numbers of 1.4 to 1.6 have been recently
used in engines of this kind. These systems have been achieved by advanced design,
using sophisticated computational design tools and extensive experimentation.
The steady state performance of a compressor is usually described by a plot of the
averaged mass flow rate versus the total pressure ratio. This plot is called the
characteristic or performance map of the compressor. Figure 1.3 shows a typical
compressor performance map for axial and centrifugal compressors. Axial compressors
tend to have a steeper drop aft of the peak of the compressor performance map compared
to centrifugal compressors.
1.2 COMPRESSOR STABILITY
Stability in a compressor is the ability of a compressor to recover from disturbances
that alter the compressor operation about an operational equilibrium point. Disturbances
may be considered as transient or deliberate changes to the operating point. In the case
of transient disturbances, the system is stable if it returns to its original operating point.
If the disturbances drive the compressor away from the original point, the system is
unstable. The steady state match between a compressor and its drive turbine or jet
nozzle, which is perturbed by a transient change of mass-flow, is a good example of this
5
case. When there are deliberate changes to the operating point, the performance is
considered stable if a new operational equilibrium point can be achieved, e.g., shifting
the operating point by changing the compressor shaft speed. If steady state operation at
a new operating point is not possible, the system is unstable.
Stability in compressors may be studied from two different perspectives. The first
is called operational stability, which deals with the matching of compressor performance
with a downstream flow device such as a turbine or throttle. The second is aerodynamic
stability, which deals with deteriorations in the operation due to flow separation, stall or
surge.
The operational stability of a compression system depends on the characteristic of
both the compressor and the downstream flow device. Mathematically, if the slope of
compressor performance map is less than the slope of characteristic map of the throttle
(points P
1
and P
2
shown in Figure 1.4a) the system is stable. Otherwise, as shown in
Figure 1.4b for point P
3
, the system is not stable. Compressors, by design, usually
operate near point P
1
on the performance map shown in Figure 1.4. Operations at lower
mass flow ratios (near point P
2
) can trigger instabilities as discussed later.
The stable range of operation of axial and centrifugal compressors is limited at both
very high and very low mass flow rates, as shown in Figure 1.5. If the mass flow rate is
too high, shocks will form and the flow through the compressor will be choked (sonic
condition). On the other hand, as the mass flow rate through the compressor decreases,
flow instabilities will occur. These instabilities include rotating stall and surge. If they
6
are allowed to persist or grow, catastrophic damage to the compressor and the engine
will occur. Surge, in particular, is to be avoided at all costs.
In looking at a map of the characteristic performance of a compressor, Figure
1.5, a dashed line known as the surge or stall line, can be seen. Rotating stall and surge
usually occur at low flow rates, but may still occur on the right side of the surge line if
the flow becomes unstable as a result of the instability. Therefore, a second line parallel
to the surge line is usually introduced as a surge avoidance line. Another reason for
introducing the surge avoidance line is that the compressor characteristic, and
consequently the surge line, may be poorly known. Operating at the surge avoidance
line provides a safety margin for the compressor operation and prevents the compressor
from operating in a region where stall or surge may occur. The closer the operating
point is to the surge line, the greater the pressure ratio achieved by the compressor, but
the greater the risk of stall or surge.
1.3 OBJECTIVES AND ORGANIZATION OF THE PRESENT WORK
The main objectives of the current study are to understand the physics of
compressor stall and surge, and to develop an appropriate control methodology for the
prevention of these instabilities.
Although considerable progress in understanding and modeling of stall and surge in
axial compressors has been achieved during the past two decades, none of the models are
able to describe accurately the flow phenomena that occur in the compressor and give rise
7
to stall and surge. Detailed flow visualizations, both computational and experimental, are
necessary for understanding the nature of these instabilities.
CFD modeling of axial compressors is a well-developed field. Most compression
systems are now being designed using CFD tools. However, most numerical studies of
air breathing compression systems are done in the stable part of compressor characteristic
performance map, where the flow is “steady” in a rotating frame. The current research
attempts to provide computational tools to study unsteady aerodynamic phenomena, such
as rotating stall and surge, in axial compressors. Work has also been done in simulating
stall and designing stall control methods that extend the stable operating range of the
compressor.
This thesis is organized as follows: A review of surge and rotating stall
phenomena, both from a historical and a technical perspective, is presented in Chapter
II. Chapter III introduces the mathematical and numerical tools required for carrying out
the numerical simulations. Validation results for an axial compressor at peak efficiency
conditions are presented in Chapter IV. Simulations of the onset and growth of stall are
given in Chapter V. Results for active bleed control techniques are presented in Chapter
VI. Finally, the conclusions and recommendations for further improvements of
compressor control technology are given in Chapter VII.
8
Figure 1.1 Work input machinery classification, Reference [5].
9
Figure 1.2 Schematic diagram of changes in fluid properties and velocity through an
axial compressor stage, References [5], [11].
Rotor
Stator
Static Pressure
Absolute Velocity
Total Enthalpy
Total Pressure
1 2 3
10
Figure 1.3 Typical compressor characteristic map for axial and centrifugal
compressors.
Figure 1.4 Operational stability, matching the compressor and throttle
characteristics.
Compressor Mass Flow
Pressure
Rise
Pressure
Rise
P
1
Mass Flow Rate
a) Stable
P
2
Mass Flow Rate
b) Unstable
P
3
Throttle
Characteristic
Compressor
Performance
11
Figure 1.5 Effects of rotor RPM on compressor performance and stability.
Compressor Mass Flow
Pressure
Rise
Margin of Safety
Stall Line
Constant Rotor Speed Line
Surge Avoidance Line
Operating Point
RPM increases
12
2 CHAPTER II
AN OVERVIEW OF COMPRESSOR INSTABILITY PHENOMENA
In the pervious chapter, it was pointed out that the stable part of compressor
performance map is limited due to aerodynamic instabilities. These instabilities manifest
themselves as rotating stall or surge. This chapter is devoted to reviewing these unsteady
phenomena from both a historical and a technical perspective. Section 2.1 describes the
physical mechanisms behind rotating stall. A discussion of surge phenomenon is given in
Section 2.2. In Section 2.3, the historical developments that led to the discovery of
rotating stall and surge phenomena are presented. A brief review of methods currently
employed in compressor control, and a review of prior computational study of
compressors and compressor control are also given in Section 2.3.
2.1 FUNDAMENTALS OF ROTATING STALL
During the normal operation of a compressor, the airflow through the compressor
is essentially steady and axisymmetric in a rotating coordinate system. If a flow
instability is somehow introduced into the system (say, due to a change in the rotor speed,
flow separation at the inlet, or other type of flow distortion), instabilities may develop
and the compressor performance may deteriorate. The instability manifests itself as
either a rotating stall or surge. Rotating stall is inherently a 2-D unsteady local
13
phenomenon in which the flow is no longer uniform in the azimuthal direction. It often
takes only a few seconds for rotating stall to build up, and the compressor can operate
under rotating stall for several minutes before damage develops. Rotating stall can occur
in both compressible and incompressible flow.
The inception of rotating stall is shown in Figure 2.1. This figure illustrates the
blade row viewed from the top of the annulus. Stall is present on some of the blades. It
is not known for certain why all blades do not stall at the same time. Dimensional
tolerances could be one possible case, Reference [12]. In manufacturing and assembly, a
few blades could be produced with slightly different profiles or with higher stagger
angles. These imperfections would cause the inlet air to see these blades at slightly
different angles of attack as compared to the other blades. When one of the blades stalls,
as a consequence of some instability, the angle of the flow relative to the shaft increases.
This increase in flow angle, in addition to blockage attributed to stall, cause part of the
oncoming flow to be diverted towards the neighboring blades, thus causing an increase in
their angles of attack and leading them to stall. As the blade rotates away from the
disturbances, the angle of attack decreases, restoring normal flow over that blade. The
region of the stalled flow, known as a stall cell, continues moving from blade to blade
and propagates around the annulus.
In a coordinate system attached to the blades, rotating stall moves in a direction
opposite to the blade motion at a fraction of the rotor speed. However, in the inertial
coordinate system, the stall region propagates in the same direction as the wheel motion.
14
The reported rotational speed of rotating stall around the annulus of compressor varies
from 20 to 75 percent of the rotor speed in the direction of the rotor motion
13
. It has also
been reported that the incipient rotating stall cells move faster. Typical frequencies for
rotating stall are 10 to 50 times larger than those for surge.
The number of stall cells depends on the compressor at hand; one to nine stalled cells
has been reported. Two types of stall associated with the number of stalled cells exist,
progressive and abrupt. In progressive stall, a phenomenon involving multiple stalled
cells, the pressure ratio after stall reduces gradually. Abrupt stall results in a sudden drop
in total-to-total pressure rise, and appears to always involve a single stalled cell.
One of the characteristics of pure rotating stall is that the average flow is steady with
respect to time, but the flow has a circumferentially non-uniform mass deficit. During
rotating stall, the cyclical variation of the pressures on the blades can cause them to
fatigue and eventually break. The flow temperature may also increase due to uneven
distribution of shaft work, reducing blade life. A typical plot of the static pressure history
measurements taken at a fixed circumferential location at the inlet of an axial compressor
under rotating stall conditions is depicted in Figure 2.2
14
.
Several types of rotating stall exist
13
:
• Part-Span: As illustrated in Figure 2.3
12
, only a restricted region of the blade
passage, usually the tip, is stalled. Stall near the root has also been reported.
• Full-Span: The entire height of the annulus is stalled. Figure 2.4
12
shows the
full-span rotating stall with various stalled cells.
15
• Small/Large scale: In this case, a small/large part of annular flow path is
blocked.
Figure 2.5 shows a typical rotating stall pattern. When rotating stall occurs at point
A on the unstalled branch, the operating point then proceeds to the so-called stalled
characteristic at point B, along a straight line AB. If point B is stable, the compressor
will remain and operate there, until measures are taken to bring it back to the unstalled
branch. Sometimes the deterioration in the performance of an axial compressor with
rotating stall is small, and may not be easily detected except as an increase in the
compressor noise or by some high-frequency sensors. Recovery from rotating stall is
often more difficult than surge
13
. Rotating stall can also serve as the precursor to the
more severe and dangerous flow stability, called surge.
2.2 FUNDAMENTALS OF SURGE
Surge is a global 1-D instability that can affect the whole compression system.
Surge is characterized by large amplitude limit cycle oscillations in mass flow rate, and
pressure rise. Even a complete reversal of the flow is possible. The behavior of surge
depends on both the compressor characteristic and the characteristics of the diffuser
13
.
Figure 2.6 shows a typical plot of the plenum pressure transient response when an axial
compressor experiences surge
14
.
In contrast to rotating stall, the average flow through the compressor is unsteady but
the flow is circumferentially uniform. Many of the conditions that a compression system
experiences during rotating stall are also present in surge. The rotor blades are stressed
16
by the oscillating flow and the uneven distribution of shaft work; backpressure decreases
while the inlet pressure increases. The compressor's noise characteristic changes and
pressure fluctuations occur through out the compressor.
In high-speed compressors, the reversal flow can be triggered by a shock wave
3
. The
high pressures behind the shock may deform the casing and inlet, and resulting pitching
moments can also change the twist of the rotor/stator blades. In low-speed compressors,
the surge appears as a moderate pulsing of the flow.
Based on flow and pressure fluctuations, surge can be categorized into four different
classes
13
:
• Mild Surge: No flow reversal; small periodic pressure fluctuations governed
by the Helmholtz resonance frequency.
• Classic Surge: No flow reversal; larger oscillations at a lower frequency.
• Modified Surge: Combination of classic surge and rotating stall; entire
annulus flow fluctuates in axial direction; non-axisymmetric flow.
• Deep Surge: Strong version of classic surge; possibility of flow reversal;
axisymmetric flow.
In both axial and centrifugal compressors, while increasing the plenum pressure at
the compressor exit at a constant rotor speed, a mild surge can occur. The mild surge
may be followed by rotating stall or modified surge. A classic or a deep surge may then
follow.
17
2.3 LITERATURE SURVEY OF STUDIES ON
ROTATING STALL AND SURGE
The occurrence of the fluid dynamic instabilities was first considered one of the
normal operating features of axial compressors. Qualitative explanations of surge have
long been known. Stodola
15
and Kearton
16
described the surge phenomenon in 1927 and
1931, respectively. To this author’s knowledge, rotating stall was first detected in 1932
in a centrifugal pump impeller
17
. In 1946, it was found that the ducts linked to the
compressors could be primary contributors to surge, Reference [12].
Up to this time, researchers assumed that the surge phenomenon was sinusoidal with
respect to time, and was associated with Helmholtz resonances. In 1955, an extensive
study of rotating stall and surge was presented by Emmons, Pearson and Grant
18
. They
predicted the rotational velocity of the stall cell using linear flow theory. Rotating stall
and surge are nonlinear phenomena, and this linearized theory limited them to analyzing
weak disturbances to the flow. In 1958, Horlock
1
published a book about these two types
of flow instabilities, and detailed the state of the art about these phenomena in axial
compressors. One of the early attempts to model the one-dimensional flow through the
axial compressors in a nonlinear form was by Greitzer
19
in 1976. The first nonlinear
analysis of incipient and fully-developed rotating stall was presented by Moore
20
. The
extensions to the inlet distortion effects may be found in References [21] and [22]. In
1986, Moore and Greitzer
23-24
presented the theory of post-stall transients in axial
compressor.
18
Much of the work on rotating stall and surge control in the literature was based on
the assumption of incompressible flow. Since the beginning of 1990’s, the importance of
compressibility effects has been recognized and studied. The effects of compressibility
on surge can be found in References [25] and [26]. They are based on one-dimensional
fluid dynamic equations, and therefore are limited to the surge phenomenon.
Incorporation of compressibility effects in two-dimensional fluid dynamic models was
presented in Reference [27]. Further details on nonlinear control methods may be found
in References [28] through [30]. An exhaustive survey of rotating stall and surge control
can be found in Reference [13].
2.3.1 EXPERIMENTAL STUDIES ON COMPRESSOR CONTROL
Surge and rotating stall are highly undesirable phenomena. They can introduce
mechanical and thermal loads, and can even cause structural damage. These
aerodynamic instabilities reduce the total to total pressure rise and efficiency of the
compression system. Unrecoverable stall in gas turbines requires restarting the engine
and may also have catastrophic consequences in aircraft jet engines. These instabilities
may be avoided by operating away from the surge line. On the other hand, due to the
high performance and efficiency obtained near the surge line, it is desirable to operate the
compressor close to the surge line. To overcome this dilemma, three different
approaches exist: surge/stall avoidance, surge detection and avoidance, and increasing the
stall margin approach.
19
Surge avoidance techniques are known and have been used for a long time in
industry and commercial systems. In this approach, the control systems do not allow the
compressor to operate on the left side of the surge avoidance line. To locate the surge
avoidance line on the compressor map, a safety margin should be specified. This safety
margin may be defined based on pressure ratio, corrected mass flow, or a combination of
pressure ratio and corrected mass flow. A common safety margin, SM, is based on total
pressure ratio and is defined as:
Avoidance Surge
o
Avoidance Surge
o
Surge
o
P
P
P
P
P
P
SM

,
_

¸
¸

,
_

¸
¸

,
_

¸
¸
·
01
2
01
2
01
2
(2.1)
Here P
02,
and P
01,
are total pressure at the compressor exit and inlet, respectively. In a
multistage axial compressor in a turbojet, it is typical to have a safety margin as high as
25 percent
9
, while for a centrifugal pipeline compressor the safety margin is 10 percent
31
.
In surge detection and avoidance methods, the onset of instabilities is first detected.
The most successful techniques to detect the onset of stall are based on monitoring the
pressure and temperature variations or other parameters (e.g. their time derivatives and
oscillation frequency) at the compressor inlet or exit. These measurements are compared
to the expected values at the surge condition, stored in the control computer. When surge
or stall is detected, corrective measures (e.g. bleed) are applied. The advantage of this
20
technique is that it is not necessary to define a large safety margin, and the compressor
can operate close to the surge line. The disadvantages of this technique are the need for
large control forces and a very fast-acting control system that will prevent the growth of
instabilities into surge. Another weakness of this technique is that it is highly dependent
on the compressor being controlled, since different compressors exhibit different
behaviors during the onset of surge.
The third control methodology involves increasing the stall margin. This approach
may be divided into two different classes: passive surge/stall control and active
surge/stall control.
In both active and passive control the characteristic performance map of the
compressor is modified and the surge line is shifted to a lower mass flow. By shifting the
surge line, the surge avoidance line is also shifted. In other words, some part of the
unstable area in the performance map is being stabilized by this approach. An advantage
of this methodology is that the compressor now can operate near peak efficiency and
high-pressure ratios at lower mass flow rates.
In passive surge/stall, the geometry of the compressor is altered to modify the stall
margin. Casing treatments
32-35
and variable guide vanes
36-39
are some the different ways
of achieving passive surge/stall control.
Casing treatments, which have been investigated more in axial compressors, the rotor
casing is designed so that the amount of blockage in a flow passage is decreased. Thus,
rotating stall is suppressed. In this method, the casing is designed with various shaped
21
grooves: perforated, honeycomb, circumferentially grooved, axial slotted, or blade angle
slotted. The effects of a porous casing on stall margin for both uniform and distorted
inlet conditions may be found in Reference [32].
The use of variable inlet guide vanes is another way of increasing the stall margin
and has been used in both axial and centrifugal compressors. In this method, the incident
angle in compressors at lower mass flow rates is reduced and the leading edge separation
is prevented. With inlet guide vanes, the direction of the flow at the leading edge is
turned such that the angle of attack decreases. Variable inlet guide vanes are also
commonly used when starting and accelerating engines to avoid crossing the surge line.
In active stall or surge control, the compressor is equipped with devices such as a
bleed valve that can be switched on or off. Generally, active surge/stall control may be
divided into two classes: open-loop and closed-loop. In closed-loop control, a feedback
law is used to activate the controller, while in the open-loop control no feedback signals
are used. Air injection
40
, bleeding
40-42
, and recirculation (a combination of injection and
bleeding) are examples of active surge/stall control.
Air injection is another way of increasing the stall margin and has been used in both
axial and centrifugal compressors. In this method, a small amount of high pressure and
high velocity air is injected into the compressor upstream of the compressor face. As a
result, the flow is energized and the axial velocity component is increased. This reduces
the local angles of attack, and the leading edge separation is prevented. The injected air
22
may be supplied from the diffuser downstream of the compressor or from a separate
device.
One of the oldest, and the most investigated approach for increasing the stall margin
in axial compressors is bleeding. This technique has been used in both axial and
centrifugal compressors. Since the early days of jet engines, bleeding has been the most
common approach for avoiding surge during engine acceleration and start-up. Beside the
start-up applications, bleeding has also been used to achieve a wide range of operating
conditions. The bleed valve is typically located either in the plenum exit or downstream
of rotor on the shroud. The concept of bleeding is that at lower mass flow the working
fluid does not have enough momentum to overcome the viscous and adverse pressure
gradient forces in the plenum. By removing some of the highly pressurized flow
downstream of the compressor, flow acceleration can increase and surge-free operation is
achieved.
Closed-loop active control was first reported by Epstein et al.
43
in 1989, and the
literature on this approach has become extensive over the last decade. This method
promises to be an integral part of future engines, the so-called smart or intelligent
engines. The closed-loop control devices use a sensor for detecting the growth of
instabilities (precursor waves) when a compressor experiences stall conditions. In this
method, a control unit processes measured flowfield data, such as temperature, pressure
or axial velocity, from a stall-detection device. The stall-detection devices are usually
located on the circumference of the compressor casing, either upstream or downstream of
23
compressor. A feedback law connecting the sensed fluctuations to the rate of bleed is
used to stabilize the compressor. The control unit activates a set of actuator devices.
There are several types of actuators in use for stabilizing the compression system.
Among these, bleed valve actuators have been the most commonly used. Pinsely et al.
42
,
studied the use of the throttle valve as actuator in a centrifugal compressor surge control,
and achieved a 25% reduction in the mass flow rate. Bleed valve control is discussed in
References [40]-[42].
Other types of actuators include variable inlet guide vanes, re-circulation,
loudspeakers, movable plenum walls, and air injections. Figure 2.7 shows a schematic of
these passive and active control methods.
2.3.2 COMPUTATIONAL STUDIES OF COMPRESSOR PERFORMANCE
AND CONTROL
As stated in Section 2.1, many attempts have been made to increase the operating
range of the compression system using appropriate detection and control devices. Over
the past five decades, considerable research has been done on both axial (References
[21]-[24] & [44]-[46]), and centrifugal compressors (References [47]-[49]). One goal of
these studies is the prediction of component performance, e.g. pressure ratio and
efficiency. It is obvious that even a small improvement in the efficiency of a commercial
aircraft engine can result in huge saving in yearly fuel costs. Therefore, turbomachinery
designers are extremely interested in tools that give good qualitative and quantitative
predictions of turbomachinery performance, and may be used in aerodynamic design.
24
One efficient method for investigating complex flow phenomena in rotating
machinery is computational fluid dynamics. The earliest numerical study of compressor
instabilities, to this author’s knowledge, is the work by Takata and Nagano
50
in 1972.
They used a finite difference method to solve the nonlinear flow equations. The flow was
treated as incompressible, and Laplace and Poisson equations were used to model the
flow within the inlet and exit ducts, respectively.
During the past decade computational fluid dynamics has undergone an impressive
evolution, providing engineers and designers with the capability to model and study 3-D
unsteady flows. There are several reasons for the scarcity of CFD-based turbomachinery
performance calculations in the early literature. One reason is that pressure field
calculations are relatively independent of viscous effects and can be obtained with simple
models. On the other hand, system losses and efficiency are strongly dependent on
viscous effects and require careful attention to the viscous terms, artificial viscosity,
turbulence modeling, and grid resolution to obtain satisfactory results. A second reason
for the scarcity of CFD studies is the inability of early computers to perform large
viscous flow calculations. A third reason is that validation data, such as detailed pressure
and velocity measurements, are difficult to obtain in turbomachines, due to the small size
and the high speeds of the components involved
62
.
In recent literature, several researchers have presented detailed investigations of
turbomachinery performance using CFD. However, most of their calculations were
extracted from 2-D codes. For example, Davis et al.
51
predicted loss buckets for
25
transonic compressor cascades in 1986; loss and exit flow angle were calculated for
turbine and fan cascades by Chevrin and Vuillez
52
using the 2-D code of Cambier et al.
53
;
and the effects of turbulence modeling on turbine blades were studied by Boyle
54
.
Thanks to the massive increase in computing power and the development of
sophisticated post-processing and visualization tools, time-accurate 3-D simulations of
rotary machines are now possible. Furthermore, with the availability of an AGARD
Advisory Report for computational test cases of internal flow, researchers now have
access to excellent data for validation of the turbomachinery codes. A number of 3-D
CFD codes for detailed modeling of turbomachinery flow fields exist. Srivastava and
Sankar
55
, Dawes
56
, Hah et al.
57
, Adamczyk et al.
58
, Hall
59
, Hathaway et al.
60
, Wood at
al.
61
, and Chima et al.
62
, among others, have developed 3-D codes that have the capability
to analyze unsteady turbomachinery flow with multiple blade passages and/or rotor-stator
interactions. However, most of these applications have been applied only to steady-state
phenomena in axial and radial compressors. There has been little or no effort in
numerically modeling and studying off-design conditions by CFD methods using the full
Navier-Stokes equations. Many researchers have also simulated rotating stall and surge
in axial compressors using simple 1-D and 2-D codes, e.g. References [63]-[65].
In an effort to model unsteady flow within a compression system, a 3-D
compressible unsteady flow solver for turbomachinery has been developed at Georgia
Institute of Technology by Niazi, Stein and Sankar, References [66]-[70]. This solver is
capable of solving multiple flow passages in an inertial reference frame, and can be
26
extended to multistage compression systems that are currently used in gas turbines. In a
previous study, this code has been applied to a NASA low speed centrifugal compressor
(LSCC) configuration at both design and off-design conditions
66-67
. The CFD
simulations captured the onset of surge within the compression system. Two different
active control schemes, a diffuser bleed valve and air injection, were implemented. It
was shown that these control schemes could suppress the stall and significantly extend
the useful operating range of the compressor
66-67
. Stein
70
simulated surge with both
steady and pulsed air injection control in centrifugal compressors using this three-
dimensional time accurate compressible flow solver, while considering one single flow
passage. To the knowledge of this author, this is the first time such flow control
simulations based on the full set of 3-D Navier-Stokes equations have been done.
27
Figure 2.1 Rotating stall inception.
Figure 2.2 Transient response of system in rotating stall, Reference [14].
28
Figure 2.3 Part-Span rotating stall with different stalled cells, Reference [12].
Figure 2.4 Full-Span rotating stall with different stalled cells, Reference [12].
29
Pressure
Mass Flow Rate
Stalled
Unstalled
A
B
Figure 2.5 Compressor map with the stalled flow characteristic, Reference [13].
Figure 2.6 Transient response of system in surge, Reference [14].
30
Figure 2.7 Types of active and passive compressor control schemes.
Air Injection
Movable Plenum Walls
Guide Vanes
Bleed Valves
Pressure
Sensors
Controller
Bleed Valve
Air
Injection
Recirculation Control
31
3 CHAPTER III
MATHEMATICAL AND NUMERICAL FORMULATION
In order to analyze flow details in compressors, solution of the 3-D Navier-Stokes
equations is required. The complex nature of the governing equations limits the
analytical solutions to simple flows and configurations. Therefore, numerical techniques
are required for more complex problems. In this chapter the mathematical formulation
and numerical tools employed in this study are documented. A comprehensive overview
of CFD methods may be found in References [71] and [72].
In Section 3.1 unsteady compressible flow equations are presented. The numerical
discretization process and an approximate factorization solution algorithm used in the
flow solver are given in Sections 3.2 and 3.3, respectively. The turbulence modeling
method implemented in this work is discussed in Section 3.4. In Section 3.5 the initial
and boundary conditions are described.
3.1 GOVERNING EQUATIONS
The system of partial differential equations for the conservation of mass, momentum
and energy in fluid flow are known as the Navier-Stokes equations. These equations are
derived from first-principles and from thermodynamic considerations. The Navier-
Stokes equations describe the physics of 3-D, unsteady compressible viscous flow,
32
subject to some stress-strain rate relationships. In this study, calorically perfect
Newtonian fluids, obeying the Stokes linear stress-strain rate law, have been considered.
In three-dimensional Cartesian coordinates, the conservative form of the equations in
vector form is given below:
z
T
y
S
x
R
z
G
y
F
x
E
t
q


+


+


·


+


+


+


(3-1)
Here, q is the state vector with unknown flow variables: density ρ, velocity components
in x, y, z direction (u, v, w, respectively) and total energy, E
t
. The quantities E, F, G are
inviscid flux terms, and R, S, T represent viscous terms. The state vector and inviscid
flux terms are:
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
·
t
E
w
v
u
q
ρ
ρ
ρ
ρ
,
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+
+
·
u p E
uw
uv
p u
u
E
t
) (
2
ρ
ρ
ρ
ρ
,
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+
+ ·
v p E
vw
p v
uv
v
F
t
) (
2
ρ
ρ
ρ
ρ
,
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+
+
·
w p E
p w
uw
uw
w
G
t
) (
2
ρ
ρ
ρ
ρ
(3-2)
Furthermore,
1
]
1

¸

+ + + · ) (
2
1
2 2 2
w v u T C E
V t
ρ (3-3)
33
Here, C
V
is the specific heat at constant volume and T is the temperature. Also, p is
pressure, and is related to total energy and velocity as follows:
RT p ρ · (3-4)
1
]
1

¸

+ + − − · ) (
2
1
) 1 (
2 2 2
w v u E p
t
ρ γ (3-5)
In Equation (3-5), γ is the specific heat ratio, and since the working fluid is air, a value of
1.4 is used. The viscous terms are:
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+ + +
·
x xz xy xx
xz
xy
xx
q w v u
R
τ τ τ
τ
τ
τ
0
,
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+ + +
·
y yz yy y
yz
yy
yx
q w v x u
S
τ τ τ
τ
τ
τ
0
,
¹
¹
¹
¹
¹
¹
¹
;
¹
¹
¹
¹
¹
¹
¹
¹
'
¹
+ + +
·
z zz zy zx
zz
zy
zx
q w v u
T
τ τ τ
τ
τ
τ
0
(3-6)
where
34

z z y x zz
y z zy yz
y z y x yy
x z zx xz
x y yx xy
x z y x xx
u w v u
w v
u w v u
w u
v u
u w v u
µ λ τ
µ τ τ
µ λ τ
µ τ τ
µ τ τ
µ λ τ
2 ) (
) (
2 ) (
) (
) (
2 ) (
+ + + ·
+ · ·
+ + + ·
+ · ·
+ · ·
+ + + ·
(3-7)
and

z
T
k q
y
T
k q
x
T
k q
z
y
x


− ·


− ·


− ·
(3-8)
In these equations, µ is the molecular viscosity and k is the thermal heat conduction
coefficient of the fluid. In Equation (3-7), λ , from the Stokes hypothesis, is µ
3
2
− .
All quantities in the Navier-Stokes equations have been non-dimensionalized by
their corresponding reference values. The following reference parameters have been used
in this work:

γ
ρ
ρ ρ ρ
2
inlet the of upstream ambient,
inlet the of upstream ambient,
,
ref ref
ref
ref
ref
ref
V
p
a a V
edge trailing blade the at rotor the of Diameter L
·
· ·
· ·
·


(3-9)
35
3.2 DISCRETIZATION OF THE GOVERNING EQUATIONS
Analytical solution of the Navier-Stokes equations is limited to simple geometries.
Therefore, for complex geometry and flows with highly non-linearity, numerical
techniques must be used to find approximate solutions. In numerical methods, solutions
are found for discrete points at different time levels. Several techniques, such as finite
difference methods, finite volume methods, and finite element methods, exist for
numerically solving the Navier-Stokes equations. The finite volume method is commonly
used in fluid dynamic problems. One of the advantages of this method is that
discontinuous phenomena, such as shock waves, can be handled. In the finite volume
method in three-dimensional space, the flow is divided into a finite number of hexagonal
cells.
In this study, the 3-D unsteady compressible Reynolds-averaged Navier-Stokes
equations are recast in integral form, and solved using a finite volume scheme.
The integral form of Equation (3-1) is:
∫∫∫ ∫∫ ∫∫ ∫∫
⋅ + + · ⋅ − ⋅ + + +


V S S
G
S
ds n k T j S i R dS n V q dS n k G j F i E dV
t
q r
r r r
r
r
r
r r r
) ( ) (
(3-10)
Here, V and S refer to the control volume and control surface area respectively, and
n
r
represents the outward normal vector to surface S. The term
G
V
r
refers to the velocity
36
of the surface S. In Equation (3-10), the state vector q is evaluated at the cell vertices, as
shown in Figure 3.1, and the surface integrals are computed at the six faces surrounding
the control volume:
[ ]
1
]
1

¸

+
+
1
]
1

¸

+ +
1
]
1

¸

+ ·
∆ ⋅ − ∆ + + · ⋅ − ⋅ + +
− +
− + − +

∫∫ ∫∫
2
1
, ,
2
1
, ,
,
2
1
, ,
2
1
, , ,
2
1
, ,
2
1
) ( ) (
k j i k j i
k j i k j i k j i k j i
G
Faces All
z y x
s S
G
G G
F F E E
S n V q S Gn Fn En dS n V q dS n k G j F i E
) )
) ) ) )
r
r
r
r
r
r r r
(3-11)
where [ ]
k j i
S n V q Gn Fn En E
G z y x
k j i
, ,
2
1
) (
, ,
2
1
t
∆ ⋅ − + + ·
t
r
r )
[ ]
k j i
S n V q Gn Fn En F
G z y x
k j i
,
2
1
,
) (
,
2
1
,
t
∆ ⋅ − + + ·
t
r
r )
(3-12)
[ ]
2
1
, ,
) (
2
1
, ,
t
∆ ⋅ − + + ·
t
k j i
S n V q Gn Fn En G
G z y x
k j i
r
r )
Viscous fluxes, R, S, and T, likewise are handled as follows:
S Tn Sn Rn dS n k T j S i R
Faces All
z y x
s
∆ + + · ⋅ + +

∫∫
) ( ) (
r
r r r
(3-13)
37
The inviscid fluxes, E
)
, F
)
, and G
)
are calculated implicitly using Roe's flux
difference scheme
73-78
as described by Liu and Vinokur
79
. At a cell interface, the
numerical flux f
Num
(which is an approximation to E
)
, F
)
, or G
)
) is given by:
[ ] [ ] S q q q q A q f q f f
L R R L R L Num

1
1
1
]
1

¸

− − + ·
4 4 4 4 8 4 4 4 4 7 6 4 4 4 8 4 4 4 7 6
Term Viscosity Artificial Term Flux Physical
) ( ) , (
~
2
1
) ( ) (
2
1
(3-14)
where
1
1
1
1
1
1
]
1

¸


+
+
+
·
1
1
1
1
1
1
]
1

¸


+
+
+
·
t R R R
z R R R R
y R R R R
x R R R R
R R
R
t L L L
z L L L L
y L L L L
x L L L L
L L
L
n p H U
n p w U
n p v U
n p u U
U
q f
n p H U
n p w U
n p v U
n p u U
U
q f
0 0
) ( , ) (
ρ
ρ
ρ
ρ
ρ
ρ
ρ
ρ
(3-15)
and ∆S is the cell area of the face where f
Num
is calculated. The quantity A
~
is defined as:
1
~

Λ · T T A , assuming that the original matrix is given by
1 −
Λ ·


· T T
q
f
A . The matrix
T contains the eigenvectors of A, which Λ contains the eigenvalues. These quantities are
in general complex number, but are real numbers for Euler equations. In Equation (3-
14),
L
q and
R
q are the primitive variables vectors to the left and the right of the cell
interfaces and are defined as:
38
1
1
1
1
1
1
]
1

¸

·
1
1
1
1
1
1
]
1

¸

·
R
R
R
R
R
R
L
L
L
L
L
L
p
w
v
u
q
p
w
v
u
q
ρ ρ
, (3-16)
The quantities, U and H
0
are the relative velocity normal to the cell face, and total
enthalpy, respectively and are defined as:
n V V U
G
r
r r
⋅ − · ) ( (3-17)
p E H
t
+ ·
0
(3-18)
and
n V n
G t
r
r
⋅ − · (3-19)
Again,
G
V
r
is the grid velocity at the face where f
Num
is being calculated, and
n
r

is the unit
normal vector to this face
.
A four-point stencil, with Roe’s Superbee limiter
78
to control the high frequency
oscillations near shocks, was used to compute the left and right values of primitive
variable vectors at each cell interface. For example,
L
q and
R
q at cell face
2
1
+ i , as
shown in Figure 3.2, can be written as:
39
) (
6
1
) (
3
1
) (
3
1
) (
6
1
1 2
2
3 1
2
1 1
1
2
1 1
2
1
+ +

+
+
+
+
+
+

+

+

− Φ − − Φ − ·
− Φ + − Φ + ·
i i
i
i i
i
i R
i i
i
i i
i
i L
q q q q q q
q q q q q q
(3-20)
and
[ ] ) 2 , min( ), 1 , 2 min( , 0 max ) (
,
) ( , ) (
1
1
2
1
1
1
2
1
2
1
2
1
2
1
2
1
r r r
q q
q q
r
q q
q q
r
r r
i i
i i
i
i i
i i
i
i i i i
· Φ


·


·
Φ · Φ Φ · Φ
+
− −
+

+ +


+

+
+

+

(3-21)
The second term in Equation (3-14), using Liu and Vinokur
79
formula, can be
expanded as:
n L R R L
N q q q q q q A
2
*
1 1
~
~
) ( ) , (
~
δ δ λ + + ∆ · − (3-22)
where
a
U
a
p
~
~
2
~ ~
~
2
~ ~
~ 3 2
2
3 2
1 1


+

,
_

¸
¸
+
+ − ·
ρ
λ λ λ λ
λ δ (3-23)
40
a
p
U
~
2
~ ~
~
2
~ ~
~ 3 2 3 2
1 2


+ ∆

,
_

¸
¸
+
+ − ·
λ λ
ρ
λ λ
λ δ (3-24)
S
U
n
n
n
N
H
w
v
u
q
z
y
x
n

1
1
1
1
1
1
1
]
1

¸

·
1
1
1
1
1
1
1
1
1
]
1

¸

·
~
0
,
~
~
~
~
~
1
~
0
*
ρ
(3-25)
and
a U
a U
U
~
~ ~
~
~ ~
~ ~
3
2
1
− ·
+ ·
·
λ
λ
λ
(3-26)
In Equations (3-22) through (3-26), the quantities marked by “~” are calculated using
Roe averaging, defined as:
R L
ρ ρ ρ ·
~
(3-27)
R L
R R L L
u u
u
ρ ρ
ρ ρ
+
+
·
~
(3-28)
41
R L
R R L L
v v
v
ρ ρ
ρ ρ
+
+
·
~
(3-29)
R L
R R L L
w w
w
ρ ρ
ρ ρ
+
+
·
~
(3-30)
R L
R R L L
H H
H
ρ ρ
ρ ρ
+
+
·
~
(3-31)
n V n k w j v i u U
G
r
r
r
r r r
⋅ − ⋅ + + · )
~ ~ ~
(
~
(3-32)
The difference, ∆, in Equations (3-22) through (3-24) represents the change in the
quantities across the cell interface, for example:
L R
p p p − · ∆ (3-33)
3.3 NUMERICAL SOLUTION OF THE DISCRETIZED EQUATIONS
The 3-D unsteady compressible Reynolds averaged Navier-Stokes equations are
solved numerically using a time marching scheme. This involves solving the governing
equations at each time step by marching in time from an initial flow condition with
appropriate boundary conditions, discussed later in Section 3.5.
As seen earlier, the discretized form of the governing equation is:
42
( )


∆ + + −
∆ − ·

∆ − ∆
+
Faces All
z y x
Faces All
Num
n
k j i
n
k j i
S Tn Sn Rn
S f
t
V q V q
, ,
1
, ,
) ( ) (
(3-34)
This equation set is highly nonlinear, and links the finite volume cell (i, j, k) to its
neighbors (i+1, j, k), (i+2,j,k) etc. Direct numerical solution of this equation set is not
feasible. A non-iterative scheme, which may be viewed as the first iteration scheme, was
used.
All terms in Equation (3-34) were first expanded about known flow quantities at the
time level “n”:
∑ ∑ ∑
∆ −


+

,
_

¸
¸
∆ ·

,
_

¸
¸

+
+
Faces All
n n Num
n
Faces All
Num
n
Faces All
Num
S q q
q
f
S f S f ) ( )
1
1
(3-35)
The viscous terms were logged by one time step. That is:
( ) ( )
n
Faces All
z y x
n
Faces All
z y x
S Tn Sn Rn S Tn Sn Rn
1
]
1

¸

∆ + + ·
1
]
1

¸

∆ + +
∑ ∑
+1
(3-36)
Bringing the unknown quantities to the left side, we get:
43
( )
( )
n
z y x
n
Num
n
n
Tn Sn Rn
S f q A V
t
q
) (
) (
) (

∑ ∑
+ + −
∆ − · ∆ +


)
(3-37)
where (∆q)
n
=q
n+1
- q
n
.
Computing
q
f
A
Num


·
)
is costly. Instead, the following approximate form was used:
S
S n V n
q
G
n
q
F
n
q
E
A
G z y x
∆ −

1
]
1

¸

⋅ −

,
_

¸
¸


+


+


·
I
max
λ
r
r )
(3-38)
where
max
λ is given by:
( ) a n V V
G
+ ⋅ − ·
r
r r
max
λ (3-39)
Finally, (∆q)
cell face
is defined as the average of (∆q) at all centers. This leads to a
seven-diagonal system linking the (∆q) value at node (i, j, k) with its five neighbors (it1,
j, k), (i, jt1, k), and (i, j, kt1).
44
This seven diagonal system of linear equations was solved by a three-factor scheme.
The factorization algorithm has been described in detail by Pulliam et al.
80
. Details of
the equations, the algorithm, and sample numerical calculations can be found in that
report, and many classical text books.
The solution of the matrix equations is obtained through a block lower-upper
decomposition (LU) and the application of the Thomas algorithm
71
. The advantage of
this algorithm is that direct inversions of the entire block matrices are not required.
3.4 TURBULENCE MODELING
In the most practical flows within turbomachines, the Reynolds number is high and
the flow is turbulent. Time and length scales of the turbulent flows are extremely small.
To capture the turbulence effects directly from the Navier-Stokes equations, the grid
resolution would need to be extremely high. To reduce the computational time, the
RANS (Reynolds Average Navier-Stokes System of equations) are solved in this work,
which simply requires replace of m and k with (µ + µ
t
) and (k + k
t
) respectively, where µ
t
and k
t
are “eddy” viscosity and “eddy” conductivity, respectively.
In this work, a one-equation turbulence model, the Spalart-Allmaras
81
model, has
been used for computing µ
t
=ρν
t
. Also,
t
p t
t
c
k
Pr
µ
· .
This model involves a number of constants:
c
b1
= 0.1335, c
b2
= 0.622, σ =2 /3, κ = 0.41,
σ κ
) 1 (
2
2
1
1
b b
w
c c
c
+
+ ·
45
cw2
= 0.3, c
w3
= 2, c
v1
= 7.1, c
t1
= 1, c
t2
= 2, c
t3
= 1.1 , c
t4
= 2
The eddy viscosity,
t
ν , is given by:
1
~
v t
f ν ν · ,
3
1
3
3
1
v
v
c
f
+
·
χ
χ
,
ν
ν
χ
~
· (3-40)
Here, ν is the molecular viscosity. The quantity ν
~
is the working variable and can be
calculated by solving the following partial differential equation
[ ]
2
1
2
12
2
1
1
2
2 2 1
~
)
~
(
~
)
~
.((
1
~
~
) 1 (
~
U f
d
f
c
f c
c S f c
Dt
D
t
b
w w
b t b
∆ +
1
]
1

¸

1
]
1

¸


+ ∇ + ∇ + ∇ + − ·
ν
κ
ν ν ν ν
σ
ν
ν
(3-41)
Where S is the magnitude of the vorticity,
2
2 2
~
~
v
f
d
S S
κ
ν
+ · ,
1
2
1
1
v
v
f
f
χ
χ
+
− · (3-42)
Here, d is the distance to the closest wall.
The function f
w
is:
46
6 / 1
6
3
6
6
3
1
1
]
1

¸

+
+
·
w
w
w
c g
c
g f , where ) (
6
2
r r c r g
w
− + · ,
2 2
~
~
d S
r
κ
ν
· (3-43)
The wall boundary condition is 0
~
· ν . In the free-stream, 0
~
· ν is best, provided
numerical errors don’t push ν
~
to negative values near the edge of the boundary layer. The
same is applied to the initial condition. The function f
t2
is:
) exp(
2
14 13 2
λ c c f
t
− · (3-44)
Finally,

the trip function f
t1
is obtained from the following equation:
[ ]
2 2 2
2
2
12 1 1
exp(
t t
t
t t t
d g d
U
c g c f +

− ·
ω
(3-45)
where ) , 1 . 0 min(
z
U
g
t
t


·
ω
and
t
ω is the wall vorticity at the trip point and d
t
is the
distance from the field point to the trip point, a user specified transition location.
This PDE was also recast in an integral form, and solved using a three-factor
approximate factorization scheme.
47
3.5 INITIAL AND BOUNDARY CONDITIONS
Since the Navier-Stokes equations are parabolic in time and elliptic in space, initial
and boundary conditions must be supplied and play a significant role in obtaining a
meaningful solution. In this section, the implementation of the initial and boundary
conditions is presented.
3.5.1 INITIAL CONDITIONS
At the start of the calculations, the flow properties everywhere in the system are
assumed to be uniform, what is called as a cold start. For example, the pressure values
equal the downstream pressure, and the u- velocity is set to a value consistent with the
specified total temperature T
0
at the inlet. As the compressor begins to draw the fluid in,
the pressure at the inlet drops relative to the outflow. The mass flow rate begins to
gradually change with time. It reaches either an asymptotically steady state or a limit
cycle oscillation, depending on whether the compressor is operating close to design
conditions, or at off-designed stalled conditions. The analysis can also use a previously
stored solution file with a corresponding grid file, to restart the computations. In this
manner, the time-accurate calculations for the compressor are simply a continuation or
restart from the previous solution. In the present method the grid and solution are stored
every few hundred time steps, based on the user input, and these files may be used to
continue the calculations or visualize the flow field.
48
3.5.2 BOUNDARY CONDITIONS
Figure 3.3 shows the different types of boundary conditions on a compressor flow
passage. The following boundary conditions have been implemented in the analysis:
Inflow Boundary: At this boundary, the stagnation temperature, T
0
, and the total
pressure, p
0
, are assumed to be known. In the simulation of a compressor drawing the air
from the atmosphere, these quantities may be set as ambient conditions. In a multistage
compressor analysis, these quantities will be known at the downstream boundary of the
previous stage. The tangential components of velocity at the inflow boundary are set to
zero, assuming that there is no swirl at the inlet. It is possible within the framework of the
present boundary conditions to prescribe swirl, a non-uniform total pressure distribution,
and a non-uniform total temperature at the inlet. The fifth equation applied at the inlet
boundary condition is the one-dimensional Riemann characteristic equation, modeling the
effects of acoustic disturbances leaving the computational field through the inlet face.
This equation is:
0 ) (
5 5
·


− +


s
R
a V
t
R
n
(3-46)
where
n
V
a
R −

·
1
2
5
γ
and s is the distance in a direction normal to the inlet face.
49
Neglecting
t
R


5
, at every step, the speed of sound, a, and normal velocity, u
n
are found at
the inflow boundary from the relation:
0
5
·


s
R
(3-47)
interior
5 5
R R
inlet
· (3-48)
Or,
1
2
0
2
2
1
2
1
2
1
2

· +



· −

γ γ
γ γ
a
n
u
a
Interior
n
u
a
n
u
a
(3-49)
After computing the speed of sound and temperature, T, isentropic relations are used to
find the values of the density ρ and pressure p at the inflow boundary. Neglecting
t
R


5
was considered acceptable because the flow properties varied very slowly with time
at the inlet.
Outflow Boundary: It was assumed in this study that the flow from the vaneless
diffuser downstream of the rotor exhausts into a plenum chamber with a constant volume,
V
p
. Figure 3.4 shows a schematic of the compressor outflow boundary coupled with the
plenum. Conservation of mass in the plenum can be written as:
50
) (
plenum
t p
m m
dt
dm
& & − ·
(3-50)
plenum plenum
) (
dt
V d
dt
dm ρ
·
(3-51)
Here,
P
m&
is the mass flow rate at the diffuser exit, and
t
m&
represents the mass flow rate
at the plenum throttle, and V is the volume of the plenum and is constant, therefore:
plenum plenum
dt
d
V
dt
dm ρ
·
(3-52)
Inserting Equation (3-52) into Equation (3-50), we have:
t p
m m
dt
d
V & & − ·
plenum
ρ
(3-53)
The relationship between density and pressure is obtained by assuming isentropic state in
the plenum:
2
plenum
plenum
a
p
·


ρ
(3-54)
51
The combination of Equation (3-54) and (3-53) gives the plenum pressure formula:
) (
2
t p
p
p p
m m
V
a
dt
dp
& & − ·
(3-55)
A first order discretization of Equation (3-55) gives:
) (
1
t p
n
p
n
p
m m K p p & & − + ·
+
(3-56)
where a value of 0.20 for K is used in this study. This value determines the natural
frequency of the plenum chamber, and indirectly determines surge. The quantity
t
m&
is
an input parameter that represents the nominal mass flow rate and lets the compressor run
at different operating conditions. All other quantities such as the density and the three
components of velocity are extrapolated from the interior.
Solid walls: At the solid walls, a no slip boundary condition is used. This condition
is applied to the hub, casing and blade surfaces. The velocity on the walls of the inlet and
the diffuser is set to zero, while the velocity for grid points on the compressor blades and
the shaft, that part which is moving, is set equal to r
r
r
× Ω . The pressure, density and
temperature values at the solid surfaces are found from the interior, using relationships
such as ∂p/∂n=∂T/∂n=∂ρ/∂n=0, where n represents the direction normal to the solid
52
surface. Density, and pressure on the hub and shroud are all directly extrapolated from
the interior. On impeller surfaces, because of the non-orthogonality of the cells around
the blade surfaces, especially near the leading and trailing edges, a normal momentum
equation, described by Pulliam et al.
85
, is used in the normal direction to the blade
surfaces.
Periodicity: Since simulation of the entire flow in a compressor requires significant
computations and is very expensive, usually just a few of the blade passages are modeled.
Therefore, it is assumed that the flow through first blade and the last blade passage are
nearly identical and a periodic boundary condition is applied. At the blade-to-blade
interface boundaries, the flow properties are computed by averaging the properties on
either side of the periodic boundary. Figure 3.5 illustrates the periodic boundary
conditions for blocks A and B.
Zonal Boundary: As shown in Figures 3.3 and 3.5, neighboring zones share grid
lines on their boundaries. The flow properties at these boundaries are computed by
taking the average of the properties on the neighboring points, as shown in Figure 3.5.
Bleed Boundary: In this study, both open-loop and closed-loop control
methodologies using steady and unsteady bleeding were studied. In both control methods,
bleed valves are implemented circumferentially downstream of rotor, located on the
shroud. In the open-loop control, a fraction of the mass flow is removed through a valve
at a fixed rate. A schematic of the open-loop control used in this study is shown in
Figure 3.6. The numerical implementation of the bleed boundary condition into the flow
53
solver requires user-specification of the bleed valves location, bleed valve area, A
b
, and
bleed mass flow rate, b m
.
. From this information, the normal velocity may be written as:
b b
b
n
A
m
u
ρ
.
· (3-57)
Density, pressure and the two tangential components of velocity are extrapolated from
the interior, consistent with the subsonic flow at this boundary.
While simulating the closed-loop control, the mass flow rate through the bleed valve
is linked to pressure upstream of the compressor face by a control unit. The mass flow
rate through the bleed valve is controlled by the instantaneous amplitude of pressure
fluctuations sensed by the “computational” probes, which in this work were located near
30% of the tip chord, upstream of the compressor face. The bleed valve was activated
when the pressure sensors upstream of compressor face experienced high-pressure
fluctuations, as discussed in Chapter VI. Figure 3.7 illustrates the schematic of the
closed-loop control system used in this study. As in open-loop control, the density and
the tangential velocities are extrapolated from the interior.
54
Figure 3.1 Control volume and cell-vertex grid points.
Figure 3.2 Computation of inviscid flux terms on a cell face.
i
j
k
(i,j,k) (i+1,j,k)
Cell face at i +
2
1
i-1 i i+1 i+2
Right
Cell face i+1/2
Stencil for q left Stencil for q right
Left
55
Zonal boundaries:
Properties are averaged
on either side of the
boundary.
Inlet
boundary:
P
0
, T
0
, v, w
are specified.
Riemann-
invariant
extrapolated
from interior.
Solid
wall boundary:
No-slip
boundary
conditions.
0 ·


·


·


n
T
n n
p ρ
Periodic
boundaries:
Properties are
averaged on
either side of the
boundary.
Outlet
boundary:
t m
.
is specified.
All the other
quantities are
extrapolated from
the interior.
Figure 3.3 Boundary conditions used in the axial compressor analysis.
56

) (
2
t p
p
p p
m m
V
a
dt
dp
& & − ·
Figure 3.4 Compressor outflow boundary conditions.
Figure 3.5 Periodic and zonal boundary conditions.
Block B Block A
Periodic boundaries
Zonal boundary
Plenum
Chamber
u(x, y, z) = 0
a
p
= constant
Isentropic
Outflow
boundary
c
m& t
m&
P
m&
57
Pressure
Sensors
Controller Unit
Bleed Valve
Impeller
Figure 3.6 Implementation of bleeding as an open-loop active control.
Figure 3.7 Implementation of bleeding as a closed-loop active control.
One tip chord length
Impeller
Bleed Valve
58
4 CHAPTER IV
CODE VALIDATION STUDIES
The numerical scheme and boundary procedures described in Chapter III were coded
in a computational flow solver, called GT-TURBO3D. Results are presented in the
following three chapters. The solver first reads a main input data file. This main input
file controls the flow type (inviscid, laminar, turbulent), and the total number of time
steps. It also specifies the computational grid size and the number of grid blocks, to
perform the necessary initialization. It also reads nominal mass flow rate,
t
m& . Next, the
code reads the boundary condition setup data file to establish the appropriate boundary
conditions described in Section 3.4 on the different faces of each of the blocks. The
solver and the boundary conditions have been written in a general form so that most
multi-block, patched grid analyses can be performed with this code.
Most calculations were carried out on a Silicon Graphics Origin 2000 system using
two of the four available processors. Although the present implicit scheme can use large
time steps, the CFL numbers was kept at or below 2 in order to resolve the unsteady flow
in a time accurate manner. The calculations were generally run for several compressor
revolutions, before a steady state or a limit cycle oscillation (indicative of separation
and/or stall) was achieved. CPU time per time step was 3x10
-5
sec per grid point. The
59
mass flow rate across several streamwise planes was monitored and used as a
convergence criterion.
This chapter is organized as follows. The two axial compressor rotors used in this
study are introduced in Section 4.1. In this section, geometry configurations and
computational grids are described as well as their design operating characteristics. A
brief literature survey of work done by other researchers on these two configurations is
presented in Section 4.2. In Section 4.3, comparisons of the experimental data with the
CFD results are given in the form of contour plots, graphs, and velocity vectors at the
peak efficiency. Some grid sensitivity studies are also given.
4.1 DESCRIPTION OF THE CONFIGURATIONS STUDIED
Two different configurations have been chosen to study the compressor stall and
passive/active control methodologies. These configurations are the NASA Rotor 67 and
NASA Rotor 37 axial compressors. In this section, the configurations and the
computational grids are presented.
4.1.1 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 67)
The first geometry chosen to validate the code and simulate compressor stall control
was a transonic high-speed axial fan rotor, called NASA Rotor 67, shown in Figure 4.1.
This low aspect ratio rotor (σ=1.56) is the first stage rotor of a two stage transonic
fan designed and tested with laser anemometer measurements at the NASA Glenn
Research Center. The design pressure ratio is 1.63, at a mass flow rate of 33.25 kg/sec.
60
The rotor has 22 blades with tip radii of 25.7 cm and 24.25 cm at the leading and trailing
edge, respectively. The hub/tip radius ratio is 0.375 at the leading edge and 0.478 at the
trailing edge. The design rotational speed is 16043 RPM, and the tip leading edge speed
is 429 m/sec with a tip relative Mach number 1.38. To determine the rotor mass flow
rate, a calibrated orifice was used. A detailed description of blade and flow-path
geometry, as well as measured data on total pressure, temperature, static pressure and
flow angle data, may be found in Reference [82]. Figure 4.2 shows the locations where
velocities were measured for this compressor. The hub consists of two stationary and one
rotational segments. The rotor hub is in the region of -3.533 cm

< x < 7.106 cm and
rotational wall conditions were applied to this region.
In order to model the rotating stall, multiple flow passage should be considered. For
this purpose, a H-H-type grid, shown in Figure 4.3, with four compressor flow passages
was generated using the “Gridgen” meshing package. The grid was generated by loading
the geometry surface data from Reference [82] into the mesh package. In each flow
passage, the grid, called the baseline grid, has 32 and 21 cells in the radial and
circumferential directions, respectively. Sixty-six cells were placed in the streamwise
direction, of which 33 were located over the blade chord. A hyperbolic tangent node
point distribution near the surfaces was used to provide adequate resolution of the
boundary layers. The clearance gap was spanned by three cells in the radial direction.
The grid at the mid-span and the grid normal to the streamwise direction are shown
in Figures 4.4 and 4.5. Figure 4.6 shows the grids in the r-x plane, known as the
61
meridional plane. In this study, results in the form of vectors or contour plots are
presented on one of these planes. This is because three-dimensional visualizations of
flowfield in compressors are difficult due to configuration complexity, e.g. twisted
blades. Grid sensitivity studies were also performed to ensure that the baseline grid has
adequate resolution to resolve the leakage flow around the tip, the solid wall boundary
layers and the shock system. Simulations were also done on another grid with twice the
number of grid points in each of the three directions (4x131x63x41). Results for this
configuration are discussed in Section 4.3 of this chapter and the next two chapters.
4.1.2 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 37)
The second configuration of interest was a high-speed axial flow compressor, called
NASA Rotor 37. Rotor 37 is a NASA stage 37 rotor designed at the NASA Lewis
Research Center as a test compressor for a high-pressure ratio (20:1) compressor core of
an aircraft engine. The rotor design pressure ratio is 2.106 at a mass flow rate of 20.2
kg/sec. The rotor has a constant meanline diameter with a hub/tip radius ratio of 0.7. The
design rotational speed is 17188 RPM, producing a speed of 454.19 m/sec at the inlet tip.
The tip relative Mach number is 1.48. The clearance at the design is 0.36 mm. Details of
this compressor may be found in references [83] and [84].
Figure 4.7 shows the Rotor 37 configuration. Figure 4.8 shows a meridional view of
this rotor and the locations of laser anemometers at different stations. For this rotor, as for
Rotor 67, some part of the hub is rotating. At the hub, stationary wall boundary
62
conditions were applied in the region x < -0.264 cm and x > 4.521 cm, while rotational
boundary conditions were applied in the region –0.264 cm < x < 4.521 cm.
The grid for NASA Rotor 37 was generated with the same methodology that was
used for NASA Rotor 67. The grid is a H-H-type with four flow passages, shown in
Figure 4.9. It has 119x71x41 grid points in each flow passage in the streamwise,
spanwise and pitchwise directions, respectively. Nine points were placed in the clearance
gap. As in the case of NASA Rotor 67, the tip clearance gap was modeled and periodic
boundary conditions were applied. Results for this configuration are given in Section 4.4.
4.2 PREVIOUSLY REPORTED STUDIES ON ROTOR 67 AND ROTOR 37
CONFIGURATIONS
Rotor 67 is an AGARD test case and is suitable for CFD validation purposes.
Several researches have studied flowfields within this compressor. Studies conducted in
the stable part of the design speed operating line for Rotor 67 may be found in References
[85]-[92].
Effects of tip clearance gap on the flowfield have also been investigated by
Adamczyk et al.
85
and Chima
86
. Chima obtained the results for two different tip
clearance gap sizes, full and half, with an accurate modeling of the clearance gap. Effects
of shock boundary layer interaction and wake development are reported by Hah et al
91
.
End-wall and casing treatment for Rotor 67 may be found in Reference [92].
Rotor 37 was used to assess the prediction capabilities of turbomachinery CFD tools
in a blind test organized by the 1994 ASME/IGTI workshop. In the CFD assessment, the
63
Rotor 37 was tested in isolation by Suder et al.
93
. Several researchers computed the
complex flow field for this rotor for the stable branch of the performance map. These
results may be found in References [90], [94] and [95]. Bright et al
96
have studied the
dynamics for the Stage 37 during rotating stall.
4.3 PEAK EFFICIENCY RESULTS FOR ROTOR 67
The viscous flow solver, described earlier, has been validated for Rotor 67 and Rotor
37 compressor configurations. The flow solver had previously been used by this author
to simulate the flowfield in a low speed centrifugal compressor, called the NASA Low
Speed Centrifugal Compressor (LSCC, built and tested at NASA Glenn Research
Center). The results for the centrifugal compressor, including the comparison of CFD
predictions and experimental measurements such as blade pressure distributions, may be
found in References [66]-[67]. High-speed centrifugal compressors have also been
studied using this flow solver and computational-experimental comparison results can be
found in References [68] and [70]. Because the predominant thrust of the present study is
to model axial compressors, the LSCC validation results are not presented here.
In the present study, the flow solver was first applied to the simulation of flow
through the NASA Rotor 67 compressor at peak efficiency conditions, for which
experimental results may be found in Reference [82]. In this section, CFD results for
both the baseline grid and the fine grid are presented and compared with the experimental
data.
64
At the start of calculations, no restart solution files exist. Therefore, as pointed out
in Section 3.4.1, the flow properties in the computational domain were assumed to be
uniform and were set to their inlet freestream values. As the flowfield was advanced in
time, the pressure at the inlet drops relative to the outflow. The drop in pressure depends
on the compressor speed. Since NASA Rotor 67 is a high-speed compressor at its design
speed, the pressure drop, especially in the blade suction side near the leading edge, is
large and can produce non-physical pressure in that region. A start-up technique similar
to that in a real compressor during acceleration was used to avoid this problem. The flow
solver was applied to the rotor at three different speed levels: 30%, 50% and 100% of the
design speed. The steady solution at each time level was used as a restart solution file for
the next speed level. By this methodology, the pressure was allowed to drop in the
suction side of the blades gradually and a steady state solution was achieved. A useful
byproduct is performance data at off-design engine speeds.
At each time step, mass flow rates across the streamwise planes were calculated and
averaged along the streamwise direction from the rotor leading edge to the rotor trailing
edge. The averaged mass flow rate was considered as the mass flow rate through the
rotor and was used as a convergence criterion. To calculate the total pressure ratio, the
radial distributions of total stagnation pressure were mass averaged across the annulus.
The formula used was:
65
∫∫
∫∫
·
VdA
VdA
p
p
p
p
ρ
ρ
01
02
01
02
(4-1)
Here, P
02
and P
01
are the stagnation pressure at station 2 and station 1, shown in Figure
4.2.
In this study 20 time steps per degree of revolution were used. Therefore, one rotor
revolution required about 7000 computational time steps. For the baseline grid, the time
required to perform one compressor revolution (where blades rotate 360°) was between 9
and 12 hours, using either one or two processors of the Origin 2000. In the fine grid, the
running time was increased to 65 hours for one rotor revolution. Figures 4.10 and 4.11
show the fluctuations of mass flow rate and total pressure ratio. The computed mean
mass flow rate was 34.23 kg/sec on the baseline grid, and the corresponding
measurement data is 34.61 kg/sec (1.2% difference). The mass flow rate on the fine grid
was identical to that on the baseline grid, up to 3 significant digits. These figures show
that the mass flow and total pressure fluctuations at this operating condition are very
small (less than 1%).
Figure 4.12 shows a comparison between the CFD predictions and the measured
normalized axial velocity distribution at the inlet survey station 1 (Figure 4.2). The
velocity is normalized by critical velocity at standard day conditions, V
STD
=310.63
m/sec. CFD results are given for both fine and baseline grids. The computed axial
velocity showed only a slight difference between the fine grid and the baseline grid.
66
Good agreement between the measurements and the predictions was obtained. It is also
seen that the grids have enough resolution to capture the boundary layer profiles near the
walls (hub and shroud).
The flow pattern inside a transonic fan like the NASA Rotor 67 is very complicated
due to phenomena such as a 3-D shock system, shock-boundary layer interaction,
clearance flow and three-dimensional separation. Figures 4.13 through 4.15 show
comparisons between the computed and measured relative Mach number at two different
span locations. Figure 4.13 shows the relative Mach number contours at the 30% span
location, measured from the hub. The corresponding experimental contours were
obtained from Reference [87]. There are several reasons the experimental data do not
usually produce good quality contour maps. Background noise, problems associated with
laser anemometer blade flash and data drop-out, non-uniform survey spacing, and
inadequate measurement locations in viscous regions and the boundary layer and wake all
produce poor quality plots. The experimental contours in Figure 4.13 were produced by
Pierzga et al.
87
, who smoothed and interpolated the measurements to fill the zero velocity
data using information from the adjacent survey points in both the streamwise and
pitchwise directions. The smoothing and interpolation procedures may be found in
Reference [87]. The inlet mach number is 0.95 and the flow exits at a Mach number of
about 0.60. As in experimental results, the CFD calculations showed that most of the
flow at this span location is subsonic. However, small regions of supersonic flow, with
relative Mach numbers about 1.05 were observed on the forward position of the suction
67
surface of the blades. The computed results in Figure 4.13 show shear layers over the
blade and in the wake. These are not visible in the experimental contour plots. As
mentioned earlier, this may be due to the lack of adequate measurement locations in the
viscous regions.
Figures 4.14 and 4.15 show the measured and computed relative Mach number at
90% span at two different azimuthal locations. The horizontal axes were normalized by
the chord at the corresponding span locations. Zero percent and 100% chord on these
plots represent the blade leading and trailing edge, respectively. The comparisons
generally agree well and a shock structure was observed. The computed relative Mach
number showed a maximum difference of five percent between the baseline grid and fine
grid.
The static pressure contours and the shock locations in the four flow passages at two
different radial locations, 30% and 70% span, are illustrated in Figure 4.16. This figure
demonstrates how the pressure through the compressor builds up. At both these radial
locations, the pressure drops on the blades suction surfaces as a consequence of flow
acceleration, in the same way an airfoil behaves. At 70% span, the pressure gradient is
much higher than at 30% span, and a bow shock occurs. All of the four flow passages
showed identical pressure distributions, indicating the flowfield is identical and well
converged in all blade passages.
Figure 4.17 shows the relative velocity profile at mid-span on three blade passages,
viewed from the top of the rotor. It is seen that the shock nearly covers the gap between
68
the blades, suggesting choke conditions are imminent. Velocity profiles are well behaved
in the blade passages with no evidence of reversed flow. Flow properties and the velocity
flowfield were almost identical from blade to blade. The relative velocity vectors on a
plane normal to the rotor axial direction is given in Figure 4.18. This figure shows the
cross flow and does not show any stall cells. Such a well-behaved flow is, of course,
expected at the peak efficiency conditions.
To visualize the shock-boundary layer interaction, relative velocity vectors near
the blade surfaces and inside the boundary layer are plotted in Figure 4.19 and 4.20.
Near the suction side just after the shock, a very strong outward flow in the radial
direction was observed, as shown in Figure 4.19. This outward flow pattern has also
been reported in References [89] through [92]. The shock is very strong from the tip to
40% of the span and causes the boundary layer to separate immediately aft of the shock.
Below 40% of span from the tip of the blade, as discussed by Hah
91
, the outward flow is
not directly due to the shock-boundary layer interaction, but is related to a conventional
adverse pressure gradient. As pointed out by Chima
89
, the flow is also strongly
influenced by a vortex roll-up on the leading edge close to the blade root. Flow near the
pressure side of the blades, shown in Figure 4.20, is more streamlined because no shock
forms on this surface boundary layer. In these figures the results shown are on both fine
and baseline grid. These figures and the preceding results showed that the baseline grid
has enough resolution to capture all the important phenomena occurring in this
compressor. Because of the large number of computations involved in surge and rotating
69
stall simulations, the baseline grid was used in simulations of off-design operating
conditions, as well as in the passive and active bleed control calculations.
Figure 4.21 illustrates the meridional velocity profile at the peak efficiency at
mid-passage. The overall flow field is well attached and behaved. Small regions of
reversed flow between the blade tip and the shroud were observed in an enlarged view,
shown in the same figure. This reversed flow and the leakage tip vortex play dominant
roles in triggering surge and rotating stall, as discussed later.
70
90
70
50
30
10
Figure 4.1 NASA Rotor 67 configuration, Reference [82].
Figure 4.2 Laser anemometer and aerodynamic locations for Rotor 67, Reference
[90].
51.4 cm
71
Figure 4.3 Computational grid for Rotor 67.
Figure 4.4 Streamwise computational grid at midspan for Rotor 67.
72
Figure 4.5 Pitchwise computational grid for Rotor 67.
Figure 4.6 Computational grid in the meridional plane for Rotor 67.
73
Figure 4.7 NASA Rotor 37 configuration, Reference [84].
Figure 4.8 Laser anemometer and aerodynamic locations for Rotor 37, Reference
[90].
74
Figure 4.9 Computational grid for Rotor 37.
Figure 4.10 Time history of mass flow rate fluctuations about mean flow at peak
efficiency conditions (Rotor 67).
-1
-0.6
-0.2
0.2
0.6
1
0 1 2 3
M
a
s
s
f
l
o
w

R
a
t
e

F
l
u
c
t
u
a
t
i
o
n
s

(
k
g
/
s
)
Rotor Revolution, Ωt/2π
Ω Ω
75
0.00
0.10
0.20
0.30
0.40
0.50
0.60
0.70
0.00 0.20 0.40 0.60 0.80 1.00
Fraction of Span from Hub to casing
U
/
V
S
T
D
CFD-Baseline Grid
Laser Measurment
CFD-Fine Grid
Figure 4.11 Pressure and mass flow rate fluctuations at peak efficiency conditions
(Rotor 67).
Figure 4.12 Comparison between experiments and present results of inlet axial
velocity at peak efficiency conditions (Rotor 67).
- 50
- 30
- 10
10
30
50
- 40 - 30 - 20 - 10 0 10 20 30 40
% Pressure
Fluctuations
% Mass Flow Rate Fluctuations
76
Figure 4.13 Comparison between experiments and present predictions of relative
Mach number contours at 30% span at peak efficiency (Rotor 67).
Measured
(Reference 87)
Computed
77
0.6
0.8
1
1.2
1.4
1.6
-125 -50 25 100 175
% Chord
M
CFD-Baseline Grid
CFD- Fine Grid
experimental
0.6
0.8
1
1.2
1.4
1.6
-125 -50 25 100 175
% Chord
M
CFD-Basedline Grid
CFD-Fine Grid
experimental
Figure 4.14 Comparison between experiments and present results of relative Mach
number at 90% span and 30% pitch at peak efficiency (Rotor 67).
Figure 4.15 Comparison between experiments and present results of relative Mach
number at 90% span and 50% pitch at peak efficiency (Rotor 67).
78
Figure 4.16 Static pressure contours at 30% and 70% span at peak efficiency
(Rotor 67).
30% Span
70% Span
79
Figure 4.17 Relative velocity profile at mid-span at peak efficiency (Rotor 67).
Figure 4.18 Secondary flow in the circumferential direction at peak efficiency
(Rotor 67).
80
Fine Grid
LE
TE
Shock
Baseline Grid
Figure 4.19 Flow field near blade suction surface at peak efficiency (Rotor 67).
81
Fine Grid
Hub
Casing
Baseline Grid
Figure 4.20 Flow field near blade pressure surface at peak efficiency (Rotor 67).
82
Casing
LE
Tip
(
Figure 4.21 Flow field at mid-passage at peak efficiency (Rotor 67).
83
5 CHAPTER V
RESULTS AT OFF-DESIGN CONDITIONS
In the previous chapter, the stable operating, peak efficiency condition results for the
compression system NASA Rotor 67 were discussed. It was shown that at this operating
point, the mass flow and pressure fluctuations are very small. Furthermore, no azimuthal
disturbances (stall cells) were observed and the flow was circumferentially uniform.
However, as pointed out in Chapters I and II, as the mass flow rate through the
compressor is decreased from this stable point to the surge line, disturbances grow and
lead the system to unstable conditions. Investigation of the flowfield at off-design
conditions is one of the goals of this study, and is documented in this chapter
In this chapter, the computational results for the two axial compressors introduced in
Chapter IV, are presented at different operating conditions. Results include the
characteristic performance maps, the time history of the flow at selected points on these
curves, velocity vectors and pressure contour plots of the flowfield, and time traces of
pressure upstream of the compressor face using computational probes located around the
circumference of the compressor. Results for the Rotor 67 compressor are presented in
Section 5.1. The calculations for the Rotor 37 are given in Section 5.2.
84
5.1 ROTOR 67 RESULTS
Results from CFD unsteady calculations for the compressor NASA Rotor 67 at two
different operating conditions are discussed in this section. In Section 5.1.1, results at
the onset of stall are presented. Stalled flow results are given in Section 5.1.2.
5.1.1 ONSET OF STALL
As mentioned earlier, the performance map of a compression system is the relation
between the total to total pressure ratio and the compressor mass flow. In real life, the
characteristic performance map of a compressor is obtained by changing the compressor
mass flow rate using a throttle valve in the plenum.
In this study, the flow solver was applied to the NASA Rotor 67 for 10 different
operating conditions to construct the overall rotor performance map. The backpressure is
calculated from Equation (3-56) with a user specified mass flow rate at the plenum exit.
From the pressure difference between the compressor inlet and diffuser exit, the mass
flow rate through the compressor can be computed, and the operating point obtained. The
radial distributions of total temperature and stagnation pressure are mass averaged across
the annulus.
The calculated total pressure ratios at full speed are plotted against the normalized
mass flow and compared to experimental data in Figure 5.1. Both calculated and
experimental data are nondimensionalized by their corresponding choked mass flow rate.
This kind of normalization removes uncertainties in the experimental mass flow rates.
The reported choked mass flow rate for the NASA Rotor 67 is 34.96 kg/sec and the
85
calculated value is 34.72 kg/sec (0.7% difference). It is seen that the performance map
agrees well with the experimental data. Three points, A, B and C, indicated on the
performance map, are the operating points where the detailed flowfield was studied.
Point A in Figure 5.1 is related to the peak efficiency condition, and its corresponding
results were studied in the previous chapter. Here, the CFD calculations for the other two
points, B and C, are discussed.
The adiabatic efficiency of a compressor is the ratio of the ideal input work needed
to raise the total pressure of a working fluid from a pressure value P
01
to a new value P
02
,
to the actual work needed to accomplish this task. The adiabatic efficiency of the
compressor can be found by using:
a
s
ad
W
W
· η (5-1)
where W
s
and W
a
are the isentropic and actual work done on the flow, respectively.
These can be found as follows:
) (
) (
01 02
.
01 02
.
T T c m W
T T c m W
p a
p s
s
− ·
− ·
(5-2)
Applying isentropic relationship for point 1 and 2:
86
γ
γ 1
01
02
01
02

,
_

¸
¸
·
p
p
T
T
s
(5-3)
and using Equations (5-1) to (5-3), we get :
1
1
01
02
1
01
02

,
_

¸
¸
·

T
T
p
p
ad
γ
γ
η (5-4)
The radial distributions of total temperature and stagnation pressure are mass
averaged across the annulus and are obtained from Equation (4-1) and the following
equation:
∫∫
∫∫
·
VdA
VdA
T
T
T
T
ρ
ρ
01
02
01
02
(5-5)
Figure 5.2 shows the comparison of measured and computed adiabatic efficiency for
different operating conditions. The predicted peak efficiency occurs at a slightly higher
mass flow rate as compared to the experimental results.
87
Stall inception in compressor experimental studies is normally detected by hot-
wire probes placed in a stationary frame around the circumference of the compressor. In
the present study, four “numerical” probes were circumferentially located in a stationary
system 30% of the tip chord upstream of the rotor blade row, as shown in Figure 5.3.
The probes were placed in the mid-pitch of each of the flow passages and at 90% of the
span from the hub. At each computational time step the probes calculated the axial
component of the flow velocity and the corresponding pressure.
Figures 5.4 shows the fluctuations of mass flow rate and total pressure ratio for
operating point B on the compressor map in Figure 5.1. The computed mean mass flow
rate was 31.6 kg/sec and the corresponding measurement data at the onset of stall is 92%
of choked flow, or 32.1 kg/sec. (~1.6% difference). Compared to the mass flow and
pressure fluctuations at point A in Figure 4.11, it can be seen that the fluctuations at this
point increased by about 15%. The amplitude of the flow oscillations at point A was less
than 1%.
The spatial average of static pressure from the probes upstream of the compressor
face is shown in Figure 5.5. The averaged pressure fluctuated between 0.57 and 0.67 of
the about value at the inlet face, 16% of its mean value. Figure 5.6 illustrates the
deviations of the static pressure of each of the four probes from their azimuthally
averaged values. For clarity, the time traces shown in Figure 5.6 are shifted vertically by
a constant interval. All the probes showed same amount of deviation and are very close
to zero, indicating the flow was periodic from blade to blade. In the other words, no
88
evidence of circumferential disturbances was observed, even though the amplitudes of the
mass and total pressure fluctuations are about 15% and 10% of the mean values,
respectively.
Figure 5.7 depicts the time traces of the spatial average of the axial velocity
component upstream of the compressor face. The axial velocity fluctuations were small
and there was no reversed flow for this operating condition.
5.1.2 ROTATING STALL AND MODIFIED SURGE SIMULATIONS
Decreasing the mass flow rate to 84.6% of the choked mass flow rate (29.4 kg/sec)
causes the compressor to experience very large amplitude fluctuations of both mass flow
rate and total pressure ratio. Figure 5.8 shows the time history of the mass flow rate for
about 25 rotor revolutions. Limit cycle oscillations with a frequency of about
70
1
blade
passage frequency, equal to 84 Hz, was observed. The total pressure ratio fluctuation
versus the mass flow fluctuation is illustrated in Figure 5.9. A comparison of Figures 5.9
and 5.4 indicates that as the mass flow rate through the compressor is decreased by six
percent, from point B to point C in Figure 5.1, the amplitude of the total pressure ratio
oscillation increases from 10% to 50%. Vector plots in meridional planes and pressure
contour plots in circumferential planes were used to study the effects of these large flow
fluctuations on the flowfield.
In Chapter IV, it was shown that at peak efficiency conditions, generally the flow
was well aligned and just a small region of reversed flow in the clearance gap was
89
observed (see Figure 4.21). Figure 5.10 illustrates the growth of this reversed flow and
flow separation at operating point C at three different time instances. These vector plots
are shown in the meridional plane at the midpitch of one of the flow passages. The rapid
growth of the tip leakage vortex over one and half rotor revolution causes some reversed
flow in the inlet near the casing.
The static pressure contours in the circumferential plane upstream of the compressor
face for operating conditions C are illustrated in Figure 5.11 at one instance in time. In
this figure, darker regions represent low-pressure areas and the lightly shade regions
correspond to high-pressure areas. It was observed that the low-pressure regions are
moving in the opposite direction of the shaft rotation, in a rotating frame. The spatial
averages of axial velocity and static pressure from the probes readings are shown in
Figure 5.12. Three different levels of amplitude with high frequency fluctuations,
indicated by sections I, II, and III, were observed in this figure. These levels can be
called as recovery level, precursor level and stall level. The amplitude of fluctuations for
both pressure and axial velocity in the recovery level, section I, were very small, about
3% of their average. After about one revolution, the amplitude of fluctuations increased
to 20% (precursor level, section II), and lasted for approximately half of the rotor
revolution. Finally the stall cycle leads to a very large fluctuations, 50%, in stall level
(section III), for one compressor revolution.
A power spectral density analysis was performed to determine the dominant
frequencies of the averaged static pressure fluctuations. The results are shown in Figure
90
5.13. This figure shows that the dominant frequency of these azimuthal disturbances is
about 100 Hz, which is about 38% of the rotor frequency.
Figure 5.14 illustrates the deviation of axial velocity from the spatial averaged axial
velocity (shown in Figure 5.12) for all of the four probes. As in Figure 5.6, the time
traces shown in this figure are shifted vertically by a constant interval. It was observed
that the probes experienced same axial velocity magnitude for about 1.5 rotor revolutions,
followed by a different velocity magnitude in each probe for approximately another 1.5
revolutions. Figures 5.11 to 5.14 indicate that the flow is not symmetric from one blade
passage to the next and a part span rotating stall occurs. From the above discussion, it is
believed that the compressor at this operating condition, point C, is experiencing a three
dimensional instability that is similar to a modified surge, which is a combination of
rotating stall and surge.
5.2 ROTOR 37 SIMULATIONS
The second configuration, which was briefly studied, was the compressor NASA
Rotor 37. As mentioned earlier, Bright et al.
96
at NASA Glenn Research Center have
obtained surge and rotating stall data for this compressor. The experimental data are for
the Stage 37, which includes Stator 37 and Rotor 37. In this study only Rotor 37 was
simulated, and the effects of stator and rotor interaction were not evaluated. No one-to-
one comparisons with the experimental measurements were attempted, due to date
availability.
91
The flow solver was applied to the NASA Rotor 37 for seven different operating
conditions. The same methodology used in Section 5.1.1 was applied to study
compressor operation at 70% of design speed, which was about 12000 RPM. This RPM
corresponds to the compressor RPM in the NASA test.
From these results, the compressor performance map was extracted and is shown in
Figure 5.15. Good agreement between the computed results and measured data was
observed.
The time history of the mass-flow rate and total pressure ratio at the selected points
A, B and C, indicated on the performance map, are plotted in Figure 5.16. The mass flow
rate for point A was 15.0 kg/sec, and at this operating point calculations showed very
small fluctuations (~1%). At point B, the mass flow was 13.5 kg/sec and the mass flow
fluctuations increased to 5% of the mean flow. The computed result for point C was
about 12 kg/sec, with 15% fluctuations in the mass flow.
To investigate azimuthal disturbances, computational probes one tip chord upstream
of the compressor face were used. Figure 5.17 shows the deviation of pressure
magnitudes from the circumferentially averaged pressure. This figure indicates that the
pressure disturbances were small compared to the 15% fluctuations in mass flow at this
operating condition. Spectral power density for these azimuthal disturbances was plotted
in Figure 5.18. This figure revealed that these disturbances had a very high content
frequency (1-8 kHz) with small amplitudes. These high frequency small disturbances
may be due to fine scale boundary layer phenomena. Figure 5.17 indicates that the
92
compressor experienced a one-dimensional disturbance, which may referred to as a mild
surge. We call this a surge because the pressure fluctuations at all four probes are nearly
identical, with no phase difference indicative of a rotating stall.
93
0.84
0.86
0.88
0.9
0.92
0.94
0.88 0.9 0.92 0.94 0.96 0.98 1
E
f
f
i
c
i
e
n
c
y
Experiment
CFD
.
.
Choked
m
m
Figure 5.1 Comparison of measured and computed characteristic performance map
(Rotor 67).
Figure 5.2 Comparison of measured and computed adiabatic efficiency (Rotor 67).
.
.
Choked
m
m
A
B
C
1.3
1.35
1.4
1.45
1.5
1.55
1.6
1.65
1.7
1.75
1.8
0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1
T
o
t
a
l

P
r
e
s
s
u
r
e

R
a
t
i
o

CFD
Experiment
94
Figure 5.3 Locations of the computational probes (Rotor 67 and Rotor 37).

I
II
III
IV
I
II
III
IV

95
Figure 5.4 Pressure and mass flow rate fluctuations at the onset of stall, operating
point B (Rotor 67).
Figure 5.5 Time history of averaged pressure computed by probes upstream of the
compressor face at operating point B (Rotor 67).
-50
-30
-10
10
30
50
-40 -30 -20 -10 0 10 20 30 40
% Pressure
Fluctuations
% Mass Flow Rate Fluctuations
Rotor Revolution, Ωt/2π
ref
P
P
96
Figure 5.6 Time history of pressure deviation from its azimuthally averaged
pressure upstream of the compressor face at operating point B
(Rotor 67).
Figure 5.7 Time history of averaged axial velocity computed by probes upstream of
the compressor face at operating point B (Rotor 67).
Rotor Revolution, Ωt/2π

a
u
Rotor Revolution, Ωt/2π
ref
P
P P −
97
-50
-30
-10
10
30
50
-40 -30 -20 -10 0 10 20 30 40
% Pressure
Fluctuations
% Mass Flow Rate Fluctuations
Figure 5.8 Time history of mass flow rate fluctuations at operating point C
(Rotor 67).
Figure 5.9 Pressure and mass flow rate fluctuations at operating point C (Rotor 67).
Rotor Revolution, Ωt/2π
.
m(kg/sec)
98
a) After ½ Rotor Cycle
b) After 1 Rotor Cycle
c) After 1.5 Rotor Cycle
Figure 5.10 Temporal growth of reversed flow at operating point C (Rotor 67).
99
Ω Ω
Low static
pressure regions
Figure 5.11 Instantaneous circumferential pressure fields at operating point C
(Rotor 67).
100
Figure 5.12 Time history of averaged axial velocity and pressure computed by
probes upstream of the compressor face at operating point C
(Rotor 67).
Rotor Revolution, Ωt/2π
ref
P
P
I
II III

a
u
101
Figure 5.13 Power spectral density of the azimuthally averaged pressure
fluctuations upstream of the compressor face at operating point C
(Rotor 67).
Figure 5.14 Time history of axial velocity deviation from its azimuthally averaged
axial velocity upstream of the compressor face at operating point C
(Rotor 67).
Rotor Revolution, Ωt/2π


a
u u
Frequency Hz
Power Spectral
Density of Pressure
102
Figure 5.15 Computed and measured characteristic performance map at 70%
design speed (Rotor 37).
Figure 5.16 Fluctuations of total pressure ratio versus mass flow rate at different
operating conditions (Rotor 37).
Mass Flow Rate (kg/sec)
A
B
C
1
1.2
1.4
1.6
10 12 14 16 18
T
o
t
a
l

P
r
e
s
s
u
r
e

r
a
t
i
o
Experiments
CFD
103
ref
P
P P −
Rotor Revolution, Ωt/2π
Figure 5.17 Time history of pressure deviation from its azimuthally averaged
pressure upstream of the compressor face at operating point C
(Rotor 37).
Figure 5.18 Power spectral density of pressure deviation from its azimuthally
averaged pressure fluctuations upstream of the compressor face at
operating point C (Rotor 37).
Power
Spectral
Density of
Pressure
Deviation
104
6 CHAPTER VI
ACTIVE CONTROL STUDIES OF ROTOR 67 CONFIGURATION
Axial compressors operate in regimes where both rotating stall and surge impose
limits on low flow rate operability. The useful operating range can be as small as 10
percent of the design flow rate. For example, in the compressor Rotor 67 described in the
previous chapters, the surge line occurs at 92% of the choked flow. In the other words,
the stable operating range in this rotor is just 8% of the maximum delivery flow.
Therefore, control methodologies are necessary for extending the stable operating range
of compressors. Control strategies designed to enhance the operability of compressors
must address rotating stall, surge, and their interaction. In recent years, there has been
much work on improving operating range of compression systems using active and
passive control methods. These methods were discussed in Section 2.3 and schematically
shown in Figure 2.7.
Ever since the Moore-Greitzer
23-24
model became available, work on feedback
stabilization methods for compressors has been extensive and a considerable number of
papers have been published. One of the most common of these methods is bleeding.
Since its application in early compressors more than 60 years ago, bleeding has been used
for engine start-up purposes. Bleeding has also been used over the past two decades to
shift the surge line to lower mass flow. Fisher
41
used a bleed valve in a centrifugal
105
compressor at various compressor shaft speeds. At the maximum speed, 78000 RPM, the
surge line was decreased by 21% toward the low mass flow rate side. Between 50000
and 70000 RPM, this shift was 28% to 31%. In a series of experiments, Pinsley et al.
42
studied centrifugal surge control using throttle valves as actuators, and a 25% reduction
of mass flow rate prior to onset of surge was achieved. The amount of bleeding has been
reported ranges from 1 to 10 percent of the mean flow
40-43
. Eveker et al.
97
were one of
the first groups to report successful experimental implementation of a bleed valve
controller. In the series of papers published by Yeung and Murray
40
, and Wang et al.
98
,
at Cal Tech, analysis of bleed valve rate requirements for suppressing rotating stall are
presented. Yeung et al.
40
experimentally showed that the rate of bleeding may be reduced
by injecting air downstream of the compressor face. They showed that by one to two
percent air injection, the rate of bleeding could be reduced by 30%. Prasad et al.
99
at
Georgia Tech used bleeding in the plenum to reject surge in axial and centrifugal
compressors.
In this study, two types of bleeding were considered. The first one is a simple
steady bleeding control, referred to as open-loop control. The second type is an unsteady
bleeding linked to the pressure fluctuations upstream of the compressor face. This is
referred to as closed-loop control. The bleed valves in both cases were located one tip-
chord downstream of the rotor blades. This chapter discusses the CFD results for the two
control methods applied to the NASA Rotor 67 compressor. In Section 6.1, results for
106
the open-loop control are presented. Section 6.2 deals with the results for the closed-loop
control.
6.1 OPEN-LOOP CONTROL STUDIES
In open-loop control, a fraction of the mass flow rate is removed through a valve
placed on the compressor casing. Such a steady bleeding is inefficient and must be turned
off during design operations. A bleed valve will have its own throttle characteristic,
which can be tailored to enhance the stable operation of the compressor. A schematic of
the steady bleeding is shown in Figure 3.6.
In this study, to simulate the open-loop control, it was assumed that flow is removed
at a constant rate in an azimuthally uniform rate. The fraction of mass flow rate and the
locations of the bleed valves are specified by the user. As pointed out in Section 3.6, the
normal velocity component is calculated from the amount of the bleed flow and the bleed
area, as shown in Equation (3-57).
To see the effects of the open-loop control on the flow field, a 3.2% of the mean flow
bleed was applied to the unstable operating condition at point C, shown in Figure 6.1.
Figure 6.2 shows the velocity vectors at the azimuthal plane at midpitch with the passive
bleed control after 1.5 rotor revolutions. It may be recalled that the compressor
experienced modified stall at the unstable branch (B-C) of the compressor map. A
comparison between Figures 6.2 and 5.10 (reproduced here for convenience) shows that
the compressor casing is seen to restore attached flow over much of the whole
compressor.
107
Figure 6.3 shows the fluctuations in mass flow rate and total pressure ratio for this
steady bleed condition. The computed mean mass flow rate entering the plenum was
30.4 kg/sec, which is 87.5% of the choked flow rate, and the mean total pressure ratio
was 1.65. This operating point is shown in the Figure 6.1, indicated by point D. In
Figure 6.1, the horizontal axis is the mass flow rate through the compressor before
bleeding, c m
.
, and the vertical axis is the total pressure ratio for the whole system
including the bleeding. As can be seen, a 3.2% bleed brings the compressor to a new
operating point, D. This is because removing some high-pressure air from the
compressor causes a reduction in the total pressure at the rotor exit, and reduces the total-
to-total pressure ratio of the rotor. Also, a comparison between Figure 5.9 and Figure 6.3
(reproduced here for convenience) reveal that this open-loop bleeding reduces the
amplitude of total pressure oscillations by 75%.
Referring back to the performance map with open-loop control shown in Figure
6.1, it is seen that the stable operation of the engine has been extended to 87.6% of the
choked mass flow rate, giving effectively a 38.2% increase in the operating range of the
compressor.
Another way to compare the controlled-operating point D in the performance map
with the other operating conditions is given in Figure 6.4. In this figure, the vertical axis
represents the non-dimensional pressure rise and the horizontal axis is the mass flow rate
entering the plenum, p m
.
, which is b c m m
. .
− , where b m
.
is the bleeding mass flow rate
through the bleed valve. A 3.2% bleed from the operating point C brings the compressor
108
to a new operating condition D. The reason for choosing the horizontal axis to be p m
.
is
to compare the throttle position for this point with the throttle positions for other points in
the compressor performance map. Under steady state conditions t p m m
. .
· , where t m
.
is
the throttle mass flow rate. From a knowledge of t m
.
and gage pressure ∆p at throttle
valve, the corresponding throttle characteristic K
t,
was calculated as follows:
p K m
t
t ∆ ·
.
(6-1)
The throttle characteristic map was obtained by using the calculated K
t
at point D and
Equation (6-1) for different mass flow rates, and is plotted as a dashed line in Figure 6.4.
In other words, the dashed line represents all the points having the same throttle position.
A comparison between Figure 6.1 and Figure 6.4 shows that even though the bleeding
was applied to point C, bringing the compressor to the operating point D, the operating
point D has the same throttle characteristic as point F in the performance map. From this
point of view, the branch B-D is a stable branch of the performance map. The useful
mass flow rate, representing flow that goes into downstream components, is 29.43 kg/sec.
This is a useful extension of the operating conditions. Without open-loop control the
system will enter into modified surge at mass flow rates below 31.6 kg/sec (point B).
To understand why and how the bleed-air mechanism works, the flow field was
examined. Figure 6.5 shows a comparison between the non-dimensional axial velocity at
109
point C, a stalled and unstable condition, after three different rotor cycles, and point D, a
stable condition achieved with bleed valve control. The data was acquired near the
leading edge at mid-passage of the compressor. It is seen that local reversed flow regions
first originate near the blade leading edge. If left unchecked, this flow region grew
temporally and spatially, leading the compressor to an uncontrolled growth of mass flow
rate and pressure fluctuations. When 3.2% of the total flow rate entering the inlet was
removed from the compressor, the reversed flow regions near the leading edge
disappeared and a stable operating condition was restored
As in operating points B and C, four “numerical” velocity probes were implemented
circumferentially at the upstream of the compressor face, shown in Figure 5.3, to trace
the axial velocity component. The spatially averaged values of axial velocity and static
pressure from the probes readings are shown in Figure 6.6 and 6.8. Compared with
results for point C in the Figure 5.12 (reproduced as part of Figure 6.6), the probes
showed only very small fluctuations and reversed flow was completely eliminated.
Figure 6.7 illustrates the deviation of axial velocity from the spatially averaged axial
velocity for all of the four probes. The time traces shown in this figure are shifted
vertically by a constant interval. It is observed that all of the probes are identical and no
circumferential disturbances indicative of rotating stall were observed. A comparison
between Figure 5.14 and Figure 6.7 (reproduced here for convenience) reveal that the
open-loop control eliminates the stall cells. Figures 6.6 through 6.8 show that the
compressor is in stable operation.
110
6.2 CLOSED-LOOP CONTROL STUDIES
In the previous section, the effects of open-loop control on the flowfield were
discussed. As mentioned earlier, such a steady preset bleeding is inefficient and must be
turned off when the compressor operates close to design conditions. Therefore, a closed-
loop active control methodology was studied. The closed-loop control was applied to the
NASA Rotor 67 configuration, and results are presented in this section.
In this approach, the bleed rate was linked to pressure amplitudes upstream of the
compressor face, which monitored by a control unit. This control unit received the
pressure signals from some probes located upstream of the compressor face, determined
the amount of bleeding needed, and activated the bleed valve. The mass flow of the
bleed valve is linked to the instantaneous amplitude of pressure fluctuations acquired by
the “computational” probes. The bleed valve was activated whenever the pressure
sensors in the upstream of compressor face experienced high-pressure fluctuations. The
amount of air that is removed, b m
.
, is related to the gage pressure ∆p at the bleed
location:
p A K m
b b
b ∆ ·
.
(6-2)
In Equation (6-2), K
b
is the bleed valve constant and is related to the valve geometry.
Quantity A
b
is the bleed valve area and
111

− · ∆ p p p
b
(6-3)
where
b
p and

p are the pressure at the bleed location and ambient pressure,
respectively. The pressure, density, and the two components of tangential velocity were
extrapolated from the interior. The normal velocity at the bleeding points was calculated
using Equation (6-2):
b
b
n
p k
u
ρ

· (6-4)
Figure 3.7 illustrates the schematic of the closed-loop diffuser-bleed valve control
used in this study.
The following parameters, which are user inputs, identify the active bleed
configuration:
- Bleeding locations
- Pressure sensor locations
- Bleed characteristic constant
- Permitted upper and lower limits of pressure for the pressure sensors
In this study, the “numerical” probes were located at 30% tip chord upstream of the
compressor face. The bleed valve was placed one tip chord downstream of the rotor. As
in open-loop bleeding, the bleeding under feedback control was circumferentially
uniform on each of the flow passages. At each time step, if the pressure values at the
112
probes were out of the permitted fluctuation limits, the bleed valve removed some air
from the compressor.
The closed-loop control was applied to point C, in Figure 6.1. Figure 6.9 shows the
circumferential averaged pressure fluctuation of the four probes at operating point C
without bleed control. As pointed out earlier, three levels of amplitudes were observed.
The bleed valve was not activated for the first two lower amplitude levels, the recovery
and precursor levels. As shown in Figure 6.8, the upper permitted limit of the inlet
pressure fluctuations was set to 0.67P

and the lower one was chosen 0.54P

. The
characteristic constant value K
b
in Equation (6-2) was chosen to be 3.0. This value was
chosen to ensure the rate of mass bleeding flow was less than 2% of the mean flow.
Figure 6.10 shows the fluctuations of total pressure ratio versus mass flow rate
fluctuations. This operating condition is referred as point E in the performance map,
Figure 6.1. A comparison with Figure 5.9 shows that the amplitude of fluctuations were
reduced by 30% and 40% in total pressure ratio and mass flow rate, respectively, when
closed-loop control was used. However, the reduction in the total pressure fluctuations
was larger for open-loop control. As mentioned earlier, the horizontal axis in Figure 6.1
is the mass flow rate through the compressor before bleeding. Figure 6.1 shows that the
closed-loop approach restores the stable operating condition at a higher pressure ratio.
Figure 6.11 shows the pressure rise across the compressor stage versus the mass flow
rate entering the plenum. In this figure, a methodology similar to the one used in open-
loop control was used to compute the throttle characteristic, K
t
, and plot the throttle
113
characteristic map. From Figure 6.11, it can be seen that the controlled operating point E
has the same throttle position as the unstable operating point G. It should be mentioned
that the air bleeding rates for operating points D and E were different. At the operating
point D (open-loop control) the mass flow rate p m
.
was 30.41 kg/sec, whereas at point E
(closed-loop control), the mass flow rate was 30.03 kg/sec. The averaged bleed flow rate
at operating point E was about 1.8 percent of the mean flow rate, where at point D, it was
3.2%. Also, from Figure 6.11, it can be concluded that the useful mass flow rate can be
as low as 29.49 kg/sec without instabilities.
The spatial average of axial velocity from the probes’ readings is shown in Figure
6.12. Compared to results for point C in the Figure 5.14, the amplitude of the axial
fluctuations was reduced, and no reversed flow was observed. The fluctuations were
large compare to the 3.2% steady bleeding case. Figure 6.13 illustrates the deviation of
axial velocity from the spatially averaged axial velocity shown in Figure 6.12. The
velocity time traces shown in this figure are shifted vertically by a constant interval. It is
observed that the probes show small circumferential disturbances. The amplitude of
these disturbances was only 20% of the disturbance amplitude for the unstable operating
point C, shown in Figure 5.14. From Figures 6.10 and 6.13, it may be concluded that the
compressor is in a stable operating condition at point E.
114
The table below summarizes the calculations reported in this study. It is clear that
both open-loop and closed-loop controls provide a useful extension of the stall margin in
the system.
Cases studied
c m
.
(kg/sec)
b m
.
(kg/sec)
b c p m m m
. . .
− ·
(kg/sec)
% o Extension of
operability margin
100
. .
. .
x
m m
m m
Stall of Onset Choked
c Stall of Onset


Stalled condition
operating point C 29.4 ____ 29.4 Unstable
Open-loop control
operating point D 30.41 0.973 29.437 38.2
Closed-loop control
operating point E 30.03 0.541 29.49 50.3
Table 6.1 Summary of stall margin extension for the two control schemes studied.
115
Choked
c
m
m
.
.
1.3
1.35
1.4
1.45
1.5
1.55
1.6
1.65
1.7
1.75
1.8
0.82 0.84 0.86 0.88 0.9 0.92 0.94 0.96 0.98 1
T
o
t
a
l

P
r
e
s
s
u
r
e

R
a
t
i
o
CFD without Control
CFD with Open-Loop Control
CFD with Close-Loop Control
Experiment
A
B
C
D
E
Stable Controlled
Conditions
Figure 6.1 Characteristic performance map with bleed control (Rotor 67).
Plenum
Chamber
c
m&
t
m&
P
m&
b
m&
116
Without control, operating point C.
Figure 6.2 Velocity profile with/without open-loop control (Rotor 67).
3.2% Bleed
With open-loop control, operating
point D.
117
Figure 6.3 Total pressure ratio and mass flow rate fluctuations with/without open-
loop control (Rotor 67).
-50
-30
-10
10
30
50
-40 -30 -20 -10 0 10 20 30 40
% Pressure
Fluctuations
% Mass Flow Rate Fluctuations
Without control, operating point C
% Mass Flow Rate Fluctuations
% Pressure
Fluctuations
With open-loop control, operating point D
-50
-30
-10
10
30
50
-40 -20 0 20 40
118
Figure 6.4 Characteristic performance map with throttle characteristic for open-
loop control (Rotor 67).
Plenum
Chamber
c
m&
t
m&
P
m&
b
m&
0.6
0.7
0.8
0.9
1
1.1
1.2
28 29 30 31 32 33 34 35
CFD without Control
CFD with Open-Loop Control
Throttle Characteristic
sec) / (
.
kg mp
ref
P
P ∆
A
B
C
D
F
119
Figure 6.5 Spanwise distribution of axial velocity at mid-passage near the leading
edge (Rotor 67).
0
20
40
60
80
100
-0.40 -0.20 0.00 0.20 0.40 0.60 0.80
U/Uinf
After 1.5
revolution
After 0.5
revolution
3.2%
bleeding
120
Figure 6.6 Time history of axial velocity with/without open-loop control (Rotor 67).
Rotor Revolution, Ωt/2π
Without control, operating point C

a
u
Rotor Revolution, Ωt/2π
With open-loop control, operating point D

a
u
121
Figure 6.7 Time history of axial velocity deviation from its azimuthally averaged
axial velocity with/without open-loop control (Rotor 67).
Rotor Revolution, Ωt/2π
Without control, operating point C


a
u u
Rotor Revolution, Ωt/2π
With open-loop control, operating point D


a
u u
122
Figure 6.8 Time history of static pressure under open-loop control (Rotor 67).
Figure 6.9 Upper and lower limit of pressures that trigger closed-loop control
(Rotor 67).
Rotor Revolution, Ωt/2π
ref
P
P
Rotor Revolution, Ωt/2π
ref
P
P
Pressure lower limit
Pressure upper
limit
123
Figure 6.10 Total pressure ratio and mass flow rate fluctuations under closed-loop
control (Rotor 67).
Figure 6.11 Characteristic performance map with throttle characteristic for closed-
loop control (Rotor 67).
-50
-30
-10
10
30
50
-30 -20 -10 0 10 20
% Mass Flow Rate Fluctuations
% Pressure
Fluctuations
0.6
0.7
0.8
0.9
1
1.1
1.2
28 29 30 31 32 33 34 35
CFD without Control
CFD with Close-Loop Control
Throttle Characteristic
A
B
C
E
G
sec) / (
.
kg mp
ref
P
P ∆
124
Figure 6.12 Time history of axial velocity under closed-loop control (Rotor 67).
Figure 6.13 Time history of axial velocity deviation from its azimuthally averaged
axial velocity upstream of the compressor face under closed-loop control
(Rotor 67).
Rotor Revolution, Ωt/2π


a
u u
Rotor Revolution, Ωt/2π

a
u
125
7 CHAPTER VII
CONCLUSIONS AND RECOMMENDATIONS
In the present work, a three-dimensional unsteady Navier-Stokes analysis capable of
modeling multistage turbomachinery components has been developed. This method uses
a cell-vertex finite volume scheme to solve the governing equations in a body-fitted
rotating coordinate system. The scheme is third order accurate in space and first or
second order in time. A one-equation Spalart-Allmaras turbulence model equation was
implemented in the solver to account for the turbulence effects. The analysis can also use
a previously stored solution file with the corresponding grid file, to restart the
computations.
A number of code-validation studies have been done for centrifugal and axial
configurations. Studies have also been done to understand some of the mechanisms
behind rotating stall and surge. Open-loop bleed control and closed-loop control methods
have been studied and are demonstrated to stabilize the system. To the author’s
knowledge, this is the first application of a fully 3-D unsteady viscous flow analysis for
modeling axial compressor control in this fashion.
In this chapter, the conclusions of this research are presented in Section 7.1. The
recommendations for further work in the future are given in Section 7.2.
126
7.1 CONCLUSIONS
The following items are concluded from the current work:
1- Three-dimensional unsteady flow solvers such as the one developed here are
valuable tools for modeling and understanding rotating stall and surge. These
methods are suitable for identifying “hot spots” in the flow field, for redesigning
the system (e.g. casing treatment), and for developing flow control devices for
the system.
2- The plenum-chamber downstream plays a critical role in surge development.
The exchange of energy between the plenum chamber and the flow inside the
blade passages is a major mechanism for surge. Therefore, devices and
mechanisms that directly alter the pressure levels in the plenum chamber are an
effective way of controlling surge. This observation has been numerically
demonstrated in this work.
3- Inlet disturbances can trigger rotating stall. The stall cells in the present
simulation appear to rotate at 38% of the rotor RPM, in the direction of the rotor
rotation. This phenomenon is consistent with experimental observations. It
should be possible to mitigate rotating stall through careful redesign of the inlet,
through guide vanes, or through active control (e.g. jets).
4- Rotating stall and surge phenomena are often accompanied by temporally
growing pressure and velocity fluctuations in the inlet upstream of the
compressor face. These oscillations may roughly be divided into these
127
categories: precursor, stall, and recovery. It should therefore be possible to use
these oscillations, in particular the precursors, to trigger open-loop or closed-loop
control. While this has been known for some time, this is the first time precursor
waves have been used in a numerical simulation to detect and automatically
control rotating stall, surge, and modified surge.
5- It should be possible to establish simple rule-based control laws linking the inlet
pressure fluctuations to the amount of actuation needed. A very simple law that
triggers bleeding whenever the upstream pressures exceeded a preset level user
was found to be effective. This law also accounted for the bleed valve
characteristics.
7.2 RECOMMENDATIONS
The following items are suggested to extend the current work:
1- The outlet boundary conditions at off-design conditions play a dominant role in
simulating the instability developments appropriately. In this work, a simple
unsteady boundary condition was introduced, and a nominal fixed mass flow rate
controlled the compressor operation. Another type of outlet boundary condition, that
should be tested, is to let this nominal mass flow through the throttle also fluctuate
with time. In this manner both backpressure and throttle mass flow vary at each time
step. This boundary may be written as:
128

) (
2
t c
p
p p
m m
V
a
dt
dp
& & − ·
(7-1)
P A K m
t t
t ∆ ·
.
(7-2)
where K
t
is the throttle characteristic constant and is related to the throttle geometry,
and A
t
is the throttle valve area. Here, ∆P is the pressure difference between the
plenum pressure or compressor backpressure and the components downstream, e.g. a
second stage or a combustor. This may be done stage by stage, providing a weak 1-D
coupling between the various components of the compression system and the
combustor.
2- This work focused on the analysis of flowfield and the fluid dynamic phenomena in
single stage compressors with multiple blade passages. The work should be
extended to the simulation of multi stage compressors considering rotor-stator
interactions. In rotor-stator interactions, the rotor grid block, similar to the present
simulation of the rotors, will be moving in the pitchwise direction, while the stator
grid block remains stationary. At the boundaries connecting the moving and
stationary regions, an interpolation approach may be used to transfer the information
from the stationary block to the rotating block.
3- In the present research, the open-loop bleeding control removed a fraction of high-
pressurized air at a fixed rate from the compressor. The bleed rate can be
129
sinusoidally varied at the rotating stall frequency with an appropriate phase lag. The
bleeding mass flow rate, b m
.
, may be written as follows:
) sin(
.
Mean
. .
ϕ ω + + · t m m m
s
f b b
(7-3)
Here, Mean
.
b m is the mean mass flow removed from the compressor, f m
.
is the
amplitude of the pulsed bleeding, ω
s
is the rotating speed of the stalled cells, and ϕ
is the phase lag. The benefit of this methodology is that the controller removes the air
only when a stalled cell passes by the bleed valve.
4- The present approach underestimates the total pressure rise across the compressor.
The one-equation turbulence model may be a source of this discrepancy. Although a
limited number of grid sensitivity studies have been done, an inadequate number of
grid points may also have contributed to this error. These discrepancies must be
systematically studied and eliminated. Thanks to the availability powerful high-
performance computers, one near-term possibility is the use of LES methodology on
sufficiently fine grids.
5- Different types of control devices, such as inlet guide vanes, casing treatment, and so
on, may be simulated with this solver and their effects on the flowfield can be
investigated. Recently, air injection control methodology has been computationally
simulated
70
. Experimental evidence also exists that shows that air injection may
130
reduce the required rate of bleeding
40
. This work should be extended to a systematic
study of these concepts.
In summary, a first principles-based methodology for modeling steady-state and off-
design operation of axial compressors has been developed. This method has been
validated and used to study stall and surge control concepts. It is hoped that this work
will serve as a useful step for future investigations in the exciting area of non-linear flow
control.
131
8 REFERENCES
1. Horlock, J. H., Axial Flow Compressors, Butterworths Scientific Publications,
London, 1958.
2. Gostelow, J. P., Cascade Aerodynamics, Pergamon Press, New York, NY, 1984.
3. Sebghati, J. M., Monitoring Axial-Flow Compressor Rotating Stall and Surge,
Master’s Thesis, The University of Tennessee, Knoxville, May 1992.
4. Ferguson, T. B., The Centrifugal Compressor Stage, Butterworths, London, 1963.
5. Japikse, D. and Baines, N. C., Introduction to Turbomachinery, Concepts ETI,
Inc. and Oxford University Press, 1994.
6. Cherkassky, V. M., Pumps, Fans, Compressors, Mir Publications, Moscow, 1980.
7. Cohen, H., Rogers, G. F. C. and Saravanamuttoo, H. I. H., Gas Turbine Theory,
4th ed., Longman, Essex, 1996.
8. Hill, P. G., and Peterson, C. R., Mechanics and Thermodynamics of Propulsion,
2nd ed., Addison Wesley Publication Co., 1992.
9. ¥cer, A. S., Stow, P., and Hirsch, C., Thermodynamics and Fluid Mechanics of
Turbomachinery, Vol. I and II, 1985.
10. Cumpsty, N. A., Compressor Aerodynamics, Longman Scientific and Technical,
Essex, 1989.
11. Gravdahl, J. T. and Egeland, O., Compressor Surge and Rotating Stall Modeling
and Control, Springer, Verlay London Limited, 1999.
12. Pampreen, R. C., Compressor Surge and Stall, Concepts ETI, Inc., Norwich,
Vermont, 1993.
132
13. de Jager, B., “Rotating Stall and Surge Control,” Proceedings of the 34
th
Conference on Decision & Control, New Orleans, LA, Dec. 1995.
14. Badmus, O. O., Nonlinear Dynamic Analysis and Control of Surge and Rotating
Stall in Axial Compression System, Ph.D. Dissertation, Georgia Institute of
Technology, March 1994.
15. Stodola, A., Steam and Gas Turbine, McGraw-Hill Book Co., New York, 1927.
16. Kearton, W. J., Turbo-Blowers and Compressors, Sir Isaac Pitman and Sons
Publishers, London, 1931.
17. Fischer, K. and Thoma, D., “Investigations of the Flow Conditions in a
Centrifugal Pump,” Transactions of ASME, Vol. 54, pp. 141-155, Nov. 1932.
18. Emmons, H. W., Pearson, C. E. and Grant, H. P., “Compressor Surge and Stall
Propagation,” Transactions of ASME, Vol. 77, pp. 455-469, 1955.
19. Greitzer, E. M., “Surge and Rotating Stall in Axial Flow Compressor: Part 1,
Theoretical Compression System Model, and Part 2, Experimental Results and
Comparison with Theory,” ASME Journal of Engineering for Power Transactions
of ASME, Vol. 98, pp. 190-217, April 1976.
20. Moore, F. K., “A Theory of Rotating Stall in Multistage Compressor, Parts I-III,”
ASME Journal of Engineering for Power, No. 106, pp. 313-336, 1984.
21. Moore, F. K., “Stall Transient of Axial Compression Systems with Inlet
Distortion,” Journal of Propulsion and Power, Vol. 2, No. 6, pp. 552-561, Nov.-
Dec. 1986.
22. Hynes, T. P. and Greitzer, E. M., “A Method for Assessing the Effects of
Circumferential Flow Distortion on Compressor Stability,” Journal of
Turbomachinery, No. 109, pp. 371-380, July 1987.
133
23. Moore, F. K. and Greitzer, E. M., “ A Theory of Post-Stall Transients in Axial
Compression Systems Part I-Development of Equations,” Journal of Engineering
for Gas Turbines and Power, Vol. 108, pp. 68-76, Jan. 1986.
24. Greitzer, E. M and Moore, F. K., “ A Theory of Post-Stall Transients in Axial
Compression Systems Part II-Application,” Journal of Engineering for Gas
Turbines and Power, Vol. 108, pp. 231-239, April 1986.
25. Davis, M. W. and O’Brien, W. F., “Stage-by Stage Post Stall Compression
System Modeling Technique,” Journal of Propulsion and Power, No. 7, pp. 997-
1005, 1991.
26. Eveker, K. M., Model Development for Active Control of Stall Phenomena in
Aircraft Gas Turbine Engines, Ph.D. Dissertation, Georgia Institute of
Technology, 1993.
27. Bonnaure, L. P. Modeling High Speed Multistage Compressor Stability, Master’s
Thesis, Massachusetts Institute of Technology, 1991.
28. McCaughan, F. E., “Application of Bifurcation Theory to Axial Flow Compressor
Instability,” Journal of Turbomachinery, No. 111, pp. 426-433, Oct. 1989.
29. McCaughan, F. E., “Numerical Results for Axial Flow Compressor Instability,”
Journal of Turbomachinery, No. 111, pp. 434-441, Oct. 1989.
30. Leonessa, A., Haddad, W. M. and Li, H., “Global Stabilization of Centrifugal
Compressors via Stability-Based Switching Controller,” Proceedings of the IEEE
International Conference Control Applications (CCA) and IEEE International
Symposium on Computer Aided Control System Design (CACSD), Vol. 2, pp.
1383-1388, Aug. 1999.
31. Botros, K. K. and Henderson, J. F., “Developments in Centrifugal Compressor
Surge Control-A Technology Assessment,” Journal of Turbomachinery, No. 116,
pp. 240-249, 1994.
134
32. Bailey, E. E. and Voit, C. H., “Some Observations of Effects of Porous Casing on
Operating Range of a Single Axial-Flow Compressor Rotor,” NASA TM X 2120,
1970.
33. Holman, F. F. and Kidwell, J. R., “Effects of Casing Treatment on a Small
Transonic Axial-Flow Compressor,” ASME Paper 75-WA/GT-5, 1975.
34. Lee, N. K. W. and Greitzer, E. M., “Effects of Endwall Suction and Blowing on
Compressor Stability Enhancement,” Journal of Turbomachinery, Vol. 112, pp.
133-144, 1990.
35. Khalid, S., “A Practical Compressor Casing Treatment,” ASME Paper 97-GT-
375, April 1997.
36. Whitfield, A. Wallace, F. J. and Atkey, R. C., “The Effect of Variable Geometry
on the Operating Range and Surge Margin of a Centrifugal Compressor,” ASME
Paper 76-GT-98, Jan. 1976.
37. Jones, B. A., “Single Stage Experimental Evaluation of Variable Geometry Inlet
Guide Vanes and Stator Blading. Part VI Final Report,” NASA CR-54559, 1970.
38. Pampreen, R. C., “The Use of Variable Inlet Guide Vanes for Automotive Gas
Turbine Engine Augmentation and Load Control,” SAE Paper 760285, 1976.
39. Sheridan, D. C., Nordenson, G. E. and Amann, C. A., “Variable Compressor
Geometry in the Single-Shaft Automotive Turbine Engine,” SAE Preprint
740166, 1974.
40. Yeung, S. and Murray, R. M., “Reduction of Bleed Valve Requirements for
Control of Rotating Stall Using Continues Air Injection,” Proceedings of the
1997 IEEE International Conference on Control Application, Hartford, CT, pp.
683-690, Oct. 1997.
41. Fisher, F. B., “Application of Map Width Enhancement Devices to Turbocharger
Compressor Stages,” SAE Paper 880794, April 1988.
135
42. Pinsley, J. E., Guenette, G. R., Epstein, A. H., and Greitzer, E. M., “Active
Stabilization of Centrifugal Compressor Surge,” Journal of Turbomachinery, No.
113, pp. 723-732, 1991.
43. Epstein, A. H., Ffows Williams, J. E. and Greitzer, E. M, “Active Suppression of
Aerodynamic Instabilities in Turbomachinery,” Journal of Propulsion and Power,
No. 5, pp. 204-211, 1989.
44. Iura, T., and Rannie, W. D., “Experimental Investigations of Propagating Stall in
Axial Compressors,” Transactions of the ASME, Vol. 116, pp. 463-471, 1954.
45. Longley, J. P., “A Review of Nonlinear Flow Models for Compressor Stability,”
ASME Journal of Turbomachinery, pp. 202-215, 1994.
46. Paduano, J. D., “Active Control of Rotating Stall in Axial Compressors,” Ph.D.
Dissertation, Massachusetts Institute of Technology, 1991.
47. Gysling, D. L., Dugundji, J., Greitzer, E. M. and Epstein, A. H., “Dynamic
Control of Centrifugal Compressors Using Tailored Structures,” ASME Paper 90-
GT-122, June 1990.
48. Lawless, P., and Fleeter, S., “Active Unsteady Aerodynamic Suppression of
Rotating Stall in an Incompressible Flow Centrifugal Compressor with Vaned
Diffuser,” AIAA/SAE/ASME/ASEE Joint Propulsion Conference, AIAA Paper
91-1898, 1991.
49. Toyama, K., Runstadler, P. W., and Dean, R. C., “An Experimental Study of
Surge in Centrifugal Compressors,” Journal of Fluids Engineering Transaction of
ASME, Vol. 99, pp. 115-131, 1977.
50. Takata, H., and Nagano, S., “Nonlinear Analysis of Rotating Stall,” Journal of
Engineering for Power Transactions of ASME, pp. 279-293, 1972.
136
51. Davis, R. L., Hobbs, D. E., and Weingold, H. D., “Prediction of Compressor
Cascade Performance Using a Navier-Stokes Technique,” ASME Paper 88-GT-
96, June 1986.
52. Chevrin, P. A., and Vuillez, C., “Viscous Flow Computations in Turbomachine
Cascades,” ASME Paper 90-GT-76, June 1990.
53. Cambier, L., Escande, B., and Veuillot, J. P., “Computation of Internal Flows at
High Reynolds Number by Numerical Solution of the Navier-Stokes Equations,”
Rech. Aerosp., No. 1986-6, pp. 27-44, 1986.
54. Boyle, R. J., “Navier-Stokes Analysis of Turbine Blade Heat Transfer,” ASME
Paper 90-GT-42, June 1990.
55. Srivastava, R., and Sankar, Lakshmi N., “Efficient Hybrid Scheme for the
Analysis of Counter-Rotating Propellers,” Journal of Propulsion and Power, Vol.
9, No. 3, pp. 382-388, May-Jun 1993.
56. Dawes, W. N., “ Numerical Study of the 3D Flowfield in a Transonic Compressor
Rotor with a Modeling of the Tip Clearance Flow,” AGARD Conference
Proceedings, No. 401, Neuilly Sur Seine, March 1987.
57. Hah, C., and Wennerstrom, A. J., “Three-Dimensional Flowfields Inside a
Transonic Compressor with Swept Blades,” Journal of Turbomachinery, Vol. 113,
pp. 241-251, 1991.
58. Adamczyk, J. J., Mulac, R. A., and Celestina, M. L., “ A Model for Closing the
Inviscid Form of the Average Passage Equations,” ASME Paper 86-GT-227,
June 1986.
59. Hall, E. J., “Aerodynamic Modeling of Multistage Compressor Flowfields - Part
1: Analysis of Rotor/Stator/Rotor Aerodynamic Interaction,” ASME Paper 97-
GT-344, 1997.
137
60. Hathaway, M. D., and Wood, J. R., “Application of a Multi-Block CFD Code to
Investigate the Impact of Geometry Modeling on Centrifugal Compressor Flow
Field Predictions, ” Journal of Turbomachinery, Vol. 119, pp. 820-830, Oct.
1997.
61. Wood, J. R., Adam, P. W., and Buggele, A. E., “NASA Low Speed Centrifugal
Compressor for Fundamental Research,” NASA TM 83398, 1983.
62. Chima, R. V., and Yokota, J. W., “Numerical Analysis of Three-Dimensional
Viscous Internal Flow,” AIAA Journal, Vol. 28, No. 5, pp. 798-806, May 1990.
63. Rivera, C. J., “Numerical Simulation of Dynamic Stall Phenomena in Axial Flow
Compressor Blade Rows,” Ph.D. Dissertation, Georgia Institute of Technology,
1998.
64. Escuret, J. F., and Garnier, V., “Numerical Simulations of Surge and Rotating-
Stall in Multi-Stage Axial-Flow Compressors,” AIAA Paper 94-3202, 1994.
65. He, L., “Computational Study of Rotating Stall Inception in Axial Compressor,”
Journal of Propulsion and Power, Vol. 13, No. 1, pp. 31-38, Jan.-Feb. 1997.
66. Niazi, S., Stein, A., and Sankar, L. N., “Development and Application of a CFD
Solver to the Simulation of Centrifugal Compressors,” AIAA Paper 98-0934, Jan.
1998.
67. Stein, A., Niazi, S., and Sankar, L. N., “Computational Analysis of Stall and
Separation Control in Centrifugal Compressors,” Journal of Propulsion and
Power, Vol. 16, No. 1, pp. 65-71, Jan. 2000,
68. Stein, A., Niazi, S. and Sankar, L. N., “Numerical Analysis of Stall and Surge in a
High-Speed Centrifugal Compressor,” AIAA Paper 2000-0226, Jan. 2000.
69. Niazi, S., Stein, A., and Sankar, L. N., “ Numerical Study of Surge and Stall
Alleviation in a High-Speed Transonic Fan Rotor,” AIAA Paper 2000-0225, Jan.
2000.
138
70. Stein, A., Computational Analysis of Stall and Separation Control in Centrifugal
Compressors, Ph.D. Dissertation, Georgia Institute of Technology, Aerospace
Engineering, May 2000.
71. Anderson, D. A., Tannehill, J. C., and Pletcher, R. H., Computational Fluid
Mechanics and Heat Transfer, 2nd

ed., McGraw-Hill, New York, 1984.
72. Hirsch, C., Numerical Computation of Internal and External Flows, Vol. I and II,
1st ed., Wiley, New York, 1988.
73. Roe, P. L., “Approximate Reimann Solvers, Parameter Vectors and Difference
Schemes,” Journal of Computational Physics, Vol. 135, pp. 250-258, 1981.
74. Roe, P. L. and Pike, J., “Efficient Construction and Utilization of Approximate
Riemann Solutions,” Computing Methods in Applied Sciences and Engineering,
Vol. 6, Elsevier Science Publishers, INRIA, pp. 499-518, 1984.
75. Roe, P. L., “Discrete Models for the Numerical Analysis of Time-Dependent
Multidimensional Gas Dynamics,” NASA Contract Report CR-172574, Mar.
1985.
76. Roe, P. L., “Some Contributions to the Modeling of Discontinuous Flows,”
Lectures in Applied Mathematics, Vol. 22, pp. 163-193, 1985.
77. Roe, P. L., “Characteristic-Based Schemes for the Euler Equations,” Annual
Review of Fluid Mechanics, Vol. 18, pp. 337-365, 1986.
78. Roe, P. L., “Some Contributions to the Modeling of Discontinuous Flows,”
Large-Scale Computations in Fluid Mechanics, Edited by Engquist, B. E., Osher
S. and Somerville, R. C. J., Vol. 22, Part. 2, Lectures in Applied Mathematics,
ASME, Providence, RI, pp. 163-193, 1985.
79. Liu, Y. and Vinokur, M., “Upwind Algorithms for General Thermo-Chemical
Nonequilibrium Flows,” AIAA Paper 89-0201, Jan. 1989.
139
80. Pulliam, H. P. and Steger, J. L., “Implicit Finite-Difference Simulations of Three-
Dimensional Compressible Flow,” AIAA Journal, Vol. 18, pp. 159-167, Feb.
1980.
81. Spalart, P. R., and Allmaras, S. R., “A One-Equation Turbulence Model for
Aerodynamic Flows,” AIAA Paper 92-0439, 1992.
82. Strazisar, A. J., Wood, J. R., Hathaway, M. D., Suder, K. L., “Laser Anemometer
Measurements in a Transonic Axial Flow Fan Rotor,” NASA TP-2879, Nov.
1989.
83. Moore, R. D. and Reid, L., “Performance of Single-Stage Axial-Flow Transonic
Compressor with Rotor and Stator Aspect Ratios of 1.19 and 1.26 Respectively,
and with Design Pressure Ratio of 2.05,” NASA Paper TM-1659, 1980.
84. Reid, L. and Moore, R. D., “Design and Overall Performance of Four Highly
Loaded, High-Speed Inlet Stages for an Advanced High-Pressure-Ratio Core
Compressor,” NASA Paper TM-1337, 1978.
85. Adamczyk, J. J., Celestina, M. L., and Grietzer, E. M., “The Role of Tip
Clearance in High-Speed Fan Stall,” Transactions of the ASME, Vol. 115,
January 1993.
86. Chima, R. V., “Calculation of Tip Clearance Effects in a Transonic Compressor
Rotor,” NASA TM 107216, May 1996.
87. Pierzga, M. J. and Wood, J. R., “Investigation of the Three-Dimensional Flow
Field Within a Transonic Fan Rotor: Experiment and Analysis,” Transactions of
the ASME, Vol. 107, pp. 436-449, April 1985.
88. Arnone, A., “Viscous Analysis of Three-Dimensional Rotor Flow Using A
Multigrid Method,” ASME 93-GT-19, Sep. 1993.
140
89. Chima, R. V., “Viscous Three-Dimensional Calculations of Transonic Fan
Performance,” AGARD 77th Symposium on CFD Techniques for Propulsion
Applications, Paper No. 21, 1991.
90. Arima, T. Sonoda, T., Shirotori, M., Tamura, A., and Kikuchi, K., “A Numerical
Investigation of Transonic Axial Compressor Rotor Flow Using a Low Reynolds
Number k-ε Turbulence Model,” Transactions of the ASME, Paper 97-GT-82,
1997.
91. Hah, C., and Reid, L., “A Viscous Flow Study of Shock-Boundary Layer
Interaction, Radial Transport, and Wake Development in a Transonic
Compressor,” Transactions of the ASME, Vol. 114, July 1992.
92. Crook, A. J., Greitzer, E. M., Tan, C. S., and Adamczyk J. J., “ Numerical
Simulation of Compressor Endwall and Casing Treatment Flow Phenomena,”
Journal of Turbomachinery, Vol. 115, pp. 501-511, July 1993.
93. Suder, K, L., and Celesttina, M. L., “Experimental and Computational
Investigation of the Tip Clearance Flow in a Transonic Axial Compressor,”
ASME Paper 94-GT-365, June 1994.
94. Shabir, A., Zhu, J., and Celestina, M., “Assessment of Three Turbulence Models
in a Compressor Rotor,” ASME 96-GT-198, 1996.
95. Dalbert, P., and Wiss, D. H., “Numerical Transonic Flow Field Predictions for
NASA Compressor Rotor 37,” ASME Paper 95-GT-326, June 1995.
96. Bright, B. M., Qammar, H. K., and Hartley, T. T., “Dimension Determination of
Precursive Stall Events in a Single Stage High Speed Compressor,” NASA TM
107268, 1996.
97. Eveker, K. M., and Nett, C. N., “Model Development of Active Surge Control
Rotating Stall Avoidance in Aircraft Gas Turbine Engines,” Proceedings of the
American Control Conference, pp. 3166-3172, 1991.
141
98. Wang, Y. and Murray, R. M., “Effects of the Shapes of Compressor
Characteristics on Actuator Requirements for Rotating Stall Control,”
Proceedings of the 1998 American Control Conference, 1998.
99. Prasad, J. V., Neumeier, Y., and Krichene, A., “Active Control of Compressor
Surge Using a Real Time Observer,” Proceedings of the NATO Applied Vehicle
Technology Conference, Braunschweig, May 2000.
142
9 VITA
Saeid Niazi was born in Sepidan, Iran on May 17, 1961. He graduated from the
Shiraz University, Iran, with a Bachelor of Science (B.Sc.) degree in Mechanical
Engineering in September 1986. In August 1991, he graduated form the Shiraz
University in Iran with a Master of Science (M.Sc.) degree in Fluid Mechanic and Heat
Transfer Engineering. From September 1989 to August 1995, he taught at the
Hormozgan University in the city of Bandar Abbas, Persian Gulf, Iran. He was the Dean
of the central library of the Hormozgan University during the period 1992-1995. He
commenced his graduate work at Georgia Institute of Technology in September 1996.
He earned the degree of Doctor of Philosophy (Ph.D.) in Aerospace Engineering in July
2000. He is a member of the American Institute of Aeronautics and Astronautics.

NUMERICAL SIMULATION OF ROTATING STALL AND SURGE ALLEVIATION IN AXIAL COMPRESSORS

Approved: ______________________________ Lakshmi N. Sankar, Chairman

______________________________ J.V.R. Prasad

______________________________ Suresh Menon

______________________________ Stephen M. Ruffin

______________________________ Prasanna V. Kadaba

Date Approved _________________

ii

IN THE NAME OF GOD

Dedicated to my parents Behjat Golbahar Haghighi and Sadrollah Niazi for their most precious gift of selfless love

iii

Prasad for his helpful suggestions and technical expertise in the area of flow control.ACKNOWLEDGEMENTS Many people have touched my life not only from an academic point of view. P. M. The financial assistance given by Army Research Office under the Multidisciplinary University Research Initiative (MURI) on Intelligent Turbine Engine is gratefully acknowledged. iv . I would also like to thank the past and present members of CFD lab for their warm friendship and support during my administrative responsibility in the lab. Guanpeng Xu. Sankar. Thanks to my co-worker and my friend Alex Stein for his great efforts during code development and his steadfast support throughout the research. I am really honored and grateful to have had the opportunity to work under him. Dr. for his support throughout the research period. Menon. Many thanks to Dr. but both their friendship and spiritual support have been a source of encouragement and strength to complete this work. Zhong Yang. J. S. To: Justin Russell. R. V. Kadaba for their services as members of my reading committee and their valuable comments. V. His patience and kindness with his detailed knowledge about this research topic played a key role in the development of this work. Ruffin. I would also like to thank Dr. Yi Liu. my teacher and dissertation advisor. S. Mert Berkman. Mehmet Sahin. I would like to thank Dr. Ebru Usta. Lakshmi N. and Dr.

Thanks also to my friend and roommate Konstantin Ignatiev in the School of Materials and Science Engineering for providing me assistance with computer services. Jagoda. Terry M. to whom this work is dedicated. Masayoshi Senga. and the staff of the Aerospace Engineering Office: Revonda B. Parrott. and Gang Wang. A special note of thanks to my brother and friend.Zhijian Liu. Thanks also to all my friends whose advice and support made this achievement possible. for their generous support during the pursuit of my degree in Atlanta. Mullis. the graduate coordinator. Simpson for being always ready to help me. Gaines. Hilda and sweet Sara. Behjat Golbahar Haghighi and Sadrollah Niazi. I would like to thank all my sisters and brothers and their families for their great support during my study. I would like to thank Dr. J. Ojala. Hamid and his family. Margaret A. Loretta Carroll. Finally. v . Their constant encouragement and great unconditional love for myself and each of my sisters and brothers have been a source of inspiration in my life. I owe my greatest gratitude to my parents. Carole W. I would like to thank Catherine Moseley Matos for taking the time to read this dissertation and for her helpful suggestions. and Howard L.

TABLE OF CONTENTS

DEDICATIONS ACKNOWLEDGEMENTS TABLE OF CONTENTS LIST OF TABLES LIST OF FIGURES NOMENCLATURE SUMMARY

iii iv vi ix x xv xix

I

INTRODUCTION ..................................................................................................... 1
1.1 AN OVERVIEW OF COMPRESSOR OPERATIONS ................................................. 2 1.2 COMPRESSOR STABILITY......................................................................................... 4 1.3 OBJECTIVES AND ORGANIZATION OF THE PRESENT WORK.......................... 6

II

AN OVERVIEW OF COMPRESSOR INSTABILITY PHENOMENA ........... 12
2.1 FUNDAMENTALS OF ROTATING STALL ............................................................. 12 2.2 FUNDAMENTALS OF SURGE.................................................................................. 15 2.3 LITERATURE SURVEY OF STUDIES ON ROTATING STALL AND SURGE .... 17
2.3.1 EXPERIMENTAL STUDIES ON COMPRESSOR CONTROL..................................... 18 2.3.2 COMPUTATIONAL STUDIES OF COMPRESSOR PERFORMANCE

vi

AND CONTROL ............................................................................................................. 23

III MATHEMATICAL AND NUMERICAL FORMULATION ............................. 31
3.1 GOVERNING EQUATIONS ....................................................................................... 31 3.2 DISCRETIZATION OF THE GOVERNING EQUATIONS....................................... 35 3.3 NUMERICAL SOLUTION OF THE DISCRETIZED EQUATIONS......................... 41 3.4 TURBULENCE MODELING ...................................................................................... 44 3.5 INITIAL AND BOUNDARY CONDITIONS ............................................................. 47
3.5.1 INITIAL CONDITIONS .................................................................................................. 47 3.5.2 BOUNDARY CONDITIONS .......................................................................................... 48

IV CODE VALIDATION STUDIES .......................................................................... 58
4.1 DESCRIPTION OF THE CONFIGURATIONS STUDIED....................................... 59
4.1.1 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 67)............................................. 59 4.1.2 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 37)............................................. 61

4.2 PREVIOUSLY REPORTED STUDIES ON ROTOR 67 AND ROTOR 37 CONFIGURATIONS .................................................................................................. 62 4.3 PEAK EFFICIENCY RESULTS FOR ROTOR 67...................................................... 63

V

RESULTS AT OFF-DESIGN CONDITIONS...................................................... 83
5.1 ROTOR 67 RESULTS.................................................................................................. 84
5.1.1 ONSET OF STALL.......................................................................................................... 84 5.1.2 ROTATING STALL AND MODIFIED SURGE SIMULATIONS................................. 88

vii

5.2 ROTOR 37 SIMULATIONS ........................................................................................ 90

VI ACTIVE CONTROL STUDIES OF ROTOR 67 CONFIGURATION ........... 104
6.1 OPEN-LOOP CONTROL STUDIES ......................................................................... 106 6.2 CLOSED-LOOP CONTROL STUDIES .................................................................... 110

VII CONCLUSIONS AND RECOMMENDATIONS .............................................. 125
7.1 CONCLUSIONS......................................................................................................... 126 7.2 RECOMMENDATIONS ............................................................................................ 127

REFERENCES.............................................................................................................. 131

VITA ............................................................................................................................. 142

viii

LIST OF TABLES Table 6.1 Summary of stall margin extension for the two control schemes studied………………………………………………………...114 ix .

. Reference [12]........... 28 Figure 2............... 54 Figure 3........................ Reference [14]............. 55 Figure 3...........................4 Compressor outflow boundary conditions............7 Types of active and passive compressor control schemes.....................................................................................3 Part-Span rotating stall with different stalled cells................................. ........ 30 Figure 3...........6 Transient response of system in surge.....2 Transient response of system in rotating stall....... Reference [5]... 27 Figure 2............................................ 54 Figure 3... Reference [14]....5 Compressor map with the stalled flow characteristic.LIST OF FIGURES Figure 1....1 Control volume and cell-vertex grid points........................................3 Boundary conditions used in the axial compressor analysis...... ..................... ................................4 Operational stability......4 Full-Span rotating stall with different stalled cells........ 56 Figure 3...... 11 Figure 2........ 27 Figure 2.......2 Computation of inviscid flux terms on a cell face.............. 8 Figure 1... References [5]............ [11].......... Reference [13]..........................1 Rotating stall inception........................ 56 x .3 Typical compressor characteristic map for axial and centrifugal compressors. matching the compressor and throttle characteristics....... 29 Figure 2...................................... 28 Figure 2..... 9 Figure 1..............5 Periodic and zonal boundary conditions....2 Schematic diagram of changes in fluid properties and velocity through an axial compressor stage................1 Work input machinery classification....... Reference [12]........................................ 10 Figure 1... 10 Figure 1.......... 29 Figure 2........5 Effects of rotor RPM on compressor performance and stability......

..............13 Comparison between experiments and present predictions of relative Mach number contours at 30% span at peak efficiency (Rotor 67)..... ..6 Computational grid in the meridional plane for Rotor 67................... Reference [90].................. 76 Figure 4.....Figure 3.. 57 Figure 3... ........... ................ 74 Figure 4.......... 70 Figure 4... Reference [90]....6 Implementation of bleeding as an open-loop active control.................................9 Computational grid for Rotor 37.......... 75 Figure 4..... .... 71 Figure 4............5 Pitchwise computational grid for Rotor 67.........................7 NASA Rotor 37 configuration............ 74 Figure 4.4 Streamwise computational grid at midspan for Rotor 67............................ 73 Figure 4.....................73 Figure 4...... 71 Figure 4.. .........8 Laser anemometer and aerodynamic locations for Rotor 37...................................... Reference [84]........... ..................................2 Laser anemometer and aerodynamic locations for Rotor 67................. . ................................... ......................... 72 Figure 4.............................3 Computational grid for Rotor 67..............................................1 NASA Rotor 67 configuration.................. 75 Figure 4......................... ........................... Reference [82].... 57 Figure 4.....10 Time history of mass flow rate fluctuations about mean flow at peak efficiency conditions (Rotor 67)....... 77 xi ................. ........ 72 Figure 4........70 Figure 4.....11 Pressure and mass flow rate fluctuations at peak efficiency conditions (Rotor 67)........14 Comparison between experiments and present results of relative Mach number at 90% span and 30% pitch at peak efficiency (Rotor 67)......................12 Comparison between experiments and present results of inlet axial velocity at peak efficiency conditions (Rotor 67).....7 Implementation of bleeding as a closed-loop active control.....

.. .....7 Time history of averaged axial velocity computed by probes upstream of the xii ......................20 Flow field near blade pressure surface at peak efficiency (Rotor 67)..................................... ..... 78 Figure 4.......... ......................................... 93 Figure 5..... 80 Figure 4.................................................. 94 Figure 5.... ................... ...............4 Pressure and mass flow rate fluctuations at the onset of stall...............3 Locations of the computational probes (Rotor 67 and Rotor 37)...................................... 79 Figure 4...16 Static pressure contours at 30% and 70% span at peak efficiency (Rotor 67)....................... 81 Figure 4... . 96 Figure 5....21 Flow field at mid-passage at peak efficiency (Rotor 67)......................1 Comparison of measured and computed characteristic performance map (Rotor 67)........ 79 Figure 4..........Figure 4............. 93 Figure 5........................ operating point B (Rotor 67)......2 Comparison of measured and computed adiabatic efficiency (Rotor 67)..... 95 Figure 5..... ............................15 Comparison between experiments and present results of relative Mach number at 90% span and 50% pitch at peak efficiency (Rotor 67).....................19 Flow field near blade suction surface at peak efficiency (Rotor 67)................... ..18 Secondary flow in the circumferential direction at peak efficiency (Rotor 67).................................. .. 77 Figure 4......6 Time history of pressure deviation from its azimuthally averaged pressure upstream of the compressor face at operating point B (Rotor 67).............17 Relative velocity profile at mid-span at peak efficiency (Rotor 67)....................... ............ 95 Figure 5....5 Time history of averaged pressure computed by probes upstream of the compressor face at operating point B (Rotor 67)............ 82 Figure 5.

...........14 Time history of axial velocity deviation from its azimuthally averaged axial velocity upstream of the compressor face at operating point C (Rotor 67)............... ..................... .................................. 98 Figure 5..15 Computed and measured characteristic performance map at 70% design speed (Rotor 37)....................9 Pressure and mass flow rate fluctuations at operating point C (Rotor 67)............. ..................... . 96 Figure 5............... ........compressor face at operating point B (Rotor 67).....13 Power spectral density of the azimuthally averaged pressure fluctuations upstream of the compressor face at operating point C (Rotor 67)..........18 Power spectral density of pressure deviation from its azimuthally averaged pressure fluctuations upstream of the compressor face at operating point C (Rotor 37). 101 Figure 5................ .. 103 Figure 5...11 Instantaneous circumferential pressure fields at operating point C (Rotor 67)..............................................................................................8 Time history of mass flow rate fluctuations at operating point C (Rotor 67)..... .............................17 Time history of pressure deviation from its azimuthally averaged pressure upstream of the compressor face at operating point C (Rotor 37)...................... 97 Figure 5................... 100 Figure 5...16 Fluctuations of total pressure ratio versus mass flow rate at different operating conditions (Rotor 37). ............................. 101 Figure 5............... 97 Figure 5.... 102 Figure 5....................................................... 103 xiii .................. 102 Figure 5.....12 Time history of averaged axial velocity and pressure computed by probes upstream of the compressor face at operating point C (Rotor 67). ................................. 99 Figure 5............................10 Temporal growth of reversed flow at operating point C (Rotor 67).............

.................................. 116 Figure 6......................9 Upper and lower limit of pressures that trigger closed-loop control (Rotor 67)......... ............................... 117 Figure 6.............1 Characteristic performance map with bleed control (Rotor 67).................. 120 Figure 6...............2 Velocity profile with/without open-loop control (Rotor 67).... 123 Figure 6...................................................................4 Characteristic performance map with throttle characteristic for open-loop control (Rotor 67)............................................5 Spanwise distribution of axial velocity at mid-passage near the leading edge (Rotor 67).... 122 Figure 6.................................. 121 Figure 6.....................................................3 Total pressure ratio and mass flow rate fluctuations with/without open-loop control (Rotor 67)............. 124 xiv .............................................................6 Time history of axial velocity with/without open-loop control (Rotor 67).8 Time history of static pressure under open-loop control (Rotor 67)......................12 Time history of axial velocity under closed-loop control (Rotor 67)............. .. 122 Figure 6.10 Total pressure ratio and mass flow rate fluctuations under closed-loop control (Rotor 67)..............................Figure 6............................ 115 Figure 6........ . ............7 Time history of axial velocity deviation from its azimuthally averaged axial velocity with/without open-loop control (Rotor 67)........................... .......................................... 124 Figure 6................................... 118 Figure 6................... 119 Figure 6...............13 Time history of axial velocity deviation from its azimuthally averaged axial velocity upstream of the compressor face under closed-loop control (Rotor 67). 123 Figure 6...........11 Characteristic performance map with throttle characteristic for closed-loop control (Rotor 67)..................................... ..........................

xv .

ct2. fw g. used in Spalart-Allmaras turbulence model inviscid flux vectors internal energy per unit volume functions in Spalart-Allmaras turbulence model functions in Spalart-Allmaras turbulence model functions in Spalart-Allmaras turbulence model identity matrix bleed valve constant throttle valve constant characteristic length of the compressor. fv2. G e ft1. F. ct3.NOMENCLATURE A Ab At a CV cb1. cb2 ct1. ct4 cv1 cw1. gt I Kb Kt Lref Jacobian matrix bleed area throttle area speed of sound specific heat at constant volume constants in Spalart-Allmaras turbulence model constants in Spalart-Allmaras turbulence model constants in Spalart-Allmaras turbulence model constants in Spalart-Allmaras turbulence model distance to closest wall. ft2 fv1. cw2. cw3 d E. usually the rotor diameter xvi .

. w Vp x. S. mb p q R. z ∆q ∆U ∆t γ time relative velocity normal to a cell face Cartesian velocity components plenum volume Cartesian coordinates change in conserved flow variables from one time step to next trip point velocity in Spalart-Allmaras turbulence model time step specific heat ratio xvii . T S ~ S T T t U u. mass flow rate through the compressor throttle mass flow rate bleed mass flow rate pressure conserved flow variables viscous stresses and heat fluxes at a cell face vorticity magnitude in Spalart-Allmaras turbulence model modified vorticity magnitude temperature also refers to matrix containing eigenvectors of the Jacobian matrix mt . mc . y. v.

t ∞ stagnation quantity bleed plenum turbulence quantity Cartesian and time derivatives free stream quantity. z. y. upstream of the inlet xviii .κ Λ µ ν νt ~ ν von Karman constant matrix containing eigenvalues of the Jacobian matrix molecular viscosity eddy viscosity turbulent viscosity working variable in Spalart-Allmaras turbulence model density constant in Spalart-Allmaras model viscous stress tensor trip point wall vorticity ρ σ τij ωt Subscripts 0 b p t x.

Superscripts n. n+1 two adjacent time level Overbars ~ →  used to indicate Roe averages used to indicate vectors time or spatial averages xix .

numerical investigating these adverse aerodynamic phenomena. and to develop control strategies for alleviation of rotating stall and surge. A three-dimensional unsteady Navier-Stokes analysis capable of modeling multistage turbomachinery components has been developed. The scheme is implicit in time. In some instances.SUMMARY Axial compression systems are widely used in many aerodynamic applications. called modified surge. and its interaction with the tip shock leads to compressor rotating stall and modified surge. The analysis is cast in a very general form so that a variety of configurations -centrifugal compressors and multistage compressors. These instabilities lead the compressor to rotating stall or surge. Calculations have been done both at design and off-design conditions for an axial compressor tested at NASA Glenn Research Center. a combination of rotating stall and surge. At off-design conditions the calculations show that the tip leakage flow becomes strong. Both global variations to xx . and first or second order in time. permitting the use of large time steps. investigations have been performed to study these phenomena. the operability of such systems is limited at low-mass flow rates by fluid dynamic instabilities. A one-equation Spalart-Allmaras model is used to model the effects of turbulence.may be analyzed with minor modifications to the analysis. Experimental and computational methods are two approaches for In this study. However. has also been observed. This method uses a finite volume approach that is third order accurate in space.

In the open-loop case mass is removed at a fixed. and azimuthal variations in flow conditions indicative of rotating stall. preset rate from the diffuser. It is observed that smaller amounts of compressed air need to be removed with the closed-loop control. It is demonstrated that these adverse phenomena may be eliminated. Calculations show that both types of bleeding eliminate both rotating stall and modified surge. were observed. the rate of bleed is linked to pressure fluctuations upstream of the compressor face. and suppress the precursor disturbances upstream of the compressor face. In the closed-loop case.the mass flow rate. Two types of controls were examined: open-loop and closed-loop. The bleed valve is activated when the amplitude of pressure fluctuations sensed by the probes exceeds a certain range. as compared to open-loop control. associated with surge. by the use of bleed valves located on the diffuser walls. xxi . and stable operation restored.

xxii .

Parsons built and patented an axial flow compressor in 19011. Since that time. Parsons 1901. and set at a suitable angle to the axis of rotation. compressors have significantly evolved. They are a primary component in turbojet engines used in aerospace propulsion. and to prevent damage. the pressure ratio per stage. the movable blades being more widely spaced than in my steam turbine. and constructed with curved surfaces on the delivery side. or dealt with soon after their inception. flow instabilities such as rotating stall and surge must be avoided. The fixed blades may have a similar configuration and be similarly arranged on the containing casing at any suitable angle. Compressors can vary in size from a few feet to tens of feet in diameter. and processors in chemical industry to pressurize gas or fluids. Compressors have a wide variety of applications.” taken from Reference [1]” In 1853 the basic fundamentals of the operations of a multistage axial compressor were first presented to the French Academie des Sciences2-3.1 CHAPTER I INTRODUCTION “My invention consists in a compressor or pump of the turbine type operating by the motion of sets of movable blades or vanes between sets of fixed blades. and a decrease in weight. There have been continuous improvements leading to increases in efficiency. industrial gas turbines that generate power. In turbomachinery applications. safe and efficient operation of the compression system is imperative. To run a compressor as efficiently as possible. 1 .

axial compressors can handle a higher mass flow rate for the same frontal area. They can be divided into four general groups: rotary. As Ferguson4 points out. compressors increase the pressure in different ways. This is one of the reasons axial compressors have been used more in aircraft jet engines. 1. In rotary and reciprocating compressors. the pressure rise obtained from a single axial stage is lower. and does not need the complex return channels required in multiple centrifugal stages.Considerable interest exists in the jet propulsion community in understanding and controlling flow instabilities. However. In centrifugal compressors. a turbo-compressor may also be 2 . flow enters and leaves the compressor in the axial direction. where frontal area plays an important role. centrifugal and axial. shaft work is used to reduce the volume of gas and increase the gas pressure. the flow leaves the compressor in a direction perpendicular to the rotation axis.1 AN OVERVIEW OF COMPRESSOR OPERATIONS The basic purpose of a compressor is to increase the total pressure of the working fluid using shaft work. the high kinetic energy of the fluid is converted into pressure by decelerating the gas in stator blade passages or in a diffuser. reciprocating. In axial and centrifugal compressors. Because an axial compressor does not benefit from the increase in radius that occurs in a centrifugal compressor. In axial compressors. also known as turbo-compressors. the fluid is first accelerated through moving blades. Depending on their type. In the next step. Another advantage of axial compressors is that multistaging is much easier. compared to centrifugal compressors.

and more advanced topics on compressors may be found in References [10]-[12]. fan. Since modern applications require compression systems with higher-pressure ratios and mass flow rates. fans are the first stage of the compression system in jet engines. through the stage. It shows how pressure rises through the rotor and stator passages. and roughly equals the work input per unit mass flow. The working fluid passes through these blades without significant change in radius. booster.called as blower.2.2. consists of a row of rotor blades followed by a row of stator blades.1. and the distinctions between these are vague. A variety of turbo-machines and their ranges of utilization in terms of basic nondimensional parameters are shown in Figure 1. taken from Reference [5]. A schematic diagram of the changes in velocity and fluid properties through an axial compressor stage is shown in Figure 1. Early axial compressors had entirely subsonic flow.pressure compressors. Generally speaking. or tangential velocity. The compressors considered in this study are from the family of axial compressors. which is a dimensionless measure of the total enthalpy change through the stage. designers 3 . and are low. turbo-charger or exhauster. An axial compressor. The vertical axis shows the head coefficient. The literature on compressors is vast. Blowers may be thought of medium pressure compressors. Energy is transferred to the fluid by changing its swirl. References [5]-[9] give a basic introduction to turbomachinery. shown in Figure 1. which is a non-dimensional volume flow rate. The horizontal axis represents the flow coefficient.

4 to 1. A transonic compression system is now one of the main components of high-bypass ratio engines. These systems have been achieved by advanced design.have permitted supersonic flow. which is perturbed by a transient change of mass-flow. The steady state match between a compressor and its drive turbine or jet nozzle. 1. characteristic or performance map of the compressor. In the case of transient disturbances. is a good example of this 4 . Large fans with inlet relative Mach numbers of 1. Today. where regions of subsonic and supersonic flow both exist in the blade passages. the system is unstable. The steady state performance of a compressor is usually described by a plot of the averaged mass flow rate versus the total pressure ratio.3 shows a typical compressor performance map for axial and centrifugal compressors. Disturbances may be considered as transient or deliberate changes to the operating point. the system is stable if it returns to its original operating point.6 have been recently used in engines of this kind. If the disturbances drive the compressor away from the original point. This plot is called the Figure 1. Axial compressors tend to have a steeper drop aft of the peak of the compressor performance map compared to centrifugal compressors.2 COMPRESSOR STABILITY Stability in a compressor is the ability of a compressor to recover from disturbances that alter the compressor operation about an operational equilibrium point. particularly near the leading edge tip where the highest total velocity occurs. using sophisticated computational design tools and extensive experimentation. most high performance compression stages are transonic.

e. Otherwise. If they 5 . If the mass flow rate is too high. which deals with the matching of compressor performance with a downstream flow device such as a turbine or throttle.5. Stability in compressors may be studied from two different perspectives. Mathematically. as the mass flow rate through the compressor decreases. These instabilities include rotating stall and surge. flow instabilities will occur.. the system is unstable. The second is aerodynamic stability. The stable range of operation of axial and centrifugal compressors is limited at both very high and very low mass flow rates. the system is not stable. The operational stability of a compression system depends on the characteristic of both the compressor and the downstream flow device. If steady state operation at a new operating point is not possible. When there are deliberate changes to the operating point. which deals with deteriorations in the operation due to flow separation. by design. if the slope of compressor performance map is less than the slope of characteristic map of the throttle (points P1 and P2 shown in Figure 1.case.4. The first is called operational stability. stall or surge. Operations at lower mass flow ratios (near point P2) can trigger instabilities as discussed later. On the other hand.g.4b for point P3. as shown in Figure 1. as shown in Figure 1. Compressors. the performance is considered stable if a new operational equilibrium point can be achieved. shocks will form and the flow through the compressor will be choked (sonic condition). usually operate near point P1 on the performance map shown in Figure 1. shifting the operating point by changing the compressor shaft speed.4a) the system is stable.

but may still occur on the right side of the surge line if the flow becomes unstable as a result of the instability. none of the models are able to describe accurately the flow phenomena that occur in the compressor and give rise 6 . is to be avoided at all costs. Figure 1. the greater the pressure ratio achieved by the compressor. and to develop an appropriate control methodology for the prevention of these instabilities.5. in particular. Although considerable progress in understanding and modeling of stall and surge in axial compressors has been achieved during the past two decades. but the greater the risk of stall or surge. Rotating stall and surge usually occur at low flow rates. Another reason for introducing the surge avoidance line is that the compressor characteristic. Therefore. a second line parallel to the surge line is usually introduced as a surge avoidance line. may be poorly known.are allowed to persist or grow. In looking at a map of the characteristic performance of a compressor. can be seen. catastrophic damage to the compressor and the engine will occur. and consequently the surge line. The closer the operating point is to the surge line. Surge. 1. Operating at the surge avoidance line provides a safety margin for the compressor operation and prevents the compressor from operating in a region where stall or surge may occur.3 OBJECTIVES AND ORGANIZATION OF THE PRESENT WORK The main objectives of the current study are to understand the physics of compressor stall and surge. a dashed line known as the surge or stall line.

Results for active bleed control techniques are presented in Chapter VI.to stall and surge. most numerical studies of air breathing compression systems are done in the stable part of compressor characteristic performance map. both computational and experimental. However. are necessary for understanding the nature of these instabilities. where the flow is “steady” in a rotating frame. Simulations of the onset and growth of stall are given in Chapter V. in axial compressors. Detailed flow visualizations. 7 . Work has also been done in simulating stall and designing stall control methods that extend the stable operating range of the compressor. Validation results for an axial compressor at peak efficiency conditions are presented in Chapter IV. This thesis is organized as follows: A review of surge and rotating stall phenomena. Chapter III introduces the mathematical and numerical tools required for carrying out the numerical simulations. such as rotating stall and surge. the conclusions and recommendations for further improvements of compressor control technology are given in Chapter VII. both from a historical and a technical perspective. The current research attempts to provide computational tools to study unsteady aerodynamic phenomena. Finally. Most compression systems are now being designed using CFD tools. is presented in Chapter II. CFD modeling of axial compressors is a well-developed field.

8 .Figure 1. Reference [5].1 Work input machinery classification.

Stator Rotor Total Pressure Total Enthalpy Absolute Velocity Static Pressure 1 2 3 Figure 1. [11]. References [5].2 Schematic diagram of changes in fluid properties and velocity through an axial compressor stage. 9 .

a) Stable b) Unstable Compressor Performance P3 P2 Pressure Rise P1 Throttle Characteristic Mass Flow Rate Mass Flow Rate Figure 1. matching the compressor and throttle characteristics.4 Operational stability.Pressure Rise Compressor Mass Flow Figure 1. 10 .3 Typical compressor characteristic map for axial and centrifugal compressors.

Margin of Safety Stall Line Pressure Rise Operating Point Surge Avoidance Line Constant Rotor Speed Line RPM increases Compressor Mass Flow Figure 1.5 Effects of rotor RPM on compressor performance and stability. 11 .

This chapter is devoted to reviewing these unsteady phenomena from both a historical and a technical perspective.3.3. it was pointed out that the stable part of compressor performance map is limited due to aerodynamic instabilities. flow separation at the inlet. or other type of flow distortion). A brief review of methods currently employed in compressor control.1 describes the physical mechanisms behind rotating stall.2 CHAPTER II AN OVERVIEW OF COMPRESSOR INSTABILITY PHENOMENA In the pervious chapter. 2. instabilities may develop and the compressor performance may deteriorate. The instability manifests itself as either a rotating stall or surge. These instabilities manifest themselves as rotating stall or surge. A discussion of surge phenomenon is given in Section 2. the historical developments that led to the discovery of rotating stall and surge phenomena are presented. If a flow instability is somehow introduced into the system (say.2. In Section 2. due to a change in the rotor speed.1 FUNDAMENTALS OF ROTATING STALL During the normal operation of a compressor. the airflow through the compressor is essentially steady and axisymmetric in a rotating coordinate system. Section 2. and a review of prior computational study of compressors and compressor control are also given in Section 2. Rotating stall is inherently a 2-D unsteady local 12 .

the angle of attack decreases. In manufacturing and assembly. in the inertial coordinate system. It often takes only a few seconds for rotating stall to build up. This increase in flow angle. thus causing an increase in their angles of attack and leading them to stall. Rotating stall can occur in both compressible and incompressible flow. As the blade rotates away from the disturbances. known as a stall cell. in addition to blockage attributed to stall. When one of the blades stalls. 13 . and the compressor can operate under rotating stall for several minutes before damage develops. In a coordinate system attached to the blades. restoring normal flow over that blade. Reference [12]. Dimensional tolerances could be one possible case. cause part of the oncoming flow to be diverted towards the neighboring blades. the stall region propagates in the same direction as the wheel motion. rotating stall moves in a direction opposite to the blade motion at a fraction of the rotor speed. Stall is present on some of the blades. the angle of the flow relative to the shaft increases.1. This figure illustrates the blade row viewed from the top of the annulus.phenomenon in which the flow is no longer uniform in the azimuthal direction. It is not known for certain why all blades do not stall at the same time. The inception of rotating stall is shown in Figure 2. The region of the stalled flow. continues moving from blade to blade and propagates around the annulus. These imperfections would cause the inlet air to see these blades at slightly different angles of attack as compared to the other blades. a few blades could be produced with slightly different profiles or with higher stagger angles. as a consequence of some instability. However.

Stall near the root has also been reported. reducing blade life. Abrupt stall results in a sudden drop in total-to-total pressure rise. A typical plot of the static pressure history measurements taken at a fixed circumferential location at the inlet of an axial compressor under rotating stall conditions is depicted in Figure 2. 14 .412 shows the full-span rotating stall with various stalled cells. One of the characteristics of pure rotating stall is that the average flow is steady with respect to time. Two types of stall associated with the number of stalled cells exist. The number of stall cells depends on the compressor at hand. and appears to always involve a single stalled cell.The reported rotational speed of rotating stall around the annulus of compressor varies from 20 to 75 percent of the rotor speed in the direction of the rotor motion13. Typical frequencies for rotating stall are 10 to 50 times larger than those for surge. the cyclical variation of the pressures on the blades can cause them to fatigue and eventually break. but the flow has a circumferentially non-uniform mass deficit. The flow temperature may also increase due to uneven distribution of shaft work. progressive and abrupt. It has also been reported that the incipient rotating stall cells move faster. the pressure ratio after stall reduces gradually. a phenomenon involving multiple stalled cells. is stalled. usually the tip.214.312. • Full-Span: The entire height of the annulus is stalled. only a restricted region of the blade passage. one to nine stalled cells has been reported. Several types of rotating stall exist13: • Part-Span: As illustrated in Figure 2. In progressive stall. During rotating stall. Figure 2.

along a straight line AB. Surge is characterized by large amplitude limit cycle oscillations in mass flow rate. If point B is stable. until measures are taken to bring it back to the unstalled branch.• Small/Large scale: In this case. Many of the conditions that a compression system experiences during rotating stall are also present in surge. called surge. the compressor will remain and operate there. 2.2 FUNDAMENTALS OF SURGE Surge is a global 1-D instability that can affect the whole compression system. the average flow through the compressor is unsteady but the flow is circumferentially uniform. Rotating stall can also serve as the precursor to the more severe and dangerous flow stability. Figure 2.6 shows a typical plot of the plenum pressure transient response when an axial compressor experiences surge14. Recovery from rotating stall is often more difficult than surge13. Even a complete reversal of the flow is possible. Sometimes the deterioration in the performance of an axial compressor with rotating stall is small. The behavior of surge depends on both the compressor characteristic and the characteristics of the diffuser13. In contrast to rotating stall. a small/large part of annular flow path is blocked. When rotating stall occurs at point A on the unstalled branch. The rotor blades are stressed 15 . Figure 2.5 shows a typical rotating stall pattern. and may not be easily detected except as an increase in the compressor noise or by some high-frequency sensors. the operating point then proceeds to the so-called stalled characteristic at point B. and pressure rise.

In high-speed compressors. entire annulus flow fluctuates in axial direction. In both axial and centrifugal compressors. small periodic pressure fluctuations governed by the Helmholtz resonance frequency. while increasing the plenum pressure at the compressor exit at a constant rotor speed. • • Classic Surge: No flow reversal. possibility of flow reversal. 16 . The mild surge may be followed by rotating stall or modified surge. the surge appears as a moderate pulsing of the flow. surge can be categorized into four different classes13: • Mild Surge: No flow reversal. and resulting pitching moments can also change the twist of the rotor/stator blades. backpressure decreases while the inlet pressure increases. the reversal flow can be triggered by a shock wave3. Based on flow and pressure fluctuations. A classic or a deep surge may then follow. larger oscillations at a lower frequency. Modified Surge: Combination of classic surge and rotating stall. In low-speed compressors. axisymmetric flow. The high pressures behind the shock may deform the casing and inlet. • Deep Surge: Strong version of classic surge. non-axisymmetric flow. a mild surge can occur.by the oscillating flow and the uneven distribution of shaft work. The compressor's noise characteristic changes and pressure fluctuations occur through out the compressor.

The extensions to the inlet distortion effects may be found in References [21] and [22]. rotating stall was first detected in 1932 in a centrifugal pump impeller17. and this linearized theory limited them to analyzing weak disturbances to the flow. Qualitative explanations of surge have long been known. The first nonlinear analysis of incipient and fully-developed rotating stall was presented by Moore20. Moore and Greitzer23-24 presented the theory of post-stall transients in axial compressor. and was associated with Helmholtz resonances. it was found that the ducts linked to the compressors could be primary contributors to surge. Reference [12]. Pearson and Grant18.2. Rotating stall and surge are nonlinear phenomena. and detailed the state of the art about these phenomena in axial compressors. Up to this time. To this author’s knowledge. In 1955. 17 . respectively. They predicted the rotational velocity of the stall cell using linear flow theory. In 1986. Horlock1 published a book about these two types of flow instabilities. In 1958.3 LITERATURE SURVEY OF STUDIES ON ROTATING STALL AND SURGE The occurrence of the fluid dynamic instabilities was first considered one of the normal operating features of axial compressors. an extensive study of rotating stall and surge was presented by Emmons. In 1946. researchers assumed that the surge phenomenon was sinusoidal with respect to time. One of the early attempts to model the one-dimensional flow through the axial compressors in a nonlinear form was by Greitzer19 in 1976. Stodola15 and Kearton16 described the surge phenomenon in 1927 and 1931.

Unrecoverable stall in gas turbines requires restarting the engine and may also have catastrophic consequences in aircraft jet engines.1 EXPERIMENTAL STUDIES ON COMPRESSOR CONTROL Surge and rotating stall are highly undesirable phenomena. surge detection and avoidance. To overcome this dilemma. An exhaustive survey of rotating stall and surge control can be found in Reference [13]. and therefore are limited to the surge phenomenon. The effects of compressibility on surge can be found in References [25] and [26]. 18 . and increasing the stall margin approach. These aerodynamic instabilities reduce the total to total pressure rise and efficiency of the compression system. it is desirable to operate the compressor close to the surge line. They can introduce mechanical and thermal loads. 2. Incorporation of compressibility effects in two-dimensional fluid dynamic models was presented in Reference [27].Much of the work on rotating stall and surge control in the literature was based on the assumption of incompressible flow. and can even cause structural damage. the importance of compressibility effects has been recognized and studied. Since the beginning of 1990’s. They are based on one-dimensional fluid dynamic equations. Further details on nonlinear control methods may be found in References [28] through [30]. On the other hand.3. These instabilities may be avoided by operating away from the surge line. due to the high performance and efficiency obtained near the surge line. three different approaches exist: surge/stall avoidance.

are total pressure at the compressor exit and inlet.1) Here P02. When surge or stall is detected. In a multistage axial compressor in a turbojet. and P01. the control systems do not allow the compressor to operate on the left side of the surge avoidance line. The advantage of this 19 . is based on total pressure ratio and is defined as:  Po 2  P   −  o2  P      01  Surge  P01  Surge Avoidance SM =  Po 2   P    01  Surge Avoidance (2. respectively. their time derivatives and oscillation frequency) at the compressor inlet or exit. This safety margin may be defined based on pressure ratio. it is typical to have a safety margin as high as 25 percent9. A common safety margin. or a combination of pressure ratio and corrected mass flow.g. In this approach. These measurements are compared to the expected values at the surge condition.Surge avoidance techniques are known and have been used for a long time in industry and commercial systems. bleed) are applied. the onset of instabilities is first detected. corrective measures (e. a safety margin should be specified. The most successful techniques to detect the onset of stall are based on monitoring the pressure and temperature variations or other parameters (e. To locate the surge avoidance line on the compressor map. In surge detection and avoidance methods. while for a centrifugal pipeline compressor the safety margin is 10 percent31. corrected mass flow. stored in the control computer.g. SM.

the casing is designed with various shaped 20 . the surge avoidance line is also shifted. The disadvantages of this technique are the need for large control forces and a very fast-acting control system that will prevent the growth of instabilities into surge. Thus. the rotor casing is designed so that the amount of blockage in a flow passage is decreased. By shifting the surge line. rotating stall is suppressed. which have been investigated more in axial compressors. Another weakness of this technique is that it is highly dependent on the compressor being controlled. Casing treatments32-35 and variable guide vanes36-39 are some the different ways of achieving passive surge/stall control.technique is that it is not necessary to define a large safety margin. The third control methodology involves increasing the stall margin. In passive surge/stall. In other words. and the compressor can operate close to the surge line. An advantage of this methodology is that the compressor now can operate near peak efficiency and high-pressure ratios at lower mass flow rates. In this method. since different compressors exhibit different behaviors during the onset of surge. some part of the unstable area in the performance map is being stabilized by this approach. In both active and passive control the characteristic performance map of the compressor is modified and the surge line is shifted to a lower mass flow. This approach may be divided into two different classes: passive surge/stall control and active surge/stall control. the geometry of the compressor is altered to modify the stall margin. Casing treatments.

The effects of a porous casing on stall margin for both uniform and distorted inlet conditions may be found in Reference [32]. In this method. This reduces the local angles of attack. Variable inlet guide vanes are also commonly used when starting and accelerating engines to avoid crossing the surge line. In closed-loop control. Air injection is another way of increasing the stall margin and has been used in both axial and centrifugal compressors. a small amount of high pressure and high velocity air is injected into the compressor upstream of the compressor face. In active stall or surge control. or blade angle slotted. while in the open-loop control no feedback signals are used. As a result. and the leading edge separation is prevented. the incident angle in compressors at lower mass flow rates is reduced and the leading edge separation is prevented. the direction of the flow at the leading edge is turned such that the angle of attack decreases. the compressor is equipped with devices such as a bleed valve that can be switched on or off. axial slotted. In this method. a feedback law is used to activate the controller. The injected air 21 . and recirculation (a combination of injection and bleeding) are examples of active surge/stall control. active surge/stall control may be divided into two classes: open-loop and closed-loop.grooves: perforated. Air injection40. bleeding40-42. circumferentially grooved. honeycomb. the flow is energized and the axial velocity component is increased. With inlet guide vanes. The use of variable inlet guide vanes is another way of increasing the stall margin and has been used in both axial and centrifugal compressors. Generally.

such as temperature. The closed-loop control devices use a sensor for detecting the growth of instabilities (precursor waves) when a compressor experiences stall conditions.may be supplied from the diffuser downstream of the compressor or from a separate device. This technique has been used in both axial and centrifugal compressors. and the literature on this approach has become extensive over the last decade. The stall-detection devices are usually located on the circumference of the compressor casing. Closed-loop active control was first reported by Epstein et al. bleeding has been the most common approach for avoiding surge during engine acceleration and start-up. from a stall-detection device. the so-called smart or intelligent engines. Since the early days of jet engines. The bleed valve is typically located either in the plenum exit or downstream of rotor on the shroud. This method promises to be an integral part of future engines. a control unit processes measured flowfield data. flow acceleration can increase and surge-free operation is achieved. either upstream or downstream of 22 .43 in 1989. One of the oldest. bleeding has also been used to achieve a wide range of operating conditions. pressure or axial velocity. Beside the start-up applications. and the most investigated approach for increasing the stall margin in axial compressors is bleeding. The concept of bleeding is that at lower mass flow the working fluid does not have enough momentum to overcome the viscous and adverse pressure gradient forces in the plenum. By removing some of the highly pressurized flow downstream of the compressor. In this method.

2 COMPUTATIONAL STUDIES OF COMPRESSOR PERFORMANCE AND CONTROL As stated in Section 2. Pinsely et al. movable plenum walls. loudspeakers.3. and centrifugal compressors (References [47]-[49]).42. The control unit activates a set of actuator devices. e. A feedback law connecting the sensed fluctuations to the rate of bleed is used to stabilize the compressor.compressor. 2. re-circulation. turbomachinery designers are extremely interested in tools that give good qualitative and quantitative predictions of turbomachinery performance. and air injections.g. bleed valve actuators have been the most commonly used. Bleed valve control is discussed in References [40]-[42].7 shows a schematic of these passive and active control methods. and may be used in aerodynamic design. pressure ratio and efficiency. Over the past five decades. many attempts have been made to increase the operating range of the compression system using appropriate detection and control devices. Figure 2. One goal of these studies is the prediction of component performance. Therefore. It is obvious that even a small improvement in the efficiency of a commercial aircraft engine can result in huge saving in yearly fuel costs. 23 . studied the use of the throttle valve as actuator in a centrifugal compressor surge control. Among these. and achieved a 25% reduction in the mass flow rate. There are several types of actuators in use for stabilizing the compression system. considerable research has been done on both axial (References [21]-[24] & [44]-[46]).1. Other types of actuators include variable inlet guide vanes.

and grid resolution to obtain satisfactory results. On the other hand. There are several reasons for the scarcity of CFD-based turbomachinery performance calculations in the early literature. and Laplace and Poisson equations were used to model the flow within the inlet and exit ducts. system losses and efficiency are strongly dependent on viscous effects and require careful attention to the viscous terms. turbulence modeling. is the work by Takata and Nagano50 in 1972. artificial viscosity. The flow was treated as incompressible. most of their calculations were For example. However. A second reason for the scarcity of CFD studies is the inability of early computers to perform large viscous flow calculations. several researchers have presented detailed investigations of turbomachinery performance using CFD. extracted from 2-D codes. A third reason is that validation data. to this author’s knowledge. During the past decade computational fluid dynamics has undergone an impressive evolution. respectively. One reason is that pressure field calculations are relatively independent of viscous effects and can be obtained with simple models.One efficient method for investigating complex flow phenomena in rotating machinery is computational fluid dynamics. In recent literature. providing engineers and designers with the capability to model and study 3-D unsteady flows.51 predicted loss buckets for 24 . such as detailed pressure and velocity measurements. due to the small size and the high speeds of the components involved62. They used a finite difference method to solve the nonlinear flow equations. The earliest numerical study of compressor instabilities. Davis et al. are difficult to obtain in turbomachines.

60.transonic compressor cascades in 1986. A number of 3-D CFD codes for detailed modeling of turbomachinery flow fields exist.57. This solver is capable of solving multiple flow passages in an inertial reference frame. Many researchers have also simulated rotating stall and surge in axial compressors using simple 1-D and 2-D codes. There has been little or no effort in numerically modeling and studying off-design conditions by CFD methods using the full Navier-Stokes equations. References [63]-[65]. However. time-accurate 3-D simulations of rotary machines are now possible. Stein and Sankar.53. References [66]-[70]. a 3-D compressible unsteady flow solver for turbomachinery has been developed at Georgia Institute of Technology by Niazi. and can be 25 . with the availability of an AGARD Advisory Report for computational test cases of internal flow. Furthermore. Srivastava and Sankar55.62. Hathaway et al. Hah et al. loss and exit flow angle were calculated for turbine and fan cascades by Chevrin and Vuillez52 using the 2-D code of Cambier et al. Dawes56. e. Adamczyk et al.58. and Chima et al. In an effort to model unsteady flow within a compression system. among others.g. have developed 3-D codes that have the capability to analyze unsteady turbomachinery flow with multiple blade passages and/or rotor-stator interactions. and the effects of turbulence modeling on turbine blades were studied by Boyle54. Thanks to the massive increase in computing power and the development of sophisticated post-processing and visualization tools. most of these applications have been applied only to steady-state phenomena in axial and radial compressors. Hall59. researchers now have access to excellent data for validation of the turbomachinery codes.61. Wood at al.

a diffuser bleed valve and air injection. while considering one single flow passage. Stein70 simulated surge with both steady and pulsed air injection control in centrifugal compressors using this threedimensional time accurate compressible flow solver. were implemented. The CFD simulations captured the onset of surge within the compression system. It was shown that these control schemes could suppress the stall and significantly extend the useful operating range of the compressor66-67. this is the first time such flow control simulations based on the full set of 3-D Navier-Stokes equations have been done. 26 . this code has been applied to a NASA low speed centrifugal compressor (LSCC) configuration at both design and off-design conditions66-67.extended to multistage compression systems that are currently used in gas turbines. In a previous study. Two different active control schemes. To the knowledge of this author.

27 .Figure 2.1 Rotating stall inception. Figure 2. Reference [14].2 Transient response of system in rotating stall.

Figure 2. Reference [12].3 Part-Span rotating stall with different stalled cells.Figure 2. 28 .4 Full-Span rotating stall with different stalled cells. Reference [12].

6 Transient response of system in surge.5 Compressor map with the stalled flow characteristic. Reference [13]. Figure 2. Reference [14]. 29 .A Pressure B Stalled Unstalled Mass Flow Rate Figure 2.

30 .Air Injection Guide Vanes Bleed Valves Movable Plenum Walls Controller Air Injection Pressure Sensors Bleed Valve Recirculation Control Figure 2.7 Types of active and passive compressor control schemes.

In Section 3. respectively. These equations are derived from first-principles and from thermodynamic considerations. solution of the 3-D Navier-Stokes equations is required.4.2 and 3. unsteady compressible viscous flow. In Section 3.3 CHAPTER III MATHEMATICAL AND NUMERICAL FORMULATION In order to analyze flow details in compressors. Therefore. 3.1 GOVERNING EQUATIONS The system of partial differential equations for the conservation of mass.1 unsteady compressible flow equations are presented. 31 . In this chapter the mathematical formulation and numerical tools employed in this study are documented. A comprehensive overview of CFD methods may be found in References [71] and [72]. The turbulence modeling method implemented in this work is discussed in Section 3. numerical techniques are required for more complex problems.5 the initial and boundary conditions are described. The Navier- Stokes equations describe the physics of 3-D. The complex nature of the governing equations limits the analytical solutions to simple flows and configurations.3. momentum and energy in fluid flow are known as the Navier-Stokes equations. The numerical discretization process and an approximate factorization solution algorithm used in the flow solver are given in Sections 3.

have been considered. The state vector and inviscid flux terms are: ρ   ρu      q =  ρv   ρw    Et    . F =  ρv + p    ρuw   ρvw      ( Et + p )u  ( Et + p )v      . G are inviscid flux terms. Et. Et = ρ CV T + (u 2 + v 2 + w 2 ) 2   (3-3) 32 . z direction (u. y.subject to some stress-strain rate relationships. calorically perfect Newtonian fluids. v. and R. respectively) and total energy. velocity components in x. In this study. obeying the Stokes linear stress-strain rate law. In three-dimensional Cartesian coordinates. The quantities E. T represent viscous terms. q is the state vector with unknown flow variables: density ρ. the conservative form of the equations in vector form is given below: ∂q ∂E ∂F ∂G ∂R ∂S ∂T + + + = + + ∂t ∂x ∂y ∂z ∂x ∂y ∂z (3-1) Here. S. w. F.  ρw     ρuw    G =  ρuw   ρw2 + p    ( Et + p ) w   (3-2) 1   Furthermore.  ρu   ρv   2   ρuv   ρu + p       2  E =  ρuv .

and since the working fluid is air. CV is the specific heat at constant volume and T is the temperature. The viscous terms are:  0  0      τ yx  τ xx     R = τ xy .4 is used. S = τ yy      τ yz  τ xz uτ xx + vτ xy + wτ xz + qx  uτ y x + vτ yy + wτ yz + q y      0    τ zx    T = τ zy    τ zz  uτ zx + vτ zy + wτ zz + q z    where (3-6) 33 .Here. and is related to total energy and velocity as follows: p = ρRT 1   p = (γ − 1)  Et − ρ (u 2 + v 2 + w 2 ) 2   (3-4) (3-5) In Equation (3-5).  . p is pressure. Also. γ is the specific heat ratio. a value of 1.

The following reference parameters have been used in this work: Lref = Diameter of the rotor . 3 All quantities in the Navier-Stokes equations have been non-dimensionalized by their corresponding reference values. λ .τ xx = λ (u x + v y + wz ) + 2 µu x τ xy = τ yx = µ (u y + vx ) τ xz = τ zx = µ (uz + wx ) τ yy = λ (u x + v y + wz ) + 2 µu y τ yz = τ zy = µ (vz + wy ) τ zz = λ (u x + v y + wz ) + 2 µu z and ∂T ∂x ∂T q y = −k ∂y ∂T q z = −k ∂z q x = −k (3-7) (3-8) In these equations. In Equation (3-7). at the blade trailing edge Vref = a ambient. is − µ . upstream of the inlet = ρ ∞ p ref = 2 ρ ref Vref (3-9) γ 34 . µ is the molecular viscosity and k is the thermal heat conduction 2 coefficient of the fluid. from the Stokes hypothesis. upstream of the inlet = a ∞ ρ ref = ρ ambient.

Therefore. can be handled. In this study. The integral form of Equation (3-1) is: r r r r r r ∂q dV + ∫∫ ( E i + F j + G k ) ⋅ n dS − ∫∫ q V G ⋅ n dS = ∫∫∫ ∂t V S S r r r r ( R i + S j + T k ) ⋅ n ds ∫∫ S (3-10) Here. for complex geometry and flows with highly non-linearity. In the finite volume method in three-dimensional space. One of the advantages of this method is that discontinuous phenomena. solutions are found for discrete points at different time levels. and finite element methods. The finite volume method is commonly used in fluid dynamic problems. Several techniques. V and S refer to the control volume and control surface area respectively. such as finite difference methods. the flow is divided into a finite number of hexagonal cells. exist for numerically solving the Navier-Stokes equations. and solved using a finite volume scheme. numerical techniques must be used to find approximate solutions. finite volume methods. the 3-D unsteady compressible Reynolds-averaged Navier-Stokes equations are recast in integral form. and r r n represents the outward normal vector to surface S. such as shock waves.2 DISCRETIZATION OF THE GOVERNING EQUATIONS Analytical solution of the Navier-Stokes equations is limited to simple geometries.3. In numerical methods. The term VG refers to the velocity 35 .

k 2 (3-12) 1 i . and T. j .k 2 1 i .k i . j . j .k 2 [ ] 1 i ± . as shown in Figure 3.k 2 2   i + 2 .k − 1  2 2  (3-11) r r = En x + Fn y + Gn z − q(VG ⋅ n ) ∆S r r = En x + Fn y + Gn z − q (VG ⋅ n ) ∆S r r = En x + Fn y + Gn z − q(VG ⋅ n ) ∆S ) where E ) F ) G 1 i ± . In Equation (3-10). j . j . j .k ) )  G i . j + 2 .k ± 1 2 Viscous fluxes.of the surface S. j − . likewise are handled as follows: r r r r ( Ri + Sj + Tk ) ⋅ n dS = ∫∫ s All Faces ∑ ( Rn x + Sn y + Tn z )∆S (3-13) 36 .k ± 2 ] i . j . j ± .k 2 [ [ ] 1 i .k + 1 + G i . and the surface integrals are computed at the six faces surrounding the control volume: r r r r r r ( Ei + Fj + Gk ) ⋅ n dS − ∫∫ qVG ⋅ n dS = ∫∫ s S All Faces r r ( En x + Fn y + Gn z )∆S − qVG ⋅ n∆S ∑ [ ] ) ) )  )  = E 1 + E 1  + F 1 + F 1  + i − . the state vector q is evaluated at the cell vertices. j ± . j . S. R.k   i .1.

In Equation (314). E . assuming that the original matrix is given by A = = TΛT −1 . the ) ) ) numerical flux fNum (which is an approximation to E . which Λ contains the eigenvalues. but are real numbers for Euler equations. F . or G ) is given by: f Num Physical Flux Term Artificial  644 744 8 6444Viscosity44 8  4 4 4 4 Term 4 7 1  1 ~ =  [ f (q L ) + f (q R )] − A(q L . The quantity A is defined as: ~ ∂f A = T Λ T −1 . q R ) (q R − q L )  ∆S 2 2    [ ] (3-14) where ρ LU L    ρ LU L u L + p L n x    f (q L ) =  ρ LU L v L + p L n y . These quantities are in general complex number.) ) ) The inviscid fluxes. The matrix ∂q T contains the eigenvectors of A. ρ U w + p n  L z  L L L U L H 0 L − p L nt    ρ RU R    ρ RU R u R + p R n x    f ( q R ) =  ρ RU R v R + p R n y  ρ U w + p n  R z  R R R U R H 0 R − p R nt    (3-15) ~ and ∆S is the cell area of the face where fNum is calculated. and G are calculated implicitly using Roe's flux difference scheme73-78 as described by Liu and Vinokur79. F . At a cell interface. q L and q R are the primitive variables vectors to the left and the right of the cell interfaces and are defined as: 37 .

VG is the grid velocity at the face where fNum is being calculated.2. with Roe’s Superbee limiter78 to control the high frequency oscillations near shocks. For example. q L and q R at cell face i + shown in Figure 3. and total enthalpy.    wL   pL    ρ R  u   R q R = v R     wR   pR    (3-16) The quantities. A four-point stencil. and n is the unit normal vector to this face. as 2 38 . U and H0 are the relative velocity normal to the cell face.ρ L  u   L q L = v L  . respectively and are defined as: r r r U = (V − VG ) ⋅ n H 0 = Et + p and r r nt = −VG ⋅ n (3-17) (3-18) (3-19) r r Again. was used to compute the left and right values of primitive variable vectors at each cell interface. can be written as: 1 .

1). i− 2 i− 2 Φ − 1 = Φ (r − 1 ) i+ 2 i+ 2 r +1 = i− 2 qi +1 − qi qi − qi −1 . min(2r . using Liu and Vinokur79 formula.1 1 q L = qi + Φ + 1 (q i − q i −1 ) + Φ − 1 (qi +1 − qi ) 6 i− 2 3 i+ 2 (3-20) 1 1 q R = q i +1 − Φ + 1 (qi +1 − q i ) − Φ − 3 (q i + 2 − qi +1 ) 3 i+ 2 6 i+ 2 and Φ + 1 = Φ (r + 1 ) . can be expanded as: ~ ~ ~ A(q L . min(r . q R ) (q R − q L ) = λ1 ∆q + δ 1 q * + δ 2 N n where (3-22) ~ ~  ~ ~  ~ ~  − λ + λ 2 + λ3  ∆p λ 2 − λ3 ρ ∆U δ1 =  + 1 ~  a2 2 2 a  ~   (3-23) 39 . r −1 = i+ 2 q i − q i −1 qi +1 − q i (3-21) Φ(r ) = max[0.2)] The second term in Equation (3-14).

defined as: ~ ρ = ρL ρR ~ u= ρL uL + ρR uR ρL + ρR (3-27) (3-28) 40 .~ ~ ~ ~  λ 2 + λ3  λ 2 − λ3 ∆p ~  ρ∆U + ~ δ 2 =  − λ1 + ~   2 2 a     (3-24)     1  ~ u  ~  ~ q * = v  ~ w  ~  H0   ρ  ~   and ~ ~ λ1 = U ~ ~ ~ λ2 = U + a ~ ~ ~ λ3 = U − a 0    n x  . N n = n y  ∆S   n z  ~  U    (3-25) (3-26) In Equations (3-22) through (3-26). the quantities marked by “~” are calculated using Roe averaging.

for example: ∆p = p R − p L (3-33) 3. in Equations (3-22) through (3-24) represents the change in the quantities across the cell interface.5. This involves solving the governing equations at each time step by marching in time from an initial flow condition with appropriate boundary conditions. ∆.3 NUMERICAL SOLUTION OF THE DISCRETIZED EQUATIONS The 3-D unsteady compressible Reynolds averaged Navier-Stokes equations are solved numerically using a time marching scheme.~ v= ρ L vL + ρ R vR ρL + ρR ρ L wL + ρ R w R ρL + ρR ρL HL + ρR H R ρL + ρR (3-29) ~ w= (3-30) ~ H= (3-31) ~ ~ r ~r ~ r r r r U = (u i + v j + wk ) ⋅ n − VG ⋅ n (3-32) The difference. As seen earlier. the discretized form of the governing equation is: 41 . discussed later in Section 3.

and links the finite volume cell (i. That is:    ∑ (Rn x + Sn y + Tn z )∆S   All Faces  n +1   =  ∑ (Rn x + Sn y + Tn z )∆S   All Faces  n (3-36) Bringing the unknown quantities to the left side. was used. ∆t =− All Faces ∑f Num ∆S + Sn y + Tn z )∆S (3-34) − All Faces ∑ (Rn x This equation set is highly nonlinear. j . j.j.(q∆V ) in. we get: 42 . (i+2. j. which may be viewed as the first iteration scheme.+j1k − (q∆V ) in.k) etc. Direct numerical solution of this equation set is not feasible. k) to its neighbors (i+1. All terms in Equation (3-34) were first expanded about known flow quantities at the time level “n”:    ∑ f Num ∆S     All Faces  n +1   ∂f Num n +1 =  ∑ f Num ∆S )  + ∑ (q − q n )∆S   All Faces ∂q  All Faces  n (3-35) The viscous terms were logged by one time step.k . A non-iterative scheme. k).

k) with its five neighbors (i±1. j. (∆q)cell face is defined as the average of (∆q) at all centers. j. k). j±1. 43 .qn. k±1). k). and (i. (i. ) ∂f Computing A = Num is costly. Instead. the following approximate form was used: ∂q )  ∂E ∂F ∂G  r r  A =  nx + ny + n z  − VG ⋅ n  ∆S  ∂q ∂q ∂q      (3-38) − λ max I ∆S where λ max is given by: r r r λ max = V − VG ⋅ n + a ( ) (3-39) Finally.) (∆q ) n n V + ∑ A(∆q ) n = −(∑ f Num ∆S ) ∆t − (∑ ( Rn x + Sn y + Tn z ) ) (3-37) n where (∆q)n=qn+1 . j. This leads to a seven-diagonal system linking the (∆q) value at node (i.

the algorithm. k t = µt c p Prt . To capture the turbulence effects directly from the Navier-Stokes equations.This seven diagonal system of linear equations was solved by a three-factor scheme. and sample numerical calculations can be found in that report.80.622. The solution of the matrix equations is obtained through a block lower-upper decomposition (LU) and the application of the Thomas algorithm71. where µt and kt are “eddy” viscosity and “eddy” conductivity. This model involves a number of constants: cb1 = 0. cw1 = cb1 (1 + cb 2 ) + κ2 σ 44 . the Spalart-Allmaras81 model. The advantage of this algorithm is that direct inversions of the entire block matrices are not required. the Reynolds number is high and the flow is turbulent. Time and length scales of the turbulent flows are extremely small. the RANS (Reynolds Average Navier-Stokes System of equations) are solved in this work. has been used for computing µt=ρνt. σ =2 /3. Also. cb2 = 0.41. The factorization algorithm has been described in detail by Pulliam et al. respectively.4 TURBULENCE MODELING In the most practical flows within turbomachines. Details of the equations. 3. To reduce the computational time. κ = 0. which simply requires replace of m and k with (µ + µt) and (k + kt) respectively. the grid resolution would need to be extremely high. a one-equation turbulence model.1335. In this work. and many classical text books.

is given by: ~ ν t = ν f v1 . κ d fv2 = 1 − χ 1 + χf v1 (3-42) Here. ct1 = 1.((ν + ν )∇ν + cb 2 (∇ν ) 2 + Dt σ ~ 2 cb1   ν  2 c w1 f w − κ 2 f 12   d  + f t1 ∆U    [ ] (3-41) Where S is the magnitude of the vorticity.3.1. f v1 = χ3 . ct4 = 2 The eddy viscosity.cw2 = 0. ~ ν ~ S = S + 2 2 fv2 . d is the distance to the closest wall. The quantity ν is the working variable and can be calculated by solving the following partial differential equation ~ ~~ 1 Dν ~ ~ ~ = cb1 (1 − f t 2 ) S ν + ∇. ν is the molecular viscosity. 3 χ 3 + cv1 χ = ~ ν ν (3-40) ~ Here. The function fw is: 45 . ct3 = 1.1 . ν t . cw3 = 2. ct2 = 2. cv1 = 7.

In the free-stream. 1 + c6 3  f w = g  6 w6   g + cw3  1/ 6 ~ ν . and solved using a three-factor approximate factorization scheme. ∆U ) and ωt is the wall vorticity at the trip point and dt is the ω t ∆z distance from the field point to the trip point. This PDE was also recast in an integral form. provided ~ numerical errors don’t push ν to negative values near the edge of the boundary layer. The same is applied to the initial condition. where g = r + cw 2 (r 6 − r ) . 46 . a user specified transition location. r = ~ 2 2 Sκ d (3-43) ~ ~ The wall boundary condition is ν = 0 . the trip function ft1 is obtained from the following equation: ω t2 f t1 = ct1 g t exp(−c12 d 2 + gt2 d t2 2 ∆U [ ] (3-45) where g t = min(0. ν = 0 is best. The function ft2 is: f t 2 = c13 exp(−c14λ2 ) (3-44) Finally.1.

the pressure values equal the downstream pressure.1 INITIAL CONDITIONS At the start of the calculations. The mass flow rate begins to gradually change with time. As the compressor begins to draw the fluid in. the implementation of the initial and boundary conditions is presented.5 INITIAL AND BOUNDARY CONDITIONS Since the Navier-Stokes equations are parabolic in time and elliptic in space. In the present method the grid and solution are stored every few hundred time steps. For example. based on the user input. initial and boundary conditions must be supplied and play a significant role in obtaining a meaningful solution. depending on whether the compressor is operating close to design conditions. It reaches either an asymptotically steady state or a limit cycle oscillation. what is called as a cold start. and these files may be used to continue the calculations or visualize the flow field. to restart the computations. the flow properties everywhere in the system are assumed to be uniform.5.velocity is set to a value consistent with the specified total temperature T0 at the inlet. 3. the pressure at the inlet drops relative to the outflow. In this manner.3. 47 . and the u. or at off-designed stalled conditions. the time-accurate calculations for the compressor are simply a continuation or restart from the previous solution. The analysis can also use a previously stored solution file with a corresponding grid file. In this section.

In the simulation of a compressor drawing the air from the atmosphere. In a multistage compressor analysis. the stagnation temperature. p0 . The fifth equation applied at the inlet boundary condition is the one-dimensional Riemann characteristic equation. assuming that there is no swirl at the inlet. and a non-uniform total temperature at the inlet.5. these quantities may be set as ambient conditions. T0 .3. This equation is: ∂R5 ∂R + (Vn − a ) 5 = 0 ∂t ∂s (3-46) where R5 = 2a − Vn and s is the distance in a direction normal to the inlet face. are assumed to be known. a non-uniform total pressure distribution. The following boundary conditions have been implemented in the analysis: Inflow Boundary: At this boundary. γ −1 48 .2 BOUNDARY CONDITIONS Figure 3. and the total pressure. It is possible within the framework of the present boundary conditions to prescribe swirl. modeling the effects of acoustic disturbances leaving the computational field through the inlet face. these quantities will be known at the downstream boundary of the previous stage. The tangential components of velocity at the inflow boundary are set to zero.3 shows the different types of boundary conditions on a compressor flow passage.

4 shows a schematic of the compressor outflow boundary coupled with the plenum.Neglecting ∂R5 . at every step. the speed of sound. un are found at ∂t the inflow boundary from the relation: ∂R5 =0 ∂s R5 inlet (3-47) (3-48) = R5 interior Or. T. Neglecting ∂R5 was considered acceptable because the flow properties varied very slowly with time ∂t at the inlet. Conservation of mass in the plenum can be written as: 49 . Vp. isentropic relations are used to find the values of the density ρ and pressure p at the inflow boundary. and normal velocity. 2a 2a −u = −u γ − 1 n γ − 1 n Interior u2 a2 a2 n = 0 + γ −1 2 γ −1 (3-49) After computing the speed of sound and temperature. Outflow Boundary: It was assumed in this study that the flow from the vaneless diffuser downstream of the rotor exhausts into a plenum chamber with a constant volume. Figure 3. a.

dm dt & & = (m p − mt ) plenum (3-50) dm dt = plenum d ( ρV ) dt (3-51) plenum Here. & & m P is the mass flow rate at the diffuser exit. therefore: dm dt =V plenum dρ dt (3-52) plenum Inserting Equation (3-52) into Equation (3-50). we have: V dρ dt & & = m p − mt plenum (3-53) The relationship between density and pressure is obtained by assuming isentropic state in the plenum: ∂p ∂ρ 2 = a plenum plenum (3-54) 50 . and mt represents the mass flow rate at the plenum throttle. and V is the volume of the plenum and is constant.

The combination of Equation (3-54) and (3-53) gives the plenum pressure formula: dp p dt = a2 p Vp & & (m p − mt ) (3-55) A first order discretization of Equation (3-55) gives: & & p n +1 = p n + K (m p − mt ) p p (3-56) where a value of 0. and indirectly determines surge. All other quantities such as the density and the three components of velocity are extrapolated from the interior. This value determines the natural frequency of the plenum chamber. Solid walls: At the solid walls. using relationships such as ∂p/∂n=∂T/∂n=∂ρ/∂n=0. The pressure. This condition is applied to the hub. a no slip boundary condition is used. while the velocity for grid points on the compressor blades and r r the shaft. is set equal to Ω × r . The quantity & mt is an input parameter that represents the nominal mass flow rate and lets the compressor run at different operating conditions. The velocity on the walls of the inlet and the diffuser is set to zero. that part which is moving. density and temperature values at the solid surfaces are found from the interior.20 for K is used in this study. casing and blade surfaces. where n represents the direction normal to the solid 51 .

both open-loop and closed-loop control methodologies using steady and unsteady bleeding were studied. The numerical implementation of the bleed boundary condition into the flow Figure 3. usually just a few of the blade passages are modeled. On impeller surfaces.3 and 3. The flow properties at these boundaries are computed by taking the average of the properties on the neighboring points. the flow properties are computed by averaging the properties on either side of the periodic boundary. because of the non-orthogonality of the cells around the blade surfaces.85. neighboring zones share grid lines on their boundaries. Periodicity: Since simulation of the entire flow in a compressor requires significant computations and is very expensive. Density.5 illustrates the periodic boundary 52 . In both control methods. as shown in Figure 3.surface.5. especially near the leading and trailing edges. Bleed Boundary: In this study.6. Therefore.5. is used in the normal direction to the blade surfaces. At the blade-to-blade interface boundaries. described by Pulliam et al. a fraction of the mass flow is removed through a valve at a fixed rate. A schematic of the open-loop control used in this study is shown in Figure 3. bleed valves are implemented circumferentially downstream of rotor. conditions for blocks A and B. a normal momentum equation. located on the shroud. it is assumed that the flow through first blade and the last blade passage are nearly identical and a periodic boundary condition is applied. In the open-loop control. and pressure on the hub and shroud are all directly extrapolated from the interior. Zonal Boundary: As shown in Figures 3.

as discussed in Chapter VI. which in this work were located near 30% of the tip chord. bleed mass flow rate.7 illustrates the schematic of the closed-loop control system used in this study. mb un = ρ b Ab (3-57) Density. While simulating the closed-loop control. bleed valve area. Ab. consistent with the subsonic flow at this boundary. the normal velocity may be written as: . m b . 53 .solver requires user-specification of the bleed valves location. The mass flow rate through the bleed valve is controlled by the instantaneous amplitude of pressure fluctuations sensed by the “computational” probes. From this information. As in open-loop control. The bleed valve was activated when the pressure sensors upstream of compressor face experienced high-pressure fluctuations. and . the density and the tangential velocities are extrapolated from the interior. Figure 3. upstream of the compressor face. the mass flow rate through the bleed valve is linked to pressure upstream of the compressor face by a control unit. pressure and the two tangential components of velocity are extrapolated from the interior.

j.2 Computation of inviscid flux terms on a cell face.j (i.k) (i+1.j. Stencil for q left Left Right Stencil for q right i-1 i i+1 i+2 Cell face i+1/2 Figure 3.1 Control volume and cell-vertex grid points.k) k i Cell face at i + 1 2 Figure 3. 54 .

Inlet boundary: P0, T0, v, w are specified. Riemanninvariant extrapolated from interior.

Periodic boundaries: Properties are averaged on either side of the boundary.

Zonal boundaries: Properties are averaged on either side of the boundary.

Outlet boundary:
.

Solid wall boundary: No-slip boundary conditions. ∂p ∂ρ ∂T = = =0 ∂n ∂n ∂n

m t is specified. All the other quantities are extrapolated from the interior.

Figure 3.3 Boundary conditions used in the axial compressor analysis.

55

Plenum Chamber & mc & mP u(x, y, z) = 0 ap = constant Isentropic & mt

dp p dt

Outflow boundary

=

a2 p Vp

& & (m p − mt )

Figure 3.4 Compressor outflow boundary conditions. Zonal boundary

Block A

Block B

Periodic boundaries

Figure 3.5 Periodic and zonal boundary conditions.

56

One tip chord length

Bleed Valve

Impeller

Figure 3.6 Implementation of bleeding as an open-loop active control.

Controller Unit Pressure Sensors Bleed Valve

Impeller

Figure 3.7 Implementation of bleeding as a closed-loop active control.

57

Next.4 CHAPTER IV CODE VALIDATION STUDIES The numerical scheme and boundary procedures described in Chapter III were coded in a computational flow solver. Results are presented in the following three chapters. and the total number of time steps. The solver first reads a main input data file. before a steady state or a limit cycle oscillation (indicative of separation and/or stall) was achieved. the code reads the boundary condition setup data file to establish the appropriate boundary conditions described in Section 3. This main input file controls the flow type (inviscid. The 58 . Although the present implicit scheme can use large time steps. laminar.4 on the different faces of each of the blocks. It also specifies the computational grid size and the number of grid blocks. patched grid analyses can be performed with this code. The calculations were generally run for several compressor revolutions. the CFL numbers was kept at or below 2 in order to resolve the unsteady flow in a time accurate manner. called GT-TURBO3D. CPU time per time step was 3x10-5 sec per grid point. It also reads nominal mass flow rate. turbulent). Most calculations were carried out on a Silicon Graphics Origin 2000 system using two of the four available processors. mt . The solver and the boundary conditions have been written in a general form so that most multi-block. to & perform the necessary initialization.

comparisons of the experimental data with the CFD results are given in the form of contour plots.mass flow rate across several streamwise planes was monitored and used as a convergence criterion.1. A brief literature survey of work done by other researchers on these two configurations is presented in Section 4. In Section 4. called NASA Rotor 67. computational grids are presented. at a mass flow rate of 33. graphs. 4. The design pressure ratio is 1.56) is the first stage rotor of a two stage transonic fan designed and tested with laser anemometer measurements at the NASA Glenn Research Center. Some grid sensitivity studies are also given. These configurations are the NASA Rotor 67 and NASA Rotor 37 axial compressors.2.1 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 67) In this section.25 kg/sec. This low aspect ratio rotor (σ=1.63. 59 . This chapter is organized as follows. The two axial compressor rotors used in this study are introduced in Section 4.1. 4. In this section. the configurations and the The first geometry chosen to validate the code and simulate compressor stall control was a transonic high-speed axial fan rotor.1 DESCRIPTION OF THE CONFIGURATIONS STUDIED Two different configurations have been chosen to study the compressor stall and passive/active control methodologies. geometry configurations and computational grids are described as well as their design operating characteristics.3.1. and velocity vectors at the peak efficiency. shown in Figure 4.

with four compressor flow passages was generated using the “Gridgen” meshing package. In each flow passage.533 cm < x < 7. Sixty-six cells were placed in the streamwise direction. of which 33 were located over the blade chord. called the baseline grid. A detailed description of blade and flow-path geometry. the grid. static pressure and flow angle data. temperature.4 and 4. In order to model the rotating stall.106 cm and rotational wall conditions were applied to this region. a H-H-type grid. For this purpose. The grid was generated by loading the geometry surface data from Reference [82] into the mesh package. respectively.The rotor has 22 blades with tip radii of 25. Figure 4. shown in Figure 4. multiple flow passage should be considered.7 cm and 24. The hub/tip radius ratio is 0. The grid at the mid-span and the grid normal to the streamwise direction are shown in Figures 4. respectively. To determine the rotor mass flow rate.375 at the leading edge and 0. has 32 and 21 cells in the radial and circumferential directions.3. The clearance gap was spanned by three cells in the radial direction.6 shows the grids in the r-x plane.2 shows the locations where velocities were measured for this compressor. Figure 4.478 at the trailing edge. The rotor hub is in the region of -3.5. The design rotational speed is 16043 RPM. The hub consists of two stationary and one rotational segments.25 cm at the leading and trailing edge. A hyperbolic tangent node point distribution near the surfaces was used to provide adequate resolution of the boundary layers. known as the 60 . and the tip leading edge speed is 429 m/sec with a tip relative Mach number 1. may be found in Reference [82]. as well as measured data on total pressure.38. a calibrated orifice was used.

106 at a mass flow rate of 20. Grid sensitivity studies were also performed to ensure that the baseline grid has adequate resolution to resolve the leakage flow around the tip. Simulations were also done on another grid with twice the number of grid points in each of the three directions (4x131x63x41). In this study.1.2 kg/sec.36 mm. The rotor has a constant meanline diameter with a hub/tip radius ratio of 0. For this rotor. Figure 4. some part of the hub is rotating.g. stationary wall boundary 61 . as for Rotor 67.meridional plane. 4. This is because three-dimensional visualizations of flowfield in compressors are difficult due to configuration complexity. called NASA Rotor 37.19 m/sec at the inlet tip. the solid wall boundary layers and the shock system. results in the form of vectors or contour plots are presented on one of these planes. producing a speed of 454. Rotor 37 is a NASA stage 37 rotor designed at the NASA Lewis Research Center as a test compressor for a high-pressure ratio (20:1) compressor core of an aircraft engine. The rotor design pressure ratio is 2.3 of this chapter and the next two chapters. The tip relative Mach number is 1.7 shows the Rotor 37 configuration. twisted blades. The design rotational speed is 17188 RPM. e. Results for this configuration are discussed in Section 4.48.8 shows a meridional view of this rotor and the locations of laser anemometers at different stations.7. Figure 4. At the hub.2 NASA TRANSONIC AXIAL FAN ROTOR (ROTOR 37) The second configuration of interest was a high-speed axial flow compressor. The clearance at the design is 0. Details of this compressor may be found in references [83] and [84].

In the CFD assessment. Effects of tip clearance gap on the flowfield have also been investigated by Adamczyk et al.264 cm and x > 4. Nine points were placed in the clearance gap. while rotational boundary conditions were applied in the region –0. 4.4. As in the case of NASA Rotor 67.conditions were applied in the region x < -0. the tip clearance gap was modeled and periodic boundary conditions were applied. the 62 .264 cm < x < 4.521 cm. Rotor 37 was used to assess the prediction capabilities of turbomachinery CFD tools in a blind test organized by the 1994 ASME/IGTI workshop.85 and Chima86. Effects of shock boundary layer interaction and wake development are reported by Hah et al91.2 PREVIOUSLY REPORTED STUDIES ON ROTOR 67 AND ROTOR 37 CONFIGURATIONS Rotor 67 is an AGARD test case and is suitable for CFD validation purposes. It has 119x71x41 grid points in each flow passage in the streamwise. full and half. Studies conducted in the stable part of the design speed operating line for Rotor 67 may be found in References [85]-[92]. with an accurate modeling of the clearance gap. spanwise and pitchwise directions. respectively. The grid is a H-H-type with four flow passages. Several researches have studied flowfields within this compressor. shown in Figure 4. Results for this configuration are given in Section 4. End-wall and casing treatment for Rotor 67 may be found in Reference [92]. The grid for NASA Rotor 37 was generated with the same methodology that was used for NASA Rotor 67.9. Chima obtained the results for two different tip clearance gap sizes.521 cm.

called the NASA Low Speed Centrifugal Compressor (LSCC. [94] and [95]. has been validated for Rotor 67 and Rotor 37 compressor configurations. the flow solver was first applied to the simulation of flow through the NASA Rotor 67 compressor at peak efficiency conditions. CFD results for both the baseline grid and the fine grid are presented and compared with the experimental data. the LSCC validation results are not presented here. Because the predominant thrust of the present study is to model axial compressors. In this section.93. High-speed centrifugal compressors have also been studied using this flow solver and computational-experimental comparison results can be found in References [68] and [70].Rotor 37 was tested in isolation by Suder et al. In the present study. 4. including the comparison of CFD predictions and experimental measurements such as blade pressure distributions. Bright et al96 have studied the dynamics for the Stage 37 during rotating stall. The results for the centrifugal compressor. Several researchers computed the complex flow field for this rotor for the stable branch of the performance map. These results may be found in References [90]. may be found in References [66]-[67]. for which experimental results may be found in Reference [82]. built and tested at NASA Glenn Research Center). described earlier. 63 .3 PEAK EFFICIENCY RESULTS FOR ROTOR 67 The viscous flow solver. The flow solver had previously been used by this author to simulate the flowfield in a low speed centrifugal compressor.

as pointed out in Section 3. the pressure drop. the radial distributions of total stagnation pressure were mass averaged across the annulus. especially in the blade suction side near the leading edge.1. The steady solution at each time level was used as a restart solution file for the next speed level. the pressure at the inlet drops relative to the outflow. The flow solver was applied to the rotor at three different speed levels: 30%. is large and can produce non-physical pressure in that region. Since NASA Rotor 67 is a high-speed compressor at its design speed. By this methodology. no restart solution files exist. the flow properties in the computational domain were assumed to be uniform and were set to their inlet freestream values.At the start of calculations. To calculate the total pressure ratio. 50% and 100% of the design speed. A start-up technique similar to that in a real compressor during acceleration was used to avoid this problem. The formula used was: 64 . As the flowfield was advanced in time. the pressure was allowed to drop in the suction side of the blades gradually and a steady state solution was achieved. The drop in pressure depends on the compressor speed.4. mass flow rates across the streamwise planes were calculated and averaged along the streamwise direction from the rotor leading edge to the rotor trailing edge. At each time step. Therefore. A useful byproduct is performance data at off-design engine speeds. The averaged mass flow rate was considered as the mass flow rate through the rotor and was used as a convergence criterion.

The computed axial velocity showed only a slight difference between the fine grid and the baseline grid. using either one or two processors of the Origin 2000. the time required to perform one compressor revolution (where blades rotate 360°) was between 9 and 12 hours.61 kg/sec (1. up to 3 significant digits.12 shows a comparison between the CFD predictions and the measured normalized axial velocity distribution at the inlet survey station 1 (Figure 4.10 and 4. VSTD =310.2. Figure 4. 65 . The mass flow rate on the fine grid was identical to that on the baseline grid. The velocity is normalized by critical velocity at standard day conditions.11 show the fluctuations of mass flow rate and total pressure ratio.2% difference). Therefore. shown in Figure 4. CFD results are given for both fine and baseline grids.63 m/sec. Figures 4. For the baseline grid.23 kg/sec on the baseline grid.p02 = p01 ∫∫ p p02 01 ρVdA (4-1) ∫∫ ρVdA Here. and the corresponding measurement data is 34. The computed mean mass flow rate was 34. P02 and P01 are the stagnation pressure at station 2 and station 1. one rotor revolution required about 7000 computational time steps. These figures show that the mass flow and total pressure fluctuations at this operating condition are very small (less than 1%). In this study 20 time steps per degree of revolution were used.2). In the fine grid. the running time was increased to 65 hours for one rotor revolution.

Figure 4.13 shows the relative Mach number contours at the 30% span location.05 were observed on the forward position of the suction 66 . Figures 4.Good agreement between the measurements and the predictions was obtained. The smoothing and interpolation procedures may be found in Reference [87].87. The flow pattern inside a transonic fan like the NASA Rotor 67 is very complicated due to phenomena such as a 3-D shock system. problems associated with laser anemometer blade flash and data drop-out. There are several reasons the experimental data do not usually produce good quality contour maps. the CFD calculations showed that most of the flow at this span location is subsonic.13 were produced by Pierzga et al. Background noise. shock-boundary layer interaction. As in experimental results. small regions of supersonic flow. The corresponding experimental contours were obtained from Reference [87]. measured from the hub. non-uniform survey spacing.95 and the flow exits at a Mach number of about 0. It is also seen that the grids have enough resolution to capture the boundary layer profiles near the walls (hub and shroud).13 through 4.15 show comparisons between the computed and measured relative Mach number at two different span locations. and inadequate measurement locations in viscous regions and the boundary layer and wake all produce poor quality plots. clearance flow and three-dimensional separation. with relative Mach numbers about 1. who smoothed and interpolated the measurements to fill the zero velocity data using information from the adjacent survey points in both the streamwise and pitchwise directions. The experimental contours in Figure 4.60. However. The inlet mach number is 0.

The computed results in Figure 4. and a bow shock occurs. These are not visible in the experimental contour plots. At both these radial locations.surface of the blades. the pressure gradient is much higher than at 30% span.15 show the measured and computed relative Mach number at 90% span at two different azimuthal locations.16. the pressure drops on the blades suction surfaces as a consequence of flow acceleration. It is seen that the shock nearly covers the gap between 67 . Figures 4.13 show shear layers over the blade and in the wake. respectively.14 and 4. this may be due to the lack of adequate measurement locations in the viscous regions. This figure demonstrates how the pressure through the compressor builds up. Figure 4. The horizontal axes were normalized by the chord at the corresponding span locations. The comparisons generally agree well and a shock structure was observed. The static pressure contours and the shock locations in the four flow passages at two different radial locations. At 70% span. The computed relative Mach number showed a maximum difference of five percent between the baseline grid and fine grid. All of the four flow passages showed identical pressure distributions. 30% and 70% span. indicating the flowfield is identical and well converged in all blade passages. Zero percent and 100% chord on these plots represent the blade leading and trailing edge. are illustrated in Figure 4.17 shows the relative velocity profile at mid-span on three blade passages. As mentioned earlier. in the same way an airfoil behaves. viewed from the top of the rotor.

19 and 4. as shown in Figure 4. as discussed by Hah91. These figures and the preceding results showed that the baseline grid has enough resolution to capture all the important phenomena occurring in this compressor. relative velocity vectors near the blade surfaces and inside the boundary layer are plotted in Figure 4. Below 40% of span from the tip of the blade. The shock is very strong from the tip to 40% of the span and causes the boundary layer to separate immediately aft of the shock. The relative velocity vectors on a plane normal to the rotor axial direction is given in Figure 4. Velocity profiles are well behaved in the blade passages with no evidence of reversed flow.19.18. To visualize the shock-boundary layer interaction.20. is more streamlined because no shock forms on this surface boundary layer. Such a well-behaved flow is. but is related to a conventional adverse pressure gradient.20. Flow properties and the velocity flowfield were almost identical from blade to blade. shown in Figure 4. the flow is also strongly influenced by a vortex roll-up on the leading edge close to the blade root.the blades. of course. Near the suction side just after the shock. This figure shows the cross flow and does not show any stall cells. suggesting choke conditions are imminent. Flow near the pressure side of the blades. a very strong outward flow in the radial direction was observed. In these figures the results shown are on both fine and baseline grid. Because of the large number of computations involved in surge and rotating 68 . As pointed out by Chima89. expected at the peak efficiency conditions. This outward flow pattern has also been reported in References [89] through [92]. the outward flow is not directly due to the shock-boundary layer interaction.

as well as in the passive and active bleed control calculations. Small regions of reversed flow between the blade tip and the shroud were observed in an enlarged view. This reversed flow and the leakage tip vortex play dominant roles in triggering surge and rotating stall. The overall flow field is well attached and behaved. Figure 4. the baseline grid was used in simulations of off-design operating conditions. as discussed later. shown in the same figure.stall simulations. 69 .21 illustrates the meridional velocity profile at the peak efficiency at mid-passage.

Reference [82].4 cm Figure 4. 90 70 50 30 10 Figure 4.2 Laser anemometer and aerodynamic locations for Rotor 67.1 NASA Rotor 67 configuration. Reference [90]. 70 .51.

3 Computational grid for Rotor 67. Figure 4.4 Streamwise computational grid at midspan for Rotor 67. 71 .Figure 4.

5 Pitchwise computational grid for Rotor 67.6 Computational grid in the meridional plane for Rotor 67. Figure 4.Figure 4. 72 .

8 Laser anemometer and aerodynamic locations for Rotor 37.Figure 4.7 NASA Rotor 37 configuration. Figure 4. 73 . Reference [90]. Reference [84].

2 0 -0.9 Computational grid for Rotor 37.6 0.10 Time history of mass flow rate fluctuations about mean flow at peak efficiency conditions (Rotor 67). 74 . Ωt/2π Figure 4.2 -0.Ω Figure 4.6 -1 1 2 3 Rotor Revolution. Massflow Rate Fluctuations (kg/s) 1 0.

00 Fraction of Span from Hub to casing Figure 4.70 0.60 0.80 1.50 30 % Pressure Fluctuations -40 -30 -20 -10 10 -10 0 10 20 30 40 -30 -50 % Mass Flow Rate Fluctuations Figure 4.11 Pressure and mass flow rate fluctuations at peak efficiency conditions (Rotor 67). 0.20 0.60 0.40 0.10 0.12 Comparison between experiments and present results of inlet axial velocity at peak efficiency conditions (Rotor 67).00 0. 75 .50 U/VSTD 0.00 CFD-Baseline Grid Laser Measurment CFD-Fine Grid 0.40 0.20 0.30 0.

76 .13 Comparison between experiments and present predictions of relative Mach number contours at 30% span at peak efficiency (Rotor 67).Computed Measured (Reference 87) Figure 4.

8 0.4 1.2 M 1 CFD-Basedline Grid CFD-Fine Grid experimental 0.6 1. 77 .15 Comparison between experiments and present results of relative Mach number at 90% span and 50% pitch at peak efficiency (Rotor 67).6 -125 -50 25 % Chord 100 175 Figure 4.1.6 1.8 0.6 -125 -50 25 % Chord 100 175 Figure 4.Fine Grid experimental 0. 1.4 1.14 Comparison between experiments and present results of relative Mach number at 90% span and 30% pitch at peak efficiency (Rotor 67).2 M 1 CFD-Baseline Grid CFD.

78 .16 Static pressure contours at 30% and 70% span at peak efficiency (Rotor 67).30% Span 70% Span Figure 4.

Figure 4. 79 .Figure 4.18 Secondary flow in the circumferential direction at peak efficiency (Rotor 67).17 Relative velocity profile at mid-span at peak efficiency (Rotor 67).

80 .Shock TE LE Baseline Grid Fine Grid Figure 4.19 Flow field near blade suction surface at peak efficiency (Rotor 67).

Casing Hub Baseline Grid Fine Grid Figure 4.20 Flow field near blade pressure surface at peak efficiency (Rotor 67). 81 .

Casing Tip LE ( Figure 4. 82 .21 Flow field at mid-passage at peak efficiency (Rotor 67).

velocity vectors and pressure contour plots of the flowfield. the stable operating. no azimuthal disturbances (stall cells) were observed and the flow was circumferentially uniform. Results include the characteristic performance maps. Investigation of the flowfield at off-design conditions is one of the goals of this study. and time traces of pressure upstream of the compressor face using computational probes located around the circumference of the compressor. Furthermore. as the mass flow rate through the compressor is decreased from this stable point to the surge line. are presented at different operating conditions. disturbances grow and lead the system to unstable conditions. the time history of the flow at selected points on these curves. the mass flow and pressure fluctuations are very small. Results for the Rotor 67 compressor are presented in Section 5.1.2. peak efficiency condition results for the compression system NASA Rotor 67 were discussed. However. 83 . as pointed out in Chapters I and II.5 CHAPTER V RESULTS AT OFF-DESIGN CONDITIONS In the previous chapter. The calculations for the Rotor 37 are given in Section 5. It was shown that at this operating point. and is documented in this chapter In this chapter. the computational results for the two axial compressors introduced in Chapter IV.

The radial distributions of total temperature and stagnation pressure are mass averaged across the annulus. and the operating point obtained. In Section 5.1. the characteristic performance map of a compressor is obtained by changing the compressor mass flow rate using a throttle valve in the plenum.1.1. The backpressure is calculated from Equation (3-56) with a user specified mass flow rate at the plenum exit. From the pressure difference between the compressor inlet and diffuser exit. Stalled flow results are given in Section 5. In this study. the mass flow rate through the compressor can be computed.5. results at the onset of stall are presented.1 ROTOR 67 RESULTS Results from CFD unsteady calculations for the compressor NASA Rotor 67 at two different operating conditions are discussed in this section. This kind of normalization removes uncertainties in the experimental mass flow rates.1 ONSET OF STALL As mentioned earlier. the flow solver was applied to the NASA Rotor 67 for 10 different operating conditions to construct the overall rotor performance map.96 kg/sec and the 84 .1. 5.1. The calculated total pressure ratios at full speed are plotted against the normalized mass flow and compared to experimental data in Figure 5.2. The reported choked mass flow rate for the NASA Rotor 67 is 34. Both calculated and experimental data are nondimensionalized by their corresponding choked mass flow rate. the performance map of a compression system is the relation between the total to total pressure ratio and the compressor mass flow. In real life.

7% difference). Point A in Figure 5. The adiabatic efficiency of a compressor is the ratio of the ideal input work needed to raise the total pressure of a working fluid from a pressure value P01 to a new value P02. the CFD calculations for the other two points. A.calculated value is 34. and its corresponding results were studied in the previous chapter. indicated on the performance map. It is seen that the performance map agrees well with the experimental data. . The adiabatic efficiency of the compressor can be found by using: η ad = Ws Wa (5-1) where Ws and Wa are the isentropic and actual work done on the flow. B and C.1 is related to the peak efficiency condition.72 kg/sec (0. are the operating points where the detailed flowfield was studied. to the actual work needed to accomplish this task. B and C. Three points. Here. respectively. (5-2) Applying isentropic relationship for point 1 and 2: 85 . These can be found as follows: Ws = m c p (T02 s − T01 ) Wa = m c p (T02 − T01 ) . are discussed.

T02 T01 s p  =  02  p   01  γ −1 γ (5-3) and using Equations (5-1) to (5-3). we get : η ad  p02  γ   p  −1   01  = T02 −1 T01 γ −1 (5-4) The radial distributions of total temperature and stagnation pressure are mass averaged across the annulus and are obtained from Equation (4-1) and the following equation: T02 = T01 ∫∫ T T02 01 ρVdA (5-5) ∫∫ ρVdA Figure 5. The predicted peak efficiency occurs at a slightly higher mass flow rate as compared to the experimental results. 86 .2 shows the comparison of measured and computed adiabatic efficiency for different operating conditions.

All the probes showed same amount of deviation and are very close to zero. 16% of its mean value.6% difference). four “numerical” probes were circumferentially located in a stationary system 30% of the tip chord upstream of the rotor blade row. or 32.11. no 87 .57 and 0.Stall inception in compressor experimental studies is normally detected by hotwire probes placed in a stationary frame around the circumference of the compressor.5.6 are shifted vertically by a constant interval. the time traces shown in Figure 5. The computed mean mass flow rate was 31.4 shows the fluctuations of mass flow rate and total pressure ratio for operating point B on the compressor map in Figure 5. At each computational time step the probes calculated the axial component of the flow velocity and the corresponding pressure. Figures 5. it can be seen that the fluctuations at this point increased by about 15%.6 kg/sec and the corresponding measurement data at the onset of stall is 92% of choked flow.6 illustrates the deviations of the static pressure of each of the four probes from their azimuthally averaged values. The spatial average of static pressure from the probes upstream of the compressor face is shown in Figure 5. For clarity. In the other words. The averaged pressure fluctuated between 0.67 of the about value at the inlet face. Compared to the mass flow and pressure fluctuations at point A in Figure 4.1 kg/sec.3. indicating the flow was periodic from blade to blade. The probes were placed in the mid-pitch of each of the flow passages and at 90% of the span from the hub. Figure 5.1. In the present study. (~1. The amplitude of the flow oscillations at point A was less than 1%. as shown in Figure 5.

A comparison of Figures 5.4 indicates that as the mass flow rate through the compressor is decreased by six percent. was observed.1.9 and 5. equal to 84 Hz. from point B to point C in Figure 5.6% of the choked mass flow rate (29.evidence of circumferential disturbances was observed.1. The total pressure ratio fluctuation versus the mass flow fluctuation is illustrated in Figure 5. the amplitude of the total pressure ratio oscillation increases from 10% to 50%. Limit cycle oscillations with a frequency of about 1 blade 70 passage frequency. The axial velocity fluctuations were small and there was no reversed flow for this operating condition. even though the amplitudes of the mass and total pressure fluctuations are about 15% and 10% of the mean values. it was shown that at peak efficiency conditions.8 shows the time history of the mass flow rate for about 25 rotor revolutions.7 depicts the time traces of the spatial average of the axial velocity component upstream of the compressor face. generally the flow was well aligned and just a small region of reversed flow in the clearance gap was 88 . Vector plots in meridional planes and pressure contour plots in circumferential planes were used to study the effects of these large flow fluctuations on the flowfield.4 kg/sec) causes the compressor to experience very large amplitude fluctuations of both mass flow rate and total pressure ratio. In Chapter IV.2 ROTATING STALL AND MODIFIED SURGE SIMULATIONS Decreasing the mass flow rate to 84.9. Figure 5. Figure 5. 5. respectively.

These vector plots are shown in the meridional plane at the midpitch of one of the flow passages. precursor level and stall level. and lasted for approximately half of the rotor revolution. II. were very small. The spatial averages of axial velocity and static pressure from the probes readings are shown in Figure 5. darker regions represent low-pressure areas and the lightly shade regions correspond to high-pressure areas.21). for one compressor revolution.12. It was observed that the low-pressure regions are moving in the opposite direction of the shaft rotation. in stall level (section III). Figure 5. section II). section I. indicated by sections I. These levels can be called as recovery level. Three different levels of amplitude with high frequency fluctuations. After about one revolution. about 3% of their average. A power spectral density analysis was performed to determine the dominant frequencies of the averaged static pressure fluctuations. The static pressure contours in the circumferential plane upstream of the compressor face for operating conditions C are illustrated in Figure 5. Finally the stall cycle leads to a very large fluctuations.observed (see Figure 4.10 illustrates the growth of this reversed flow and flow separation at operating point C at three different time instances. The amplitude of fluctuations for both pressure and axial velocity in the recovery level.11 at one instance in time. in a rotating frame. In this figure. were observed in this figure. the amplitude of fluctuations increased to 20% (precursor level. The rapid growth of the tip leakage vortex over one and half rotor revolution causes some reversed flow in the inlet near the casing. 50%. The results are shown in Figure 89 . and III.

In this study only Rotor 37 was simulated.13. Figure 5. As mentioned earlier. The experimental data are for the Stage 37. It was observed that the probes experienced same axial velocity magnitude for about 1. followed by a different velocity magnitude in each probe for approximately another 1. This figure shows that the dominant frequency of these azimuthal disturbances is about 100 Hz.11 to 5.5 rotor revolutions. the time traces shown in this figure are shifted vertically by a constant interval. Bright et al. it is believed that the compressor at this operating condition.2 ROTOR 37 SIMULATIONS The second configuration. As in Figure 5.12) for all of the four probes.6. which includes Stator 37 and Rotor 37. No one-toone comparisons with the experimental measurements were attempted. is experiencing a three dimensional instability that is similar to a modified surge. and the effects of stator and rotor interaction were not evaluated. point C.5 revolutions. From the above discussion. 90 . Figures 5.14 illustrates the deviation of axial velocity from the spatial averaged axial velocity (shown in Figure 5.5. which is about 38% of the rotor frequency. which was briefly studied.14 indicate that the flow is not symmetric from one blade passage to the next and a part span rotating stall occurs. was the compressor NASA Rotor 37. which is a combination of rotating stall and surge.96 at NASA Glenn Research Center have obtained surge and rotating stall data for this compressor. due to date availability. 5.

The time history of the mass-flow rate and total pressure ratio at the selected points A.15. The computed result for point C was about 12 kg/sec. B and C. Good agreement between the computed results and measured data was observed. Figure 5.18.17 indicates that the 91 .1 was applied to study compressor operation at 70% of design speed. Figure 5. Spectral power density for these azimuthal disturbances was plotted in Figure 5. and at this operating point calculations showed very small fluctuations (~1%).The flow solver was applied to the NASA Rotor 37 for seven different operating conditions. are plotted in Figure 5. To investigate azimuthal disturbances. which was about 12000 RPM.0 kg/sec. The same methodology used in Section 5. computational probes one tip chord upstream of the compressor face were used. with 15% fluctuations in the mass flow. The mass flow rate for point A was 15.1.5 kg/sec and the mass flow fluctuations increased to 5% of the mean flow. indicated on the performance map.16. the compressor performance map was extracted and is shown in Figure 5. This figure revealed that these disturbances had a very high content frequency (1-8 kHz) with small amplitudes.17 shows the deviation of pressure magnitudes from the circumferentially averaged pressure. the mass flow was 13. This figure indicates that the pressure disturbances were small compared to the 15% fluctuations in mass flow at this operating condition. These high frequency small disturbances may be due to fine scale boundary layer phenomena. From these results. This RPM corresponds to the compressor RPM in the NASA test. At point B.

92 . We call this a surge because the pressure fluctuations at all four probes are nearly identical. which may referred to as a mild surge.compressor experienced a one-dimensional disturbance. with no phase difference indicative of a rotating stall.

75 1.8 1.9 .6 1.5 1.94 0.2 Comparison of measured and computed adiabatic efficiency (Rotor 67).35 Experiment 1. m Choked Figure 5.94 0.86 Experiment CFD 0.92 0.65 Total Pressure Ratio 1.7 1. 0.94 .9 0.96 0.3 0.98 1 m . 0.86 0.84 0.96 0.84 0. C B A 0.1.88 0.4 CFD 1.92 0.9 0.45 1. mChoked Figure 5.88 0.92 Efficiency 0.98 1 m . 93 .88 0.82 0.1 Comparison of measured and computed characteristic performance map (Rotor 67).55 1.

I II III IV

IV

III

II

I

Figure 5.3 Locations of the computational probes (Rotor 67 and Rotor 37).

94

50

30

% Pressure Fluctuations
-40 -30 -20 -10

10

-10

0

10

20

30

40

-30

-50

% Mass Flow Rate Fluctuations

Figure 5.4 Pressure and mass flow rate fluctuations at the onset of stall, operating point B (Rotor 67).

P Pref

Rotor Revolution, Ωt/2π Figure 5.5 Time history of averaged pressure computed by probes upstream of the compressor face at operating point B (Rotor 67).

95

P−P Pref

Rotor Revolution, Ωt/2π Figure 5.6 Time history of pressure deviation from its azimuthally averaged pressure upstream of the compressor face at operating point B (Rotor 67).

u a∞

Rotor Revolution, Ωt/2π Figure 5.7 Time history of averaged axial velocity computed by probes upstream of the compressor face at operating point B (Rotor 67).

96

9 Pressure and mass flow rate fluctuations at operating point C (Rotor 67).8 Time history of mass flow rate fluctuations at operating point C (Rotor 67).. 97 . 50 30 % Pressure Fluctuations -40 -30 -20 10 -10 -10 0 -30 -50 10 20 30 40 % Mass Flow Rate Fluctuations Figure 5. Ωt/2π Figure 5. m (kg/sec) Rotor Revolution.

98 .a) After ½ Rotor Cycle b) After 1 Rotor Cycle c) After 1.5 Rotor Cycle Figure 5.10 Temporal growth of reversed flow at operating point C (Rotor 67).

Low static pressure regions Ω Figure 5. 99 .11 Instantaneous circumferential pressure fields at operating point C (Rotor 67).

100 .I II III u a∞ P Pref Rotor Revolution.12 Time history of averaged axial velocity and pressure computed by probes upstream of the compressor face at operating point C (Rotor 67). Ωt/2π Figure 5.

101 .13 Power spectral density of the azimuthally averaged pressure fluctuations upstream of the compressor face at operating point C (Rotor 67). u −u a∞ Rotor Revolution. Ωt/2π Figure 5.14 Time history of axial velocity deviation from its azimuthally averaged axial velocity upstream of the compressor face at operating point C (Rotor 67).Power Spectral Density of Pressure Frequency Hz Figure 5.

1.6
Total Pressure ratio
B C A

1.4

1.2
Experiments CFD

1 10 12 14 16 18
Mass Flow Rate (kg/sec) Figure 5.15 Computed and measured characteristic performance map at 70% design speed (Rotor 37).

Figure 5.16 Fluctuations of total pressure ratio versus mass flow rate at different operating conditions (Rotor 37).

102

P−P Pref

Rotor Revolution, Ωt/2π Figure 5.17 Time history of pressure deviation from its azimuthally averaged pressure upstream of the compressor face at operating point C (Rotor 37).

Power Spectral Density of Pressure Deviation

Figure 5.18 Power spectral density of pressure deviation from its azimuthally averaged pressure fluctuations upstream of the compressor face at operating point C (Rotor 37).

103

6

CHAPTER VI

ACTIVE CONTROL STUDIES OF ROTOR 67 CONFIGURATION

Axial compressors operate in regimes where both rotating stall and surge impose limits on low flow rate operability. The useful operating range can be as small as 10 percent of the design flow rate. For example, in the compressor Rotor 67 described in the previous chapters, the surge line occurs at 92% of the choked flow. In the other words, the stable operating range in this rotor is just 8% of the maximum delivery flow. Therefore, control methodologies are necessary for extending the stable operating range of compressors. Control strategies designed to enhance the operability of compressors must address rotating stall, surge, and their interaction. In recent years, there has been much work on improving operating range of compression systems using active and passive control methods. These methods were discussed in Section 2.3 and schematically shown in Figure 2.7. Ever since the Moore-Greitzer23-24 model became available, work on feedback stabilization methods for compressors has been extensive and a considerable number of papers have been published. One of the most common of these methods is bleeding. Since its application in early compressors more than 60 years ago, bleeding has been used for engine start-up purposes. Bleeding has also been used over the past two decades to shift the surge line to lower mass flow. Fisher41 used a bleed valve in a centrifugal

104

42 studied centrifugal surge control using throttle valves as actuators. At the maximum speed. In a series of experiments.98. The first one is a simple steady bleeding control. results for 105 . In Section 6. The second type is an unsteady bleeding linked to the pressure fluctuations upstream of the compressor face. and a 25% reduction of mass flow rate prior to onset of surge was achieved. The amount of bleeding has been reported ranges from 1 to 10 percent of the mean flow40-43. They showed that by one to two percent air injection. and Wang et al. In the series of papers published by Yeung and Murray 40. analysis of bleed valve rate requirements for suppressing rotating stall are presented. This is referred to as closed-loop control. Yeung et al.compressor at various compressor shaft speeds. Pinsley et al. the rate of bleeding could be reduced by 30%. referred to as open-loop control.99 at Georgia Tech used bleeding in the plenum to reject surge in axial and centrifugal compressors.1. The bleed valves in both cases were located one tipchord downstream of the rotor blades.97 were one of the first groups to report successful experimental implementation of a bleed valve controller. Prasad et al. the surge line was decreased by 21% toward the low mass flow rate side. Between 50000 and 70000 RPM. this shift was 28% to 31%. 78000 RPM. at Cal Tech. two types of bleeding were considered. In this study. Eveker et al. This chapter discusses the CFD results for the two control methods applied to the NASA Rotor 67 compressor.40experimentally showed that the rate of bleeding may be reduced by injecting air downstream of the compressor face.

As pointed out in Section 3. a 3. The fraction of mass flow rate and the locations of the bleed valves are specified by the user. 6.2% of the mean flow bleed was applied to the unstable operating condition at point C. It may be recalled that the compressor experienced modified stall at the unstable branch (B-C) of the compressor map. A comparison between Figures 6.1.5 rotor revolutions. to simulate the open-loop control.10 (reproduced here for convenience) shows that the compressor casing is seen to restore attached flow over much of the whole compressor.6.1 OPEN-LOOP CONTROL STUDIES In open-loop control. In this study.2 and 5. A schematic of the steady bleeding is shown in Figure 3. 106 . a fraction of the mass flow rate is removed through a valve placed on the compressor casing. which can be tailored to enhance the stable operation of the compressor. shown in Figure 6. as shown in Equation (3-57). A bleed valve will have its own throttle characteristic. Section 6. it was assumed that flow is removed at a constant rate in an azimuthally uniform rate. Such a steady bleeding is inefficient and must be turned off during design operations.the open-loop control are presented.2 shows the velocity vectors at the azimuthal plane at midpitch with the passive bleed control after 1. To see the effects of the open-loop control on the flow field. Figure 6. the normal velocity component is calculated from the amount of the bleed flow and the bleed area.6.2 deals with the results for the closed-loop control.

Referring back to the performance map with open-loop control shown in Figure 6. which is 87. m p . 107 . it is seen that the stable operation of the engine has been extended to 87. .9 and Figure 6. the horizontal axis is the mass flow rate through the compressor before . . and the vertical axis is the total pressure ratio for the whole system including the bleeding. which is m c − m b .2% bleed brings the compressor to a new operating point. and the mean total pressure ratio was 1.4 kg/sec. giving effectively a 38. bleeding.Figure 6. Also. Another way to compare the controlled-operating point D in the performance map with the other operating conditions is given in Figure 6.2% bleed from the operating point C brings the compressor .2% increase in the operating range of the compressor. a 3. This operating point is shown in the Figure 6.4. A 3. and reduces the totalto-total pressure ratio of the rotor. m c .3 shows the fluctuations in mass flow rate and total pressure ratio for this steady bleed condition.1.6% of the choked mass flow rate. In this figure. This is because removing some high-pressure air from the compressor causes a reduction in the total pressure at the rotor exit.5% of the choked flow rate. where m b is the bleeding mass flow rate through the bleed valve. . In Figure 6.3 (reproduced here for convenience) reveal that this open-loop bleeding reduces the amplitude of total pressure oscillations by 75%. a comparison between Figure 5.1.1. The computed mean mass flow rate entering the plenum was 30. D. indicated by point D. As can be seen. the vertical axis represents the non-dimensional pressure rise and the horizontal axis is the mass flow rate entering the plenum.65.

the corresponding throttle characteristic Kt. In other words. the operating point D has the same throttle characteristic as point F in the performance map. Without open-loop control the system will enter into modified surge at mass flow rates below 31. A comparison between Figure 6. Figure 6.4.6 kg/sec (point B).5 shows a comparison between the non-dimensional axial velocity at 108 . .. where m t is the throttle mass flow rate. valve. (6-1) The throttle characteristic map was obtained by using the calculated Kt at point D and Equation (6-1) for different mass flow rates. .4 shows that even though the bleeding was applied to point C. . From this point of view. From a knowledge of m t and gage pressure ∆p at throttle . the dashed line represents all the points having the same throttle position. to a new operating condition D. To understand why and how the bleed-air mechanism works. the branch B-D is a stable branch of the performance map. is 29. Under steady state conditions m p = m t . The reason for choosing the horizontal axis to be m p is to compare the throttle position for this point with the throttle positions for other points in the compressor performance map.43 kg/sec.1 and Figure 6. and is plotted as a dashed line in Figure 6. This is a useful extension of the operating conditions. representing flow that goes into downstream components. was calculated as follows: m t = K t ∆p . bringing the compressor to the operating point D. the flow field was examined. The useful mass flow rate.

It is seen that local reversed flow regions first originate near the blade leading edge. compressor is in stable operation.point C.7 (reproduced here for convenience) reveal that the open-loop control eliminates the stall cells. the probes showed only very small fluctuations and reversed flow was completely eliminated. If left unchecked. four “numerical” velocity probes were implemented circumferentially at the upstream of the compressor face.7 illustrates the deviation of axial velocity from the spatially averaged axial velocity for all of the four probes.6). after three different rotor cycles.8 show that the 109 . Compared with results for point C in the Figure 5. and point D.2% of the total flow rate entering the inlet was removed from the compressor.6 and 6.8. the reversed flow regions near the leading edge disappeared and a stable operating condition was restored As in operating points B and C. When 3. A comparison between Figure 5. a stalled and unstable condition. It is observed that all of the probes are identical and no circumferential disturbances indicative of rotating stall were observed. leading the compressor to an uncontrolled growth of mass flow rate and pressure fluctuations. this flow region grew temporally and spatially. The time traces shown in this figure are shifted vertically by a constant interval. The spatially averaged values of axial velocity and static pressure from the probes readings are shown in Figure 6. a stable condition achieved with bleed valve control. The data was acquired near the leading edge at mid-passage of the compressor.6 through 6.3. Figure 6.12 (reproduced as part of Figure 6. Figures 6. to trace the axial velocity component. shown in Figure 5.14 and Figure 6.

is related to the gage pressure ∆p at the bleed location: . The bleed valve was activated whenever the pressure sensors in the upstream of compressor face experienced high-pressure fluctuations. The amount of air that is removed. and activated the bleed valve. Therefore. which monitored by a control unit. Kb is the bleed valve constant and is related to the valve geometry. m b .6. the effects of open-loop control on the flowfield were discussed. The mass flow of the bleed valve is linked to the instantaneous amplitude of pressure fluctuations acquired by the “computational” probes. the bleed rate was linked to pressure amplitudes upstream of the compressor face. a closedloop active control methodology was studied. such a steady preset bleeding is inefficient and must be turned off when the compressor operates close to design conditions. determined the amount of bleeding needed. Quantity Ab is the bleed valve area and 110 .2 CLOSED-LOOP CONTROL STUDIES In the previous section. mb = K b Ab ∆p (6-2) In Equation (6-2). and results are presented in this section. This control unit received the pressure signals from some probes located upstream of the compressor face. The closed-loop control was applied to the NASA Rotor 67 configuration. In this approach. . As mentioned earlier.

The normal velocity at the bleeding points was calculated using Equation (6-2): un = k b ∆p ρb (6-4) Figure 3. The following parameters. and the two components of tangential velocity were extrapolated from the interior.7 illustrates the schematic of the closed-loop diffuser-bleed valve control used in this study. As in open-loop bleeding.∆p = pb − p∞ (6-3) where pb and p∞ are the pressure at the bleed location and ambient pressure. which are user inputs. if the pressure values at the 111 . At each time step. The bleed valve was placed one tip chord downstream of the rotor. The pressure. respectively. identify the active bleed configuration: Bleeding locations Pressure sensor locations Bleed characteristic constant Permitted upper and lower limits of pressure for the pressure sensors In this study. the bleeding under feedback control was circumferentially uniform on each of the flow passages. density. the “numerical” probes were located at 30% tip chord upstream of the compressor face.

Figure 6. Figure 6.probes were out of the permitted fluctuation limits.8. In this figure. Figure 6. and plot the throttle 112 .1 is the mass flow rate through the compressor before bleeding.54P∞. when closed-loop control was used. This operating condition is referred as point E in the performance map. As pointed out earlier. Figure 6. The characteristic constant value Kb in Equation (6-2) was chosen to be 3. the horizontal axis in Figure 6. a methodology similar to the one used in openloop control was used to compute the throttle characteristic. the upper permitted limit of the inlet pressure fluctuations was set to 0.1. the bleed valve removed some air from the compressor. The closed-loop control was applied to point C. the recovery and precursor levels. in Figure 6. As mentioned earlier.9 shows the circumferential averaged pressure fluctuation of the four probes at operating point C without bleed control. Kt. This value was chosen to ensure the rate of mass bleeding flow was less than 2% of the mean flow.10 shows the fluctuations of total pressure ratio versus mass flow rate fluctuations. Figure 6. three levels of amplitudes were observed. A comparison with Figure 5. respectively.67P∞ and the lower one was chosen 0.1. The bleed valve was not activated for the first two lower amplitude levels. As shown in Figure 6.0. the reduction in the total pressure fluctuations was larger for open-loop control.9 shows that the amplitude of fluctuations were reduced by 30% and 40% in total pressure ratio and mass flow rate.1 shows that the closed-loop approach restores the stable operating condition at a higher pressure ratio. However.11 shows the pressure rise across the compressor stage versus the mass flow rate entering the plenum.

Also. whereas at point E (closed-loop control). it may be concluded that the compressor is in a stable operating condition at point E. The averaged bleed flow rate at operating point E was about 1.12. From Figures 6.41 kg/sec.14. The spatial average of axial velocity from the probes’ readings is shown in Figure 6. the mass flow rate was 30. it can be seen that the controlled operating point E has the same throttle position as the unstable operating point G.2%. and no reversed flow was observed. The amplitude of these disturbances was only 20% of the disturbance amplitude for the unstable operating point C. from Figure 6. Figure 6.8 percent of the mean flow rate.10 and 6.13.characteristic map.11. it was 3.11. At the operating . 113 . It is observed that the probes show small circumferential disturbances.03 kg/sec.14. The fluctuations were large compare to the 3. It should be mentioned that the air bleeding rates for operating points D and E were different.13 illustrates the deviation of axial velocity from the spatially averaged axial velocity shown in Figure 6. shown in Figure 5. the amplitude of the axial fluctuations was reduced.49 kg/sec without instabilities. From Figure 6. Compared to results for point C in the Figure 5.2% steady bleeding case. it can be concluded that the useful mass flow rate can be as low as 29. point D (open-loop control) the mass flow rate m p was 30.12. The velocity time traces shown in this figure are shifted vertically by a constant interval. where at point D.

It is clear that both open-loop and closed-loop controls provide a useful extension of the stall margin in the system.The table below summarizes the calculations reported in this study. .3 Table 6.541 29. Cases studied mc mb (kg/sec) (kg/sec) m p = m c − mb (kg/sec) .437 38.49 50. . 114 .973 29.41 0.03 0. .4 Unstable 30.2 30. .4 ____ 29. . % o Extension of operability margin m Onset of Stall − m c m Choked − m Onset of Stall . . x100 Stalled condition operating point C Open-loop control operating point D Closed-loop control operating point E 29.1 Summary of stall margin extension for the two control schemes studied. .

5 1.92 0. 0.96 0.7 1.75 B C E D 1.94 0.9 . 115 .86 0.8 1.88 0.82 CFD with Open-Loop Control CFD with Close-Loop Control Experiment 0.& mb & mc & mP Plenum Chamber & mt 1. m Choked Figure 6.35 1.98 1 mc .1 Characteristic performance map with bleed control (Rotor 67).84 0.4 1.55 1.65 Total Pressure Ratio A 1.3 0.45 Stable Controlled Conditions CFD without Control 1.6 1.

Without control. 116 . 3.2 Velocity profile with/without open-loop control (Rotor 67).2% Bleed With open-loop control. Figure 6. operating point D. operating point C.

Without control. 117 .3 Total pressure ratio and mass flow rate fluctuations with/without openloop control (Rotor 67). operating point C 50 30 % Pressure Fluctuations -40 -30 -20 10 -10 -10 0 -30 -50 10 20 30 40 % Mass Flow Rate Fluctuations With open-loop control. operating point D 50 30 % Pressure Fluctuations 10 -40 -20 -10 0 20 40 -30 -50 % Mass Flow Rate Fluctuations Figure 6.

8 0.6 28 29 C D F B A CFD without Control CFD with Open-Loop Control Throttle Characteristic 30 .2 1. 31 32 33 34 35 m p (kg / sec) Figure 6.7 0.1 ∆P Pref 1 0.9 0. 118 .4 Characteristic performance map with throttle characteristic for openloop control (Rotor 67).& mb & mc & mP Plenum Chamber & mt 1.

5 Spanwise distribution of axial velocity at mid-passage near the leading edge (Rotor 67).20 0.60 0. 119 .100 80 After 1.5 revolution 3.40 0.5 revolution 60 40 20 0 0.2% bleeding -0.00 After 0.40 -0.20 U/Uinf 0.80 Figure 6.

Without control. Ωt/2π With open-loop control. operating point C u a∞ Rotor Revolution. 120 . operating point D u a∞ Rotor Revolution.6 Time history of axial velocity with/without open-loop control (Rotor 67). Ωt/2π Figure 6.

operating point C u −u a∞ Rotor Revolution. operating point D u −u a∞ Rotor Revolution. Ωt/2π With open-loop control.7 Time history of axial velocity deviation from its azimuthally averaged axial velocity with/without open-loop control (Rotor 67). 121 .Without control. Ωt/2π Figure 6.

P Pref Rotor Revolution.9 Upper and lower limit of pressures that trigger closed-loop control (Rotor 67). 122 . Pressure upper limit P Pref Pressure lower limit Rotor Revolution. Ωt/2π Figure 6.8 Time history of static pressure under open-loop control (Rotor 67). Ωt/2π Figure 6.

50 30 % Pressure Fluctuations 10 -30 -20 -10 -10 0 -30 -50 10 20 % Mass Flow Rate Fluctuations Figure 6. 32 33 34 35 m p (kg / sec) Figure 6.11 Characteristic performance map with throttle characteristic for closedloop control (Rotor 67). 1.7 0.6 28 CFD with Close-Loop Control Throttle Characteristic 29 30 31 .9 0.8 1 E A CFD without Control 0.2 1.10 Total pressure ratio and mass flow rate fluctuations under closed-loop control (Rotor 67).1 C G B ∆P Pref 0. 123 .

124 . u a∞ Rotor Revolution.13 Time history of axial velocity deviation from its azimuthally averaged axial velocity upstream of the compressor face under closed-loop control (Rotor 67).u −u a∞ Rotor Revolution. Ωt/2π Figure 6.12 Time history of axial velocity under closed-loop control (Rotor 67). Ωt/2π Figure 6.

a three-dimensional unsteady Navier-Stokes analysis capable of modeling multistage turbomachinery components has been developed. 125 . The analysis can also use a previously stored solution file with the corresponding grid file. Open-loop bleed control and closed-loop control methods have been studied and are demonstrated to stabilize the system. this is the first application of a fully 3-D unsteady viscous flow analysis for modeling axial compressor control in this fashion. A number of code-validation studies have been done for centrifugal and axial configurations. the conclusions of this research are presented in Section 7. To the author’s knowledge. This method uses a cell-vertex finite volume scheme to solve the governing equations in a body-fitted rotating coordinate system. Studies have also been done to understand some of the mechanisms behind rotating stall and surge.2. In this chapter. A one-equation Spalart-Allmaras turbulence model equation was implemented in the solver to account for the turbulence effects.7 CHAPTER VII CONCLUSIONS AND RECOMMENDATIONS In the present work. The recommendations for further work in the future are given in Section 7. The scheme is third order accurate in space and first or second order in time. to restart the computations.1.

devices and mechanisms that directly alter the pressure levels in the plenum chamber are an effective way of controlling surge.1 CONCLUSIONS The following items are concluded from the current work: 1. These methods are suitable for identifying “hot spots” in the flow field.g. This phenomenon is consistent with experimental observations. demonstrated in this work.7. 2. 3Inlet disturbances can trigger rotating stall. or through active control (e.Three-dimensional unsteady flow solvers such as the one developed here are valuable tools for modeling and understanding rotating stall and surge.g. Therefore. The exchange of energy between the plenum chamber and the flow inside the blade passages is a major mechanism for surge. and for developing flow control devices for the system. These oscillations may roughly be divided into these 126 . It should be possible to mitigate rotating stall through careful redesign of the inlet.Rotating stall and surge phenomena are often accompanied by temporally growing pressure and velocity fluctuations in the inlet upstream of the compressor face.The plenum-chamber downstream plays a critical role in surge development. through guide vanes. 4. for redesigning the system (e. casing treatment). jets). The stall cells in the present This observation has been numerically simulation appear to rotate at 38% of the rotor RPM. in the direction of the rotor rotation.

This law also accounted for the bleed valve 7. 5. This boundary may be written as: 127 . In this work. is to let this nominal mass flow through the throttle also fluctuate with time. and a nominal fixed mass flow rate controlled the compressor operation. a simple unsteady boundary condition was introduced.It should be possible to establish simple rule-based control laws linking the inlet pressure fluctuations to the amount of actuation needed.categories: precursor. stall.2 RECOMMENDATIONS The following items are suggested to extend the current work: 1. Another type of outlet boundary condition. that should be tested. this is the first time precursor waves have been used in a numerical simulation to detect and automatically control rotating stall. characteristics. to trigger open-loop or closed-loop control. and recovery. in particular the precursors. It should therefore be possible to use these oscillations. In this manner both backpressure and throttle mass flow vary at each time step. and modified surge. surge. A very simple law that triggers bleeding whenever the upstream pressures exceeded a preset level user was found to be effective.The outlet boundary conditions at off-design conditions play a dominant role in simulating the instability developments appropriately. While this has been known for some time.

and At is the throttle valve area. The work should be extended to the simulation of multi stage compressors considering rotor-stator interactions. = a2 p Vp & & (mc − mt ) (7-1) m t = K t At ∆P (7-2) where Kt is the throttle characteristic constant and is related to the throttle geometry. the open-loop bleeding control removed a fraction of highpressurized air at a fixed rate from the compressor. an interpolation approach may be used to transfer the information from the stationary block to the rotating block.dp p dt .This work focused on the analysis of flowfield and the fluid dynamic phenomena in single stage compressors with multiple blade passages.In the present research. the rotor grid block. In rotor-stator interactions. Here. will be moving in the pitchwise direction. 3. This may be done stage by stage. e. ∆P is the pressure difference between the plenum pressure or compressor backpressure and the components downstream. 2. similar to the present simulation of the rotors. a second stage or a combustor. providing a weak 1-D coupling between the various components of the compression system and the combustor. At the boundaries connecting the moving and stationary regions. The bleed rate can be 128 . while the stator grid block remains stationary.g.

ωs is the rotating speed of the stalled cells. m b Mean is the mean mass flow removed from the compressor. one near-term possibility is the use of LES methodology on sufficiently fine grids. Recently. These discrepancies must be systematically studied and eliminated. m f is the amplitude of the pulsed bleeding. bleeding mass flow rate. 5. Although a limited number of grid sensitivity studies have been done. an inadequate number of grid points may also have contributed to this error. and ϕ is the phase lag. 4.sinusoidally varied at the rotating stall frequency with an appropriate phase lag. . casing treatment. The one-equation turbulence model may be a source of this discrepancy. and so on.Different types of control devices. The benefit of this methodology is that the controller removes the air only when a stalled cell passes by the bleed valve. air injection control methodology has been computationally simulated70. Thanks to the availability powerful highperformance computers. may be written as follows: mb = mb Mean + m f sin(ω s t + ϕ ) .The present approach underestimates the total pressure rise across the compressor. . The . (7-3) Here. Experimental evidence also exists that shows that air injection may 129 . may be simulated with this solver and their effects on the flowfield can be investigated. such as inlet guide vanes. . mb . .

It is hoped that this work will serve as a useful step for future investigations in the exciting area of non-linear flow control. This method has been validated and used to study stall and surge control concepts. 130 . In summary.reduce the required rate of bleeding40. a first principles-based methodology for modeling steady-state and offdesign operation of axial compressors has been developed. This work should be extended to a systematic study of these concepts.

6. The University of Tennessee. Moscow. Compressors. Verlay London Limited.. 7. 8.. 1985. M. Thermodynamics and Fluid Mechanics of Turbomachinery. 2nd ed.. and Oxford University Press.. Essex. 1989. 9. 11. C. Longman Scientific and Technical.. 5.. P. and Saravanamuttoo. REFERENCES Horlock. Sebghati. 4th ed.. C. A. Cohen. Mir Publications. 10. H. Inc. Butterworths. Addison Wesley Publication Co. 1992.. Cherkassky. N. Pergamon Press. and Peterson.. Mechanics and Thermodynamics of Propulsion. F. 1984. 1963.8 1. London.. T. 1994. 1999. 1980.. Gravdahl.. J. J. T. New York. Vermont. N. 1958. Longman. C. H. Springer. S. Pampreen. 4.. 2. A. Butterworths Scientific Publications. D.. Ferguson. The Centrifugal Compressor Stage. Stow. Axial Flow Compressors. Compressor Surge and Stall. G. Compressor Aerodynamics. Compressor Surge and Rotating Stall Modeling and Control. Fans. Concepts ETI. Rogers. London. C. R. Hill.. 131 .. C. Vol. ¥cer. Pumps. J. Cumpsty. J. Essex. Knoxville.. I and II. Inc. NY. H. Concepts ETI. G. 1996. 1993.. Introduction to Turbomachinery. and Egeland. P. I. P. May 1992. Japikse.. V. Gas Turbine Theory. R. B. H. and Baines. Norwich. 3. Monitoring Axial-Flow Compressor Rotating Stall and Surge. Gostelow. O.. 12. Cascade Aerodynamics. M. and Hirsch. Master’s Thesis.

“Rotating Stall and Surge Control. 98.” Journal of Turbomachinery. H. pp. A. 21. No. 371-380. and Greitzer. pp. O. Nov. 6. London.” Transactions of ASME. Vol. 190-217.. No. pp. 22. Stodola. O. W. Parts I-III.. Moore. 313-336. “A Method for Assessing the Effects of Circumferential Flow Distortion on Compressor Stability. Nov. 1931. M..” Proceedings of the 34th Conference on Decision & Control. F.” ASME Journal of Engineering for Power. T. Georgia Institute of Technology.” Journal of Propulsion and Power. March 1994. Vol. and Thoma. and Part 2. P. K. Pearson.D. Sir Isaac Pitman and Sons Publishers.Dec. 106. 1927. Emmons. “Compressor Surge and Stall Propagation. C. J. 552-561..... 20.” Transactions of ASME. 1932. New Orleans.” ASME Journal of Engineering for Power Transactions of ASME. E. 109. No. B. 141-155. Moore. 1995. 1955. 14. and Grant. 19.13. 132 . Vol. 455-469. “Investigations of the Flow Conditions in a Centrifugal Pump. 17. LA. E. H. Kearton. McGraw-Hill Book Co. Nonlinear Dynamic Analysis and Control of Surge and Rotating Stall in Axial Compression System. 1984. M. Greitzer. pp. E. P. K. April 1976. Turbo-Blowers and Compressors.. Ph. “Surge and Rotating Stall in Axial Flow Compressor: Part 1. Dec. pp. Fischer.. Vol. Badmus.. 1986. Hynes. Dissertation. July 1987. 77.. pp. Theoretical Compression System Model.. de Jager. 18. “Stall Transient of Axial Compression Systems with Inlet Distortion. K. 16. 54. New York. D. 2. Experimental Results and Comparison with Theory. Steam and Gas Turbine. 15. “A Theory of Rotating Stall in Multistage Compressor. W. F.

pp.” Journal of Turbomachinery. 1991. Ph. “Numerical Results for Axial Flow Compressor Instability.” Journal of Propulsion and Power. McCaughan. 29. 133 . “ A Theory of Post-Stall Transients in Axial Compression Systems Part I-Development of Equations. No.” Journal of Turbomachinery. 111.23. Moore. 2.. 434-441. 108. 1999. K. pp. Bonnaure. J. E.” Journal of Engineering for Gas Turbines and Power. E.. 68-76. Eveker. F. pp. Greitzer. L. Haddad.. pp.. No. pp. “Stage-by Stage Post Stall Compression System Modeling Technique. 108. Vol. Modeling High Speed Multistage Compressor Stability. 1991. 111. 240-249. pp. 25. K. W. and Li. 231-239.D. 426-433. 9971005.. H. No. Botros. M.. Massachusetts Institute of Technology. F. Master’s Thesis. K. M. 1994. No. and Greitzer.. Vol.” Journal of Engineering for Gas Turbines and Power.. M. 1986. 24. F. “ A Theory of Post-Stall Transients in Axial Compression Systems Part II-Application. “Developments in Centrifugal Compressor Surge Control-A Technology Assessment. M. P. 116. 1993. M and Moore. Dissertation. 7. 27. 1383-1388.” Proceedings of the IEEE International Conference Control Applications (CCA) and IEEE International Symposium on Computer Aided Control System Design (CACSD). K. Aug. 1989. Oct. F. Jan. 30. Model Development for Active Control of Stall Phenomena in Aircraft Gas Turbine Engines. A. Leonessa. 31.” Journal of Turbomachinery. E. E. “Global Stabilization of Centrifugal Compressors via Stability-Based Switching Controller. Georgia Institute of Technology. 26. April 1986. F. F. McCaughan. and O’Brien. 28. Vol. pp. Oct. Davis. “Application of Bifurcation Theory to Axial Flow Compressor Instability. W.. and Henderson. W. K. 1989.

R. Nordenson. “Reduction of Bleed Valve Requirements for Control of Rotating Stall Using Continues Air Injection. J. R. 112.” SAE Paper 760285. 36. 134 . CT. 1974. M. 1970. Holman. 41. 40. “Some Observations of Effects of Porous Casing on Operating Range of a Single Axial-Flow Compressor Rotor. A. G.” Proceedings of the 1997 IEEE International Conference on Control Application. F. Fisher. and Amann.. 1976. Lee. C. Jan. Pampreen. 37. C. and Murray..” Journal of Turbomachinery.” NASA CR-54559.. “Single Stage Experimental Evaluation of Variable Geometry Inlet Guide Vanes and Stator Blading.. “Variable Compressor Geometry in the Single-Shaft Automotive Turbine Engine. and Atkey. S. K. A. “Effects of Casing Treatment on a Small Transonic Axial-Flow Compressor. N. F. Jones. F. April 1988. “Application of Map Width Enhancement Devices to Turbocharger Compressor Stages. B.” ASME Paper 97-GT375.. 1975. 133-144. Khalid. 1990. and Voit. C. E.. Hartford.. and Greitzer. “The Effect of Variable Geometry on the Operating Range and Surge Margin of a Centrifugal Compressor. pp. 39. Part VI Final Report. 683-690. Oct.” ASME Paper 76-GT-98. 35. pp.” NASA TM X 2120. “The Use of Variable Inlet Guide Vanes for Automotive Gas Turbine Engine Augmentation and Load Control. “A Practical Compressor Casing Treatment. “Effects of Endwall Suction and Blowing on Compressor Stability Enhancement. E. Wallace. W. Sheridan.” SAE Preprint 740166. April 1997. S. 38. C. E. and Kidwell. Yeung.. 1997. 33. 1976.. J.” ASME Paper 75-WA/GT-5. 1970. R.” SAE Paper 880794. M.32. D. 34. Whitfield. H. Vol. B. C.. R. Bailey. A. E.. F.

. H. E..” Journal of Fluids Engineering Transaction of ASME. M.. T. Ffows Williams. E.. J. 279-293. Vol. J. 48. W. L. 1991. 49..” Transactions of the ASME. “Nonlinear Analysis of Rotating Stall. and Fleeter.D. E. 99... No. 1994. E. A.” Journal of Turbomachinery.. and Greitzer. Gysling..” Journal of Engineering for Power Transactions of ASME.. AIAA Paper 91-1898. “Dynamic Control of Centrifugal Compressors Using Tailored Structures. and Rannie. P. S. D. 44.. M. and Dean. Greitzer. Pinsley. Iura. No. P. 47. and Nagano.. “An Experimental Study of Surge in Centrifugal Compressors. “Experimental Investigations of Propagating Stall in Axial Compressors. pp. 115-131. D. P. 1991.” AIAA/SAE/ASME/ASEE Joint Propulsion Conference. G. C. Epstein. 46. J. Paduano. 1977. H. 204-211. pp. 723-732. 50. Vol. Lawless. Guenette. Dissertation. Takata. June 1990. 135 .. Longley. 1989.. M. 45. 116. 5. H. 202-215. pp. pp. Epstein. A. pp.. S. “A Review of Nonlinear Flow Models for Compressor Stability. R. A. R. J. Toyama. 463-471. Massachusetts Institute of Technology.” ASME Paper 90GT-122.” Journal of Propulsion and Power. 1954. Dugundji. and Epstein..” ASME Journal of Turbomachinery. K. “Active Suppression of Aerodynamic Instabilities in Turbomachinery. D.. Runstadler. 1991. 113. 43. “Active Control of Rotating Stall in Axial Compressors. 1972. W.. and Greitzer.42. pp. H. “Active Stabilization of Centrifugal Compressor Surge. “Active Unsteady Aerodynamic Suppression of Rotating Stall in an Incompressible Flow Centrifugal Compressor with Vaned Diffuser.. J. E.” Ph.

June 1986.. E. 3. Hobbs.. 27-44.” ASME Paper 90-GT-42.Part 1: Analysis of Rotor/Stator/Rotor Aerodynamic Interaction.” AGARD Conference Proceedings. and Weingold. A. 56. “Aerodynamic Modeling of Multistage Compressor Flowfields . and Wennerstrom. B.” ASME Paper 97GT-344.” ASME Paper 86-GT-227. “ Numerical Study of the 3D Flowfield in a Transonic Compressor Rotor with a Modeling of the Tip Clearance Flow. pp. 136 . J. Hall. Mulac.. June 1986.” ASME Paper 90-GT-76.” Rech. 57. Escande. N. C. May-Jun 1993. Srivastava.. and Celestina.. pp. Chevrin. P. and Vuillez. Davis. “Navier-Stokes Analysis of Turbine Blade Heat Transfer. “Computation of Internal Flows at High Reynolds Number by Numerical Solution of the Navier-Stokes Equations. “Efficient Hybrid Scheme for the Analysis of Counter-Rotating Propellers. 58. No.. L. 1991. 1986-6. R. Hah. Boyle. P.” ASME Paper 88-GT96. H. June 1990. 52.” Journal of Turbomachinery. 54. R. June 1990. D.. L.. Neuilly Sur Seine.. Vol. “ A Model for Closing the Inviscid Form of the Average Passage Equations. and Veuillot. Dawes. Vol... “Viscous Flow Computations in Turbomachine Cascades. “Three-Dimensional Flowfields Inside a Transonic Compressor with Swept Blades. A. M. Cambier. J. R. E... 382-388. J. R. 113. 1986. “Prediction of Compressor Cascade Performance Using a Navier-Stokes Technique... J. Adamczyk. 1997. C. Aerosp. No. J. 9. A. 55. and Sankar..51... Lakshmi N. W.. D. pp. 53.” Journal of Propulsion and Power. No. 241-251. March 1987. 59. J. 401. L.

P. and Sankar. Georgia Institute of Technology. 28. “Computational Study of Rotating Stall Inception in Axial Compressor. L. Niazi. 63. 65-71. S.. 820-830. and Sankar. He. 2000. 64. A. Oct.. 2000. N. 66. Stein. 1. Jan. A. “Application of a Multi-Block CFD Code to Investigate the Impact of Geometry Modeling on Centrifugal Compressor Flow Field Predictions.. R... 137 . No. 1998. Stein.” Journal of Propulsion and Power. 16. 67. 69. 13. D. 1997. 798-806.” AIAA Journal. ” Journal of Turbomachinery. “Computational Analysis of Stall and Separation Control in Centrifugal Compressors. R. Jan. 1997.. 1. Vol. “Numerical Analysis of Three-Dimensional Viscous Internal Flow. J. Stein. 119.. Rivera. May 1990. Niazi.” NASA TM 83398.” AIAA Paper 98-0934. J. L. J. Vol. Jan. Niazi.. V. 68.” AIAA Paper 2000-0225. L. “Numerical Analysis of Stall and Surge in a High-Speed Centrifugal Compressor. J.” AIAA Paper 2000-0226.. 2000. Vol. N. Dissertation..-Feb. 1994. N.. and Garnier. W. Wood.. Adam. 1998.60.. pp. Hathaway. pp. and Buggele. 65.. Stein. “NASA Low Speed Centrifugal Compressor for Fundamental Research. 1983. E. “Numerical Simulation of Dynamic Stall Phenomena in Axial Flow Compressor Blade Rows. L.. J. A. C. Niazi. “Development and Application of a CFD Solver to the Simulation of Centrifugal Compressors. L. pp. S. and Sankar. Escuret. S. Jan. No.. A. M. Chima.” Ph. Vol. F. pp. 62. No.. and Yokota. “ Numerical Study of Surge and Stall Alleviation in a High-Speed Transonic Fan Rotor. and Sankar.D. A. 5. “Numerical Simulations of Surge and RotatingStall in Multi-Stage Axial-Flow Compressors.. Jan.. and Wood.. 61. N. V..” AIAA Paper 94-3202. 31-38.. S. R.” Journal of Propulsion and Power. W.

pp. and Somerville.” AIAA Paper 89-0201..70. P. “Some Contributions to the Modeling of Discontinuous Flows. “Approximate Reimann Solvers. “Characteristic-Based Schemes for the Euler Equations. 74. and Pletcher.. 337-365. Computational Fluid Mechanics and Heat Transfer. L. L. May 2000. 6. Vol.. 22. Vol.” Lectures in Applied Mathematics. 1986. 163-193. 71..” Journal of Computational Physics. pp..D. J. Lectures in Applied Mathematics.. Georgia Institute of Technology. C. Dissertation. Anderson. Computational Analysis of Stall and Separation Control in Centrifugal Compressors. Part. and Vinokur. Roe. L. 1985. Vol. Elsevier Science Publishers. 1984. Aerospace Engineering. 79. Roe. “Upwind Algorithms for General Thermo-Chemical Nonequilibrium Flows.” Annual Review of Fluid Mechanics. New York. Y. Parameter Vectors and Difference Schemes. INRIA. Hirsch.” Large-Scale Computations in Fluid Mechanics. L. 1989. E. “Discrete Models for the Numerical Analysis of Time-Dependent Multidimensional Gas Dynamics. Vol. Osher S. 163-193. 250-258. 1984.” NASA Contract Report CR-172574. Stein. Tannehill. 1985. Ph... J. B. 75. Roe. 18. 2.. 1985. Wiley. Roe. Edited by Engquist. D.. 138 . A. I and II. P. RI. 76. J. 2nd ed. P. 22. 499-518. P.” Computing Methods in Applied Sciences and Engineering. L. McGraw-Hill. Jan. C. Providence. 1st ed... A. ASME. Roe. 135. R. R. Liu. P. Numerical Computation of Internal and External Flows. and Pike. pp.. C. “Efficient Construction and Utilization of Approximate Riemann Solutions. 72. pp.. Vol. 78. Roe. Vol. P.. pp. L. 77. 1988. 73. “Some Contributions to the Modeling of Discontinuous Flows. Mar. M. New York. H.. 1981.

“Calculation of Tip Clearance Effects in a Transonic Compressor Rotor. “Design and Overall Performance of Four Highly Loaded. R. V. “Performance of Single-Stage Axial-Flow Transonic Compressor with Rotor and Stator Aspect Ratios of 1. D. L. 87. “Implicit Finite-Difference Simulations of ThreeDimensional Compressible Flow. 115. pp. 1980. J. A. L.” Transactions of the ASME. Wood. Celestina..” Transactions of the ASME. Vol. M. R. and Steger.26 Respectively.. R. 159-167. E. “Laser Anemometer Measurements in a Transonic Axial Flow Fan Rotor.. 436-449. Pierzga.. 1980.” NASA TP-2879. J.” NASA Paper TM-1337. 1993. S. April 1985. and Allmaras. J. Vol. Reid. “Viscous Analysis of Three-Dimensional Rotor Flow Using A Multigrid Method. P. and Wood.” AIAA Paper 92-0439. M. Sep. Nov.05. P. 107. Moore. J.. “Investigation of the Three-Dimensional Flow Field Within a Transonic Fan Rotor: Experiment and Analysis. Pulliam.80. January 1993. 84.” NASA Paper TM-1659. 1989.19 and 1. Spalart. 139 .. 86. Arnone. J. and Grietzer. A. Adamczyk. R. J. Suder.. and Moore. R.” NASA TM 107216. L. J..” AIAA Journal..” ASME 93-GT-19.. May 1996.. “A One-Equation Turbulence Model for Aerodynamic Flows. H. 81.. L. D. 1978. D.. pp. R. High-Speed Inlet Stages for an Advanced High-Pressure-Ratio Core Compressor. 85. Strazisar. “The Role of Tip Clearance in High-Speed Fan Stall. L. Feb. 82. K. R. and with Design Pressure Ratio of 2. 1992. Chima. 18... M. and Reid. 88. M. Hathaway. Vol. 83.

K. E.. C. “Viscous Three-Dimensional Calculations of Transonic Fan Performance.” ASME 96-GT-198. 1991... J. 1996. 21. 115. “A Numerical Investigation of Transonic Axial Compressor Rotor Flow Using a Low Reynolds Number k-ε Turbulence Model.” Proceedings of the American Control Conference. B. “Model Development of Active Surge Control Rotating Stall Avoidance in Aircraft Gas Turbine Engines. pp. Tan.. June 1994. June 1995. K. and Kikuchi. and Nett. M. P. Vol. T. and Wake Development in a Transonic Compressor. T. 93. J. Shirotori. 97. A.” ASME Paper 95-GT-326. 1991. R. Bright.. Shabir.. Crook. 95. July 1993.” Transactions of the ASME. H. pp. M. V. Eveker. S. 140 . and Celestina. 1996. Greitzer. A. M. L.. J. 1997. Suder. Vol.” Transactions of the ASME. H. 114. “Dimension Determination of Precursive Stall Events in a Single Stage High Speed Compressor. and Celesttina. and Adamczyk J. K.. “Numerical Transonic Flow Field Predictions for NASA Compressor Rotor 37. C.” AGARD 77th Symposium on CFD Techniques for Propulsion Applications. 3166-3172.. N.. and Hartley. L. “ Numerical Simulation of Compressor Endwall and Casing Treatment Flow Phenomena. Arima. Paper No.” NASA TM 107268. Radial Transport. M. T.. July 1992. and Reid. 96. Zhu.” ASME Paper 94-GT-365. Chima. D. 91.. C. Hah.... A. “Experimental and Computational Investigation of the Tip Clearance Flow in a Transonic Axial Compressor. and Wiss. M. Dalbert. Tamura. K.. Sonoda.. Qammar.. “Assessment of Three Turbulence Models in a Compressor Rotor. M. 90.. “A Viscous Flow Study of Shock-Boundary Layer Interaction. L. 92..89.. 94. T.” Journal of Turbomachinery. 501-511.. Paper 97-GT-82..

Prasad. Neumeier. Y. 1998. Wang. R. “Effects of the Shapes of Compressor Characteristics on Actuator Requirements for Rotating Stall Control. M. and Krichene.” Proceedings of the 1998 American Control Conference.” Proceedings of the NATO Applied Vehicle Technology Conference. Braunschweig. and Murray. Y. J.. A.. V. “Active Control of Compressor Surge Using a Real Time Observer.. May 2000.. 141 .98. 99.

He commenced his graduate work at Georgia Institute of Technology in September 1996. Iran. he taught at the Hormozgan University in the city of Bandar Abbas.Sc. Iran on May 17. Iran. with a Bachelor of Science (B. Persian Gulf. he graduated form the Shiraz University in Iran with a Master of Science (M.) degree in Mechanical Engineering in September 1986.) in Aerospace Engineering in July 2000. 1961. He graduated from the Shiraz University. From September 1989 to August 1995. He is a member of the American Institute of Aeronautics and Astronautics. 142 .Sc. In August 1991. He earned the degree of Doctor of Philosophy (Ph.9 VITA Saeid Niazi was born in Sepidan.D.) degree in Fluid Mechanic and Heat Transfer Engineering. He was the Dean of the central library of the Hormozgan University during the period 1992-1995.

Sign up to vote on this title
UsefulNot useful