DESIGN METHODOLOGY FOR AERODYNAMIC DESIGN

OF CENTRIFUGAL COMPRESSOR

A Thesis Submitted in partial fulfillment of the requirements of the degree

Of

B.TECH MECHANICAL ENGINEERING

by

ANAND VIJAYKUMAR
03BME018

SCHOOL OF MECHANICAL AND BUILDING SCIENCES

VIT
UNIVERSITY
(Estd. u/s 3 of UGC Act 1956)

Vellore-632014, Tamil Nadu, India

www.vit.ac.in

APRIL, 2007

Dedicated to

Shraman Goswami

ACKNOWLEDGEMENT

I would like to thank my Institute, Vellore Institute of Technology for providing me with
the opportunity for carrying out my project in the area of Turbo machinery and providing
me with the necessary guidance and infrastructure through which I could make this
project a grand success. I would also like to thank- Honeywell Technology Solutions Lab
for giving me this unique opportunity for carrying out my academic project in their
esteemed organization and providing me with all the guidance and support which are
needed at this level. I would also like to thanks my mentor at VIT, MR. D.R.S
Raghuraman who, with his immense experience and technical know-how provided me
with valuable inputs and guidance which made my project a truly valuable learning
experience. I would also thank my guide at Honeywell, MR. Shraman Goswami who in
spite of his busy work schedule took time off to guide me and played a very inspirational
role and help me understand the nuances of the subject with his highly technical expertise.

CERTIFICATE
This is to certify that the thesis titled Design Methodology for Aerodynamic Design
of Centrifugal Compressors is submitted by Mr. Anand Vijaykumar, 03BME018 to
the School of Mechanical and Building Sciences of VIT University, Vellore for the
award of the degree in B.Tech Mechanical Engineering is a bonafide record of work
carried out by him under supervision. The content of this thesis, in full or in parts
have not been submitted to any other institute or University for the award of any
degree or diploma.

Guide

Dean

Internal Examiner

External Examiner

ABSTRACT
The project presents a one dimensional (mean-line) design methodology for a
medium pressure ratio Centrifugal Compressor. A computational procedure for
design of a Centrifugal Compressor is established. The numerical model is based on
the conservation principles of mass, momentum and energy conservation and has
been utilized to predict the operational and aerodynamic characteristics of a small
centrifugal compressor as well as determining the performance and geometry of
compressor blades, both straight and backswept. The design code provides a basis
on which the design of the Compressor can be modeled by varying the key
parameters which include both aerodynamic and geometric details. The code shall
then predict and give a first cut solution, which will further help in zeroing on to a
particular design for the given requirements. The design which models the flow in
an Impeller, Diffuser and an annular bend takes into consideration various loss
models occurring in the complex flow of a Centrifugal Compressor. It makes use of
a Jet-Wake Model i.e. it splits the flow into Primary and Secondary zone and
performs a Mixed flow Analysis. The Design also performs an inverse flow analysis
where in we get the geometry by specifying aerodynamic details. The design has
been exhaustively validated by using a number of Test Compressors from NASA
test reports. Based on the geometry computed by the code, an impeller model is
generated using ANSYS BLADEGEN-Tool.

CONTENTS
Page No
Acknowledgement.4
Abstract..5
List of Tables.7
List of Figures/ Exhibits/ Charts8
Nomenclature.9
1

INTRODUCTION13
1.1
1.2
1.3
1.4
1.5

General Introduction.13
Background Information...14
Introduction to Project..14
Introduction to Design Process.17
Literature Review..20
DESIGN PROCEDURE

2.1 Design Process Flow Chart for Primary Calculations..22

2.2 Preliminary Design...24
2.3 Design of Diffuser and Bend29
2.4 Methodology.31
2.5 Method of Loss Analysis..32
3

PROCEDURE FOR PRIMARY FLOW CALCULATION37

3.1 Impeller Design Validation..37
3.2 Vaneless Diffuser Design Validation...41
3.3 Vaned Diffuser Design Validation...44
3.4 Bend Diffuser Design Validation......49

SECONDARY FLOW ANALYSIS.52

4.1 Secondary flow regime52
4.2 Recirculation and Disk Friction loss53
4.3 Impeller Boundary Layer Control54
4.4 Slip Factor55
4.5 Impeller Secondary Flow Calculations56
4.6 Mixing Calculations.........58

DESIGN VALIDATION
5.1 Compressor Design Validation- Test Compressor-1.......60

INVERSE DESIGN
6.1 Approach...63
6.2 Methodology 64
6.3 Other Modes of Operations...67

ANSYS BladeGen..69
7.1 Introduction to BladeGen...70
7.2 Using BladeGen.....70

Performance Variation with Geometry...74

8.1 Impeller Loss variations with Geometry....74
8.2 Impeller Exit Pressure Variations..75

CONCLUSION..76
9.1 Conclusion..76
9.2 Towards Better Design...77
9.3 Scope for Future work....78

APPENDIX.80
A.181
A.281
A.382
A.483
A.584
A.685
A.786
A.887
A.988

BIBILOGRAPHY/REFERENCES.90.

LIST OF TABLES

Table No

Title

Page No

2.1

Calculation of Impeller Exit Condition Using Spread Sheet Program26

3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8
3.9
3.10
3.11
3.12
3.13

Input for Impeller Design..37

Assumptions Made In Impeller Design.37
Calculation Performed in Impeller Design40
Inputs for Vaneless Diffuser Design..41
Assumptions Made in Vaneless Diffuser Design...41
Calculations Performed for Vaneless Diffuser Design..43
Inputs for Vaned Diffuser Design..44
Assumptions Made in Vaned Diffuser Design...45
Calculations Performed for Vaned Diffuser Design..46.
Best Combination of Blades, Width, and
Flow Angle for Maximum Pressure Rise in a Vaned Diffuser..47
Inputs for Bend Diffuser Design48
Assumptions Made in Bend Diffuser Design49
Calculations Made in Bend Diffuser Design.49

4.1
4.2

Secondary Flow Calculations.51

Mixing Flow Calculations..58

5.1
5.2
5.3

Comparison of CFD results with 1D code for Test Compressor-160

Comparison of 1D code Results with Test Results for Test Compressor-2...61
Comparison of 1D code Results with Test Results for Test Compressor-3...62

6.1

Variations in Impeller Exit Conditions as a Function of

Impeller Inlet Tip Mach number65
Various Losses Tabulated as function of Impeller Inlet Tip Radius...65
Variation of Inlet Tip Mach number with Impeller Inlet Tip Radius..66
Plot of Impeller Losses as a Function of Impeller Tip Radius71.

6.2
6.3
6.4

LIST OF FIGURES/EXHIBITS/CHARTS

Figure No.

Title

Page No

1.1
1.2
1.3

Activity Chart..................................................................................12
Meanline Diagram of the Centrifugal Compressor.......16
Meridional view of impeller..14

2.1
2.2

Impeller Design Flow Chart..24

Compressor Flow range Vs Mach number at Vane diffuser exit..36

3.1
3.2
3.3
3.4
3.5

Impeller Inlet Velocity Triangle........40

Impeller Exit Velocity Triangle.41
Vaneless Diffuser Exit Velocity Triangle..44
Vaned Diffuser Exit Velocity Triangle..46
Bend Exit Velocity Triangle..51

4.1

Mixed Flow Calculation Flow Chart.........56

6.1

Variation of Rotor Efficiency with Impeller Inlet Tip Mach number...65

8.1
8.2
8.3

Blade-Blade View, Meridional View and Meshed View..65

Meshed View of Impeller Blade........66
Shaded View of Impeller Blade71

7.1
7.2
7.3

Plot of Impeller Blade Exit Angle Vs Impeller Losses.........72

Plot of Blade Backsweep Angle Vs Stagnation Pressure at Impeller Exit73
From Aerodynamic Design of Blades to Blade Fabrication..74

8.1
8.2
8.3
8.4

Plot of Impeller Blade Exit Angle Vs Impeller Losses..74

Backsweep Vs Impeller Stagnation Pressure Ratio...74
Backsweep Vs Impeller Stagnation Pressure Rise.75

A.1
A.2
A.3
A.4
A.5
A.6
A.7
A.8
A.9

Impeller Input Text File...80

Vaneless Diffuser, Vaned Diffuser and Bend Input Text File.....81
Impeller Design Source Code..82
Impeller Source Code For design of Hub, Mean and Tip Conditions.83
Source Code for the Design of Impeller Losses..84
Vaneless Diffuser Design Source Code...85
Vaned Diffuser Design Source Code...86
Bend Design Source Code...87
.88

Nomenclature

2b
bt
c
Cd
ct
cm
Cpi
Dhyd
DF
i1, h
r1, t
r1, h
LC
mf
M
N
nrdb
Po,P
p

r
T
To
w
W
zb

absolute velocity angle

blade angle at exit of impeller
blade thickness
absolute velocity
coefficient of discharge
tangential velocity
meridional component of velocity
pressure loss coefficient (ideal)
hydraulic diameter
diffusion factor
incidence angle at hub
radius at tip
radius at hub
loss coefficient
mass flow rate
mach no
rpm
no of diffuser blades
stagnation pressure
static pressure
density
radius
static temp
stagnation temp
blade width
relative velocity
impellor blades

Suffix
t
h
m

tip
hub
meridional component

Constants
k
cp

gamma
specific heat
10

R
Re

universal gas constant

Reynolds Number

BLADE LOADING LOSS

SKIN FRICTION LOSS

DISK FRICTION LOSS

RECIRCULATION LOSS

INCIDENCE LOSS

No

station name

12345-

impeller inlet
impeller exit
vaneless diffuser exit
vaned diffuser exit
bend exit

11

Activity Chart and Schedule of Project Work.

Fig :1- Activity Chart

12

CHAPTER-1
INTRODUCTION
1.1 Introductory Remarks/Outline
GENERAL INTRODUCITION:
Over the past decade , there has been a growing interest and need for miniature energy
conversions systems, such as portable power generation for consumer electronics and
propulsion for micro fuel cells and handheld analytical instruments, as well as miniature
cooling units for electronics. Meeting this need requires the implementation of traditional
thermodynamic cycles at small scale, along with required machinery. Among the core
energy conversion technologies common large scales, compressors, pumps and turbines
are often used to convert fluid power to mechanical used for high power density and
reliability as illustrated by its prevalence in aerospace (aircraft gas turbines engines) and
power generation industries. But fabrication of such complex machines at millimeter
scale can however prove to be challenging and expensive.

The objective of current work consists of creating a knowledge base for design and micro
Turbo Machinery with special emphasis on Centrifugal Compressors Compressors.
This project focuses on the design process of single stage micros scale turbo machinery
along with CFD predictions of the key aerodynamic performance parameters required in
the design process. This work focuses on unique and unexplored design space defined by
the small scale and planar geometries characteristics of turbo pumps, compressors, gas
turbines and steam turbines, or other turbo machinery-based Microsystems.
Correlations are proposed for the loss coefficient, based on laminar- turbulent flow theory.
A critical Reynolds no is also identified, below which adjacent boundary layers merge,
inducing sharp increase in loss and deviation. This imposes practical limits on the
miniaturization of such microturbomachinery.

13

1.2 Background Information

Honeywell Technology Solutions Lab has taken a personal initiative in developing a
100N thrust Micro jet (Gas Turbine Engine). The Compressor design could be applied for
this Gas Turbine Engine which could eventually be used in a UAV (Un-Manned Aerial
Vehicle) or a Land based Generator.

1.3 INTRODUCTION TO PROJECT

Problem Definition:
To develop a design methodology for the Aerodynamic design of a medium pressure
ratio Centrifugal Compressor.

Scope of the Project:

Aerodynamic Design of a Centrifugal Compressor.
Secondary Flow Analysis and Mixed Flow calculations.
Validation of Results with Test Compressor taken from NASA reports.
Performing Inverse Mean-line design.
Creating an impeller Blade Model using ANSYS BladeGen-tool.

The objective of the project is to design and eventually develop a Centrifugal compressor
of medium to high pressure ratio for a Micro Jet Engine (Small Gas Turbine Engine).
The project deals with the design methodology for the design of a Centrifugal
Compressor. The 1D design gives an initial design solution on the basis of which it can
be decided if a complete CFD analysis of the compressor is required.

14

Technology Road Map

Fabricate an in house
Centrifugal Compressor
1

Develop a methodology for aerodynamic design of

Centrifugal compressor
2

Design and Study of Centrifugal Compressors

3

Fig : 1.1- Technology Road Map

The Project deals with the design, development and testing of a Micro jet engine which is
a Small Gas Turbine Engine. The possible application of the Micro jet engine could be
for a mini Unmanned Aerial Vehicle (UAV). It could also be used as a land-based
generator with suitable modification included.

15

The Micro jet is designed for a thrust of 100N. The engine consists of a centrifugal
compressor, reverse flow annular combustor; radial turbine and exhaust system.

Fig: 1.2- Meanline Diagram of the Centrifugal Compressor

The following components of the Compressor are designed using the 1D

code.
-

backward swept impeller

vaneless space
vaned diffuser
90 deg angular bend

16

1.4 INTRODUCTION TO DESIGN PROCESS

The design of modern centrifugal compressors involves several different disciplines,
considerable historical design experience, and usually a variety of design tools.
The classical turbo-machinery design process begins with mean line performance
modeling calculations, once a cycle specification has been set freezes the design
flow, speed and stage pressure ratio or head rise.
When basic mean-line velocity triangles have been suitable optimized then
blading methods are used to design the required blade shapes. They may involve,
either direct or inverse computational methods or they should involve both flow
solvers and fundamental design rules.
When appropriate passage contours and blade shapes, are obtained it is reasonable
to go for final levels of design optimization.
At final level, CFD is used to study the full three-dimensional (3D) flow field
within the impeller, and the designer can then focus on the fine detailed design
considerations like localized regions of backflow, unwanted secondary flows, so
on and so forth.

1.4.1 Mean Line Flow Modeling:

One Dimensional or Mean line flow modeling is the simplest and the quickest way of
looking at the flow state.
When considering mean line we are paying attention to the average to characteristics of
the flow at that location which, on a mass or energy average basis gives the correct
representation of the entire flow field. However for the impeller exit we see that this is a
considerable simplification which cannot be justified under all circumstances. Hence the
concept of Two- Zone modeling comes into the picture.
The mean line flow modeling is illustrated for the stage with the overall stage
performance data only. Some assumptions about the slip-factor and diffuser pressure
recovery coefficient are introduced. Subsequently other experimental results are
progressively introduced and using the data, these assumptions can be checked or
eliminated, thereby improving the basic consistency model considerably.

17

The modeling is for a compressor which features an impeller with radial blades or
backswept vanes at exit, a vaneless or vaned diffuser and a 90 degree annular bend.

1.4.2 Importance of Meanline Design:

Fundamental importance of meanline design is to optimize the basic velocity

triangle at each distinct, basic station throughout a turbo machine flow path. The
basic stations would include, for example, an impeller inlet, an impeller exit,
diffuser inlet, diffuser exit, and so forth. It makes no sense to try to use higher
order codes to optimize the average velocity triangle at these stations. This is the
important role of a meanline performance code.
If proper meanline optimization is not done, no amount of subsequent
optimization will correct for the inappropriate starting parameters. These starting
parameters involve the passage width, (b), and the mean radius,(r), where the
passage height is centered at each meanline station. Likewise, the mean blade
angle must be chosen according to the expected flow direction and desired
incidence. Preferred exit angles must be set. The correct mass flow rate and work
input (U2C2) level is nearly assured by these steps.
Basic stall criteria usually are assessed on a meanline level and should be
evaluated to assure adequate stable operating range or the compressor. In short,
most of the truly critical design criteria that must be met for a product to be
acceptable in the marketplace have a very important meanline design requirement.

18

To meet these requirements, good meanline modeling is essential in order to start a

design optimization process with any real hope of a cost-effective, market-competitive,
product. Additionally, most diffusers and return channels are 2D in form so most of the
design process is complete at the meanline level. After the meanline optimization work is
completed, blades are designed to assume the intended deviation and, in turn, the work
input required and to turn the flow through desired angles either before or after the
impeller.

Fig: 1.3 Meridional view of impeller.

19

1.5 Literature Review

1.) FORTRAN Program for Predicting Off- Design Performance of Centrifugal
Compressors
Provided the off design Characteristics of Centrifugal Compressor with a
Example of a Compressor built by a commercial engine manufacturer to
demonstrate the effectiveness of the correlation with limited experimental Data.
Individual Losses are Computed Using Analytical Correlations which relates loss
Levels to Velocity Diagram Characteristics and Overall Geometry.
It Provides Valuable Information about Efficiency decrements through Use of
Plots and Tabulates Performance details for Validation.
2.) A Method of Performance Prediction of Centrifugal Compressors
Provides a huge database of Experimental data and by means of Extensive
Utilization of Correlations tests Results with predicted data.
Provides correlations for finding variations in Gamma, Specific Heat and
Viscosity with variations in temperature and pressure
By means of a extensive availability of Measured data provides Tabulated Values
for finding out Loss Coefficients in Vaned Diffuser by suitably interpolating the
data.
Provides important correlations for finding Area Ratio (AR) in Vaned Diffuser.
Also provide correlations for Loss Modal like Recirculation Loss and Leakage
Losses occurring in Impeller Section.

20

CHAPTER-2
2.0 DESIGN PROCEDURE
WORK PLAN FLOW CHART
Start

OD design or
Thermodynamic Design

Initial iterations to find

out thermodynamic
properties like
temperature (To, T1),
Pressure (Po, P1) and
mass flow rate m (kg/s)

Using CFD analysis or

using Exiting Papers to
confirm Results.

1D- design or Mean Line Design

Are results
validated correctly?
No
Yes

Tools used for generating

the blade profile
BladeGen

3D design of impeller or the

geometric design of blade profile
using tools.

Inverse Mean Line Design/Optimization

Project Report Submission

Stop

21

2.1 Design Process Flow chart for Primary Flow Calculations

Start

Impeller Design

Aerodynamic Inputs
Po1, To1, N, m, blade no,
Geometrical Inputs
Rt, Rh, R2, Clearance,
Incidence angle, backsweep

Results and calculations

P1, T1, Po2, To2, P2, T2,
M1, M2, 1 (deg)

OUTLET CALCULATIONS:
U2= (2*pi ()*r2*N)/60
ct2=U2+Cm2m*tan2b
=U2 since 2b=0 (radial outlet
=1-(cos 2b) 0.5 /z 0.7 using Wiesners Correlation
ho =U2*ct2
T02m=T01+(k-1)/(kR)* ho
Po2/Po1= (1+ (nrotor *ho)/(Cp*T01))^k/ (k-1)
cm2m and M2m are calculated as before.
W2m= (U2-ct2m) ^2+cm2m
2m=-cos-1(Cm2m/W2m)
2m= tan -1 (Ct2m/Cm2m)

Calculation of rotor
efficiency

Vaneless Diffuser Design

Aerodynamic Inputs
mf3=mf2=mf1
Geometrical Inputs
d3=d2*1.08 m
Assumptions
DF1=1.1, Cd3, LC1, C3=C2/DF1

Vaneless diffuser calculations:

ct3=d2*Ct2/d3
c3=c2/DF2 (initial)
h3=ho3-c32
a3=sqrt (k*R*T3)
M3= c3/a3
P3 = Po3/pow((1+ (k3-1.)/2*M3*M3), (k3/ (k3-1)))
3 (kg/m3) =P3/(R*T3)
cm3=mf3/ (Af* 3*)
c3new=sqrt (ct3*ct3+cm3+cm3)

Output from Code

Results and Calculations of
Po3, To3, P3, T3, M3
Iteration on C3 till error in C3<0.001

No
Error in C3<0.001

22

A
LC2 is determined from table-1
Reference-1
LC2=Cpi-Cp
Cpi=1-1/AR2

Vaned Diffuser Design

Vaned Diffuser Calculations:
4 = 39.5
Tan (4) = Ct4/Cm4
c4=c3/DF2 initial
h4=ho4-c4*c4
a4=sqrt (K4*R*T4)
M4=c4/a4
P4 = Po4/pow ((1+ (k4-1.)/2*M4*M4), (k4/ (k4-1)))
4 (kg/m3) =P4/(R*T4)
cm4=mf3/ (AF* 4)
c4new=sqrt (ct4*ct4+cm4+cm4)

Aerodynamic inputs
mf4=mf3=mf2, alp4 (absolute)
Geometrical inputs
Nrdb, d4, bt4,
Assumptions
Cd4, LC2, DF2=2.2, C4=C3/DF2 (initial)

Output from code

Results and calculations of
Po4, To4, P4, T4, M4
Iteration on C4 till error in C4<0.001

No
Error in C4<0.001

Bend Design

Aerodynamic inputs
mf5=mf4=mf3=mf2, Ho5=H02,
Geometrical inputs
d5, b5, d5 (outlet), d5 (inlet)
Assumptions
LC3, Cd5, DF3=1.1,

Bend Calculations:
ct5=d4*ct4/d5
c5 = c4*DF3 (initial)
h5 = ho5 - 0.5*c5*c5
a5 =sqrt (k5*R*T5);
m5 = c5/a5;
P5= Po5/pow ((1+ (k5-1.)/2*M5*M5), (k5/ (k5-1)))
5 = P5/R/t5;
cm5 = mf5/( 5*Af5);
c5new=sqrt (ct5*ct5+cm5*cm5)

Output from code

Results and calculations of
Po5, To5, P5, T5, M5
Iteration on C5 till error in C5<0.001

No

Error in C5<0.001

END
23

2.2 PRELIMINARY DESIGN:

This report consists of evaluation of impeller and diffuser configuration that exhibits the
potential for high efficiency and adequate operating range. These configurations will be
selected for detailed aerodynamic design and fabrication.
The impeller configuration considered in this section includes radial and backward
curved impeller with conventional blading.
The diffuser configuration includes a vaneless diffuser, single-row vaned diffuser
(conventional vaned diffuser). Also included will be initial screening of possible
boundary layer control methods for adaptation to these configurations.
2.2.1 Conventional impeller
A particular operating speed is selected for conventional impeller. The tip Mach number
was limited to subsonic range to avoid shock losses.
A loss optimization study with a fixed speed and hub diameter was conducted and
included consideration of the following losses:
Impeller friction and turbulence
Impeller dump (mixing)
Impeller disk friction
Impeller axial clearance

Impeller efficiency is given by:

rotor= h Aero- h BL- h SF- h RC- h IN-h mixing

h Aero

24

2.2.2 Impeller Design Flow Chart

Fig :2.1-Impeller Design Flow Chart

The impeller exit back angle (backward sweep) was fixed or could be made variable.
Design of centrifugal section was a cut and try process where in the meridional shape
and the blade angles were varied to produce a desirable blade loadings. Initial
estimates of blade angle distribution, meridional shape, blockage, and loss
distribution were based on a previous 4:1 pressure ratio conventional impeller design.

25

2.2.3 Preliminary Design Performed using Spread Sheet Program.

Table No: 2.1- gives the Impeller Exit conditions by Performing 3 design iterations.

26

2.2.4 Boundary Layer Formation:

In any flow through a passage, a layer of low velocity, so called a boundary layer,
grows along the wall due to wall friction, where velocity varies from zero on the wall
to the main flow velocity.
In flow through a diverging passage, the pressure gradient balances with a moderate
change of velocity in the main flow, but the pressure gradient is too steep for the slow
flow in the boundary layer to keep the profile of velocity distribution. As a result the
profile of velocity in the boundary layer is distorted further and the layer gets thick
absorbing a part of the main flow, in worst case a reverse flow occurs near the wall,
and the pressure hardly increases downstream.
Rate of growth of the boundary layer and distortion of the velocity profile depends on
the pressure gradient, and the rate of growth is also proportional to the initial
thickness of the boundary layer. In cases of flow through curved ducts with
rectangular cross section, there is pressure gradient perpendicular to the curved wall,
and in the boundary layer on the flat side-walls cross-flow is induced toward the
lower pressure zone and slow flow accumulates.

2.2.5 Analysis of Impeller Boundary Layer:

Analysis of impeller boundary layer was done with use of calculations. This program
computed the characteristic parameters of two dimensional boundary layer including
compressibility effects and heat transfer to the wall.
The following sections discuss the preliminary and detailed design of the compressor
together with the methods of calculation and assumptions used in the compressor
design. Design point velocity diagrams and state conditions are given for various
stations throughout the stage. The velocity diagrams are deduced from the effective
flow areas dictated by assumed loss at various stations. Blade and Vane geometry and
surface velocity distribution are given for impeller channel diffuser, and bend.

27

2.2.6 Impeller Flow Physics:

The flow from the impeller is steady with respect to the rotating system, and a jet
zone and a slow flow zone exist side by side in a pitch of vanes. If the flows, flow out
independently keeping the respective angular momentum, the backward leaning angle
of the slow flow is larger than that of the jet flow, because the radial velocity
component of the slow flow is smaller than that of the jet flow while their
circumferential velocity components are nearly equal. The slow flow or wake flow
and the jet flow must share a common border in the rotating system. Therefore, the jet
flow pushes the slow flow perpendicular to the border, i.e. outward and forward. As a
result the slow flow components are increased. Losing its width it gradually unifies
with jet flow to finally form a uniform flow at impeller exit.
The wake flow with increased circumferential/tangential velocity highly affected a
larger friction force exerted by walls of the stationery diffuser; the pressure loss
becomes larger unless the diffuser is very wide.

28

2.3 DESIGN OF DIFFUSER AND BEND

The initial impeller meridional contour was constructed using mathematical curves
for the hub and shroud. An Initial blade shape was constructed by matching blade
angles at the inlet and exit to those required by part-power incidence and blade back
sweep with numerically smoothed curves between line endpoints. Near zero
incidence from free stream to blade mean camber line along the inducer span was
specified at design point to obtain more favorable incidence and consequently,
additional surge margin at part power operation.
A meridional plane flow solution was obtained using computer program. Iterations on
meridional shape, and blade angle distributions along the blade surface were obtained.
In the meridional analysis, the primary objectives were to obtain near-uniform
loading along the shroud meridional length with most of the diffusion occurring
within the inducer. High initial diffusion was considered acceptable because the
boundary layer is most able to withstand the adverse pressure gradient without
separating. Light diffusion along the shroud downstream of the inducer was used in
an attempt to minimize the potential for flow separation, which is most likely to occur
first at shroud section. Velocity peaks and shroud suction surface diffusion ware also
minimized to promote flow stability.
Detailed blade geometry and flow path information output from the final meridional
plane solution were used as input for a blade to blade solution as described. The final
blade angle distribution for the hub, mean and tip sections are shown and blade
surface velocity distributions from the blade to blade solutions for these three sections
are shown.
2.3.1 Vaneless Diffuser
The Vaneless Diffuser is frequently employed in process compressors, and
turbocharger compressors. A reasonable level of static pressure recovery can be
achieved with a properly designed vaneless diffuser, which is also inherently
inexpensive. It comprises essentially of two parallel walls forming an open radial
annular passage from the impeller tip to some limiting outer radius. The most
common type of vaneless diffuser is the pinched or reduced area, vaneless diffuser

29

which provides a necessary stabilization of the flow. The flow through the Vaneless
diffuser follows the conservation of mass and momentum equations
2.3.2 Channel Diffuser
The primary considerations in the detailed design of the channel diffuser were
incidence for adequate surge margin and throat area required for the design mass flow.
The wall spacing was equal to the impeller exit blade height plus the axial clearance
at impeller exit. At low flow rates the stream tubes diverge providing increased
available flow area, and diffusing flow field results. At high flow rates however due
to acceleration in the inlet region is sufficient to overcome most of diffusion in the
subsequent passage portion, and thus is hard choke and diffuser works like a nozzle.
The static pressure recovery in the Vaned diffuser section is a function of the diffuser
AR (Area Ratio) and throat blockage which measured by the boundary layer
thickness.
2.3.3 Turning Duct
The turning duct geometry is completely specified by two circular surfaces of
revolution, the inner wall radius and outer wall radius. The inlet annulus is specified
at a radial distance from the axis of rotation. Inlet static conditions are those which
correspond to exit of channel diffuser. A particular degree of swirl still remains in the
flow just inside the channel diffuser exit. Rapid expansion, assumed to occur at
constant total pressure, around the channel diffuser trailing edges result in a absolute
flow angle measured from radial direction.
Angular momentum remains relatively constant throughout the turning duct although
the slight increase in the duct streamline radius. The flow solution for the turning duct
was computed using the meridional plane analysis.

30

2.4 METHODOLOGY:
A C-language program for calculating the off-design performance of centrifugal
compressors with vaneless, channel diffuser and bend is presented. Use of the program
requires complete knowledge of the overall impeller and diffuser geometries.
Individual losses are computed using analytical equations and empirical correlations
which relate loss levels to velocity diagram characteristics and overall geometry. At flow
rates between surge and choke, individual efficiency decrements, compressor overall
efficiency and compressor total pressure ratio are tabulated.
An example case of performance comparison with a compressor built by a commercial
engine manufacturer is presented to demonstrate the correlation with limited
experimental data.
2.4.1 off Design Characteristics
The off-design performance characteristics of centrifugal compressors are of interest
because of the large effects that compressor component performance has on overall
cycle performance and because the compressor is required to operate at off-design
conditions much of the time. These losses become more pronounced at part-power
settings. In addition to good performance at off-design flow rates it is important that
the compressor operate stably over the range of flows and speeds required by the
engine operating envelope.
The usable range of the compressor pressure ratio-mass flow characteristic is
bounded by the surge and choke mass flow rates. Operation at flows less than the
surge point flow should be avoided because of potentially dangerous vibrations
induced by the intermittent flow reversals and power loss. Operation with the
compressor choked is generally avoided because of the poor compressor
efficiency and pressure ratio at the choke point. The problem undertaken in this
analysis is to determine the centrifugal compressor performance characteristics
over a range of rotative speeds and flow rates and predict the usable range of flow
rates at which the compressor can operate.

31

The method of analysis uses the loss correlations and surge and choke criteria
added to predict compressor operating range.
The program predicts centrifugal compressor performance through utilization of
empirical correlations which are related to the compressor geometry and velocity diagram
characteristics.
A complete knowledge of the compressor overall geometry and working fluid inlet total
conditions is required for its use. Working fluid state conditions and flow properties are
calculated using a mean streamline one-dimensional analysis.
The program is limited to centrifugal compressors with channel diffusers operating up to
their choke point. A comparison of calculated and experimental performance is given to
demonstrate the correlation with limited experimental data.
The experimental data presented for comparison were obtained from a compressor
developed by NASA Research Centre refernce [].
Shock losses in the rotor are neglected. Clearance losses are considered to be inherent in
the impeller losses since good performance correlation is achieved on compressors
operating with reasonable clearances.

2.5 METHOD OF LOSS ANALYSIS

Individual losses are calculated using velocity diagram characteristics and empirical
correlations determined by the input absolute velocity level and compressor geometry.
Overall compressor efficiency, total pressure ratio, and mass flow rate are tabulated for
each operating point inside the predicted range for each speed line that is input.

32

2.5.1 CALCULATED LOSSES:

Blade loading loss - Boundary layer growth in the impeller is highly dependent on the
diffusion of the working fluid internal to the impeller itself. Equation for calculating the
diffusion factor of the impeller based on a uniform velocity loading along the blade chord.
This equation is used to calculate the impeller diffusion factor for impellers without
splitters. A modified form, with reduced penalty due to aerodynamic work input, is used
for impellers which have a set of splitter blades. With the diffusion factor calculated by
these methods the blade loading loss was expressed

Skin friction loss - In addition to the losses resulting from the aerodynamic loading
Of the impeller blades, the impeller incurs losses due to skin friction of the impeller
and shroud wetted areas. Developed an equation for this loss based on fully
Developed turbulent pipe flow. In the case of impellers with splitter blade rows, the
Empirical constant appearing in the equation is modified to account for the higher mean
Channel relative velocity caused by the addition of splitters. The general equation used
For skin friction loss is

Where KSF = 5.6 for conventional impellers and KSF = 7.0 for impellers with tandem
Blades.

33

Disk friction loss - The specific loss due to windage on the compressor back face
is calculated using the equation

Recirculation loss - Losses resulting from work done on the working fluid due to
Backflow into the impeller are expressed as

Vaneless diffuser loss

Vaned diffuser loss - Curves of maximum static pressure recovery coefficient at

Given area ratio were extrapolated from the test data for square throat diffusers reported
For various combinations of vaned diffuser throat Mach number, aerodynamic blockage,
and area ratio. These data were recorded for channel diffusers with symmetrical pressure
Loadings about the channel centerline. The vaned diffuser in a Centrifugal compressor is
loaded with a pressure gradient across the channel. In this Analysis it was assumed that
the difference in the loadings between the test diffusers of and the compressor diffusers
would have no effect on calculated diffuser recovery. A one-seventh power velocity
distribution in the boundary layer along the vaneless space endwalls is used to calculate
the displacement thickness representing vaned diffuser throat blockage. The value of
static pressure recovery coefficient Cp corresponding to the vaned diffuser geometric
area ratio, inlet Mach number, and aerodynamic blockage is extrapolated from the test
data. The vaned diffuser exit critical velocity ratio is calculated using one-dimensional
continuity. Vaned diffuser loss is then calculated using the equation

34

Calculation of Compressor choking Flow Two criteria are used in the prediction of
compressor choking flow:
(1) Inducer choke.
(2) Vaned diffuser choke.
Vaned diffuser choke is predicted from one-dimensional continuity using the computed
Values of weight flow, total temperature, total pressure, aerodynamic blockage, and the
geometric throat area. The maximum value of the one-dimensional weight flow function
is calculated from the equation

Calculation of Surge:
The vaned diffuser is assumed to be the component which governs the location of the
compressor surge point. Based on a database of a number of compressors covering a
range of pressure ratios indicates that the compressor flow range can be expressed as a
function of vaned diffuser leading edge mach number.

35

Fig: 2.2-Compressor Flow range Vs Mach number at Vane diffuser exit

2.5.2 CONCLUDING REMARKS
This section summarizes the pertinent characteristics of the prediction method, discusses
the limitations of the method, and notes the effect of the limitations on the results. For
inducers with relative Mach numbers appreciably in excess of unity the predicted
compressor total pressure ratio and efficiency are greater than that attainable
experimentally. This is a result of neglecting shock losses in the inducer inlet. Predicted
static pressure recovery for vaned diffusers with throat Mach numbers greater than unity
will be too large. Extrapolation of the channel diffuser data used in the program is valid
only in the subsonic flow regime. Compressor total pressure ratio and efficiency are not
adequately predicted with the one-dimensional correlations when the compressor mass
flow rate is limited by the impeller rather than the vaned diffuser. The predicted total
pressure ratio and efficiency levels are higher than those measured experimentally.
Predicted surge point efficiencies are considerably lower than those measured on the
example compressor. This may be due in part to the difficulty encountered in making
precise measurements at this flow condition.

Overall Compressor Efficiency:

h aero- (h incidence + h blade loading + h skinfriction Loss

h vanelessdiffuser loss+ h vaned diffuser loss+ h bend loss)

h aero + h recirculation + h diskfriction loss
36

CHAPTER-3

Step By Step Procedure for Primary flow calculation

3.1 Impeller Design ValidationThe calculations are performed 4:1 Rc, 404-III single stage Centrifugal Compressor for
use in regenerative gas turbine applications. Its is an Allison Engine Company Design
and details are taken from Reference [1]
Inputs:
Name
Total Pressure
Total Temperature
N
mass flow rate
(mf1)
incidence at hub,
tip
Radius at tip
Radius at hub
Radius
Blade width (w2)
Blade Number
exit clearance

Location
impeller
inlet
impeller
inlet
impeller
impeller
inlet

Value

impeller
impeller
inlet
impeller
inlet
impeller exit
impeller exit
impeller
impeller exit

50

101.3 kpa
288.166 K
21789 rpm
4.391

0.105 m
0.041 m
0.216 m
0.017 m
15 blades , 15 splitters
0.00023 m

Table No: 3.1-Inputs to Impeller

Assumptions:
Name
Mach no (initial)
2b
rotor (initial)
Mach no (iterated)
rotor (iterated)
Mach no (initial)
Mach
no(interated)

Location
impeller
inlet
impeller exit
impeller
impeller
inlet
impeller
impeller exit

Value

impeller exit

0.919

0.3
500
0.86
0.392
0.918
0.6

Table No: 3.2- assumptions made in impeller design

37

Calculations:
Impellor inlet mean diameter

= D1

= 0.146

Impellor exit tip diameter

= D2

= 0.432

Impellor exit width

= W2

= 0.017

Rotor speed (design)

= N

= 21789

RPM

Overall Compressor Pressure ratio

= PR

= 4

Compressor efficiency

= 0.859

Number of impellor blades

= Zb

= 30

Blade thickness at the impellor at mean

= bt

= 0.001

Compressor inlet mass flow

= mf1

= 4.391

kg/s

Compressor inlet pressure (Total)

= Po1

= 101300

Pa

Compressor inlet temperature

= To1

= 288.166

Impellor exit blade angle

= 500

Compressor inlet enthalpy (determined)

= ho1

= 288577.2993 J/kg

Compressor exit enthalpy (determined)

= h02

= 467329.6959 J/kg

Discharge coefficient

= Cd

= 0.97

P02

b2

P01*(1+(nrotor*ho)/(Cp*T01))k/(k-1)

= 467.253

Pa

nrotor = 0.918 (from loss models)

Area at the impeller exit (Area2) =

*D2*W2 Zb*W2*bt2 / cos (

= 0.0228169

b2)

m2

Calculation of Impeller exit parameters (Station 2)

Ae2 = Area2 * Cd
= 0.0228169*0.97
= 0.022132393

m2

h = h02 h01
= 175894.318
U2 =

J/kg

*D2*N/60

= 2* *0.216 * 21789 / 60

38

= 492.280

m/s

h = U2 * ct2 U1 * ct1
h = U2 * ct2
ct2 = 356.888

m/s

Slip factor = SF = ct2 / ct2i

= 1 / (1 + (
(D1/D2))))

cos (

b2)

/ 2*Zb*(1-

= 0.926
ct2i = ct2 / SF
= 372.445 / 0.926
= 385.408

m/s

Slip = ct2i ct2

= 385.408 356.88
= 28.528

mf2 =

2*

m/s

Ae2 * Cm2

cm2 = mf2 /

2*

Ae2

= 4.391 / 2.37 *0.022

= 84.2155 m/s

m/s

c2 = ( cm22 + ct22)
= (84.21552 +356.882)
= 366.51

m/s

h2 = h02 - c22 /2
= 467329.6959-(366.51^2/2)
= 400164.9059
Static temperature at rotor exit (t2 ) = 395.148
CP2 = 1011.2866
2

J/kg K
K
J/kg K

= 1.391

a2 = Sonic speed = ( 2R t2)

= 397.17728

m/s

39

Absolute Mach No. M2 = c2 / a2

= 0.92
Static pressure at impellor exit (P2 ) = P02 / (1+ (( 2-1) M22 /2))( 2/

2 -1)

= 269154

Pa

= P2 / R t2
= 260968/ 287 *397.498
kg/m3

= 2.37334
2 = tan-1 (ct2 / cm2)
=
= 76.842
2

= Relative flow angle at impellor exit

= tan-1 (Vt2 / cm2)
= -58.466

( 2
is
negative
)

Table No: 3.3-Impeller Inlet Exit Calculations

Impeller Inlet Velocity Triangle

Fig:3.1-Inlet Velocity Triangle

1 (deg):
90
1 (deg):
38.421
1b (deg):
33.421

40

Impeller Exit velocity triangle

Fig: 3.2- Exit Velocity Triangle

2 (deg):
13.158
2 (deg):
31.545
2b (deg):
40
3.2 Vaneless Diffuser Design Validation:
The flow is assumed to be free vortex flow and based on this condition the tangential
component of velocity (ct3) is determined. An initial assumption of absolute velocity (c3)
is made based on the diffusion factor (DF1). Iterations are performed on the absolute
velocity (c3) and Vaneless diffuser exit conditions are determined.
Inputs:
Name
diameter
width
mass flow rate

Location
vaneless exit
vaneless exit
vaneless
entry

value
d2*1.08=0.46656 m
b2*0.98=0.01666 m
mf3=mf2=mf1

Table No: 3.4- Inputs to Vaneless Diffuser

Assumptions:
Name
diameter
width
mass flow rate

Location
vaneless exit
vaneless exit
vaneless
entry

value
d2*1.08=0.46656 m
b2*0.98=0.01666 m
mf3=mf2=mf1

Table No: 3.5- Assumptions made in Vaneless Diffuser Design

41

Calculations:
W3= width at the vane less diffuser = W2*0.98
exit
= 0.017

D3 = Vane less diffuser exit diameter

= 1.08*D2 (8% more than impellor
exit diameter)
= 0.4536

LC1=Loss coefficient = 0.07

P03 = vane less diffuser exit total pressure
= P02 LC1*( P02 - P2)
= 453385.84
Enthalpy at vane less diffuser exit =
(h03)

Pa

h02

= 467329.6959

J/kg

DF1=Diffusion factor = c2/c3

= 1.1 (first guess)
c3 = c2/DF1
= 379.700/1.1
= 348.990

m/s

h3 = h03-c32/2
= 467329.6959 348.9902/2
= 406432.6859
t3 Static temperature at vane less = 400.4606
diffuser exit for h3
CP3 = 1011.99
3

J/kg
K
J/kg K

= 1.395

a3 = Sonic speed = ( 3R t3)

= (1.39587*287*401.1
= 399.69

m/s

42

Absolute Mach No.M3 = c3 / a3

=
P3 =

0.873
P03 / (1+ (( 3-1) M32 /2))( 3/

3 -1)

= 276769
3

Pa

= P3 / R t3
= 2.4081

Area3 =

kg/m3

*D3*W3

= 0.0237

m2

ae3 = Area3* Cd
= 0.02377
From continuity equation,
mf3 =

*ae3*cm3

cm3 = 4.391*0.0237*2.4081
= 79.1624

m/s

ct3 = ct2*D2/D3
= 339.894

m/s

mf2 = mf3
ALPHA3 = tan-1 (Ct3/Cm3)
= 76.7703
Angle of radial diffuser blade = tan-1(cm3/ct3)
leading edge for zero incidence ( 3)
= 13.11

Table No: 3.6- Calculations made in Vaneless Diffuser Design

43

Vaneless Diffuser Pressure Loss

dp

f * cosec(a) * sec(a) * M * M * dr * P
g * h * R * T * r * r * (1 (U *U ) /(2 * g * cp * T )) * cos B

Vaneless Diffuse exit velocity triangle

Fig :3.3- Vaneless Diffuser Velocity Triangle

3 (deg):
13.2297
3.3 Vaned Diffuser Design Validation:
The absolute flow angle (4) is fixed from blade geometry. An initial guess of the
absolute velocity (c4) is made based on a value of diffusion factor (DF2). Iterations are
performed on the absolute velocity and vaned diffuser exit conditions are determined.
The loss coefficient is as function of area ratio, length width ratio (LWR) and blockage.
Inputs:
Name
no of diffuser blades
(nrdb)
blade thickness (bt4)
diameter (d4)

Location

Value

vaned diffuser
vaned diffuser
vaned diffuser exit

width (w4)

vaned diffuser exit

mass flow rate (mf4)

absolute angle ( 4)
Mach no (m4n)

vaned diffuser
vaned diffuse exit
vaned diffuser exit

24
0.038 m
0.72644 m
w3=w4 (no
pinching)=0.01666 m
mf4=mf3=mf2=const=4.391
kg/s
39.5 calculated from paper
0.193

Table No: 3.7- Inputs toVaned Diffuser

44

Assumptions:
Name
loss coefficient (LC2)initial
coefficient of discharge

Location

Value

vaned diffuser exit

vaned diffuser exit

0.2
0.97

Table No: 3.8- Assumptions Made in Vaned Diffuser Design

Calculations:
DF2=Diffusion factor = 2.265 initial guess
nrdb =number of radial diffuser = 24
blade
bt3=diffuser blade thickness at = 0.00
entry
bt4=diffuser blade thickness at exit = 0.038

mm
mm

W4=Diffuser width = .01666

mm

mf4=mass flow rate = 4.390

kg/s

ALPHA4=exit blade angle = exit flow

angle desired = 39.5
D4=diffuser exit diameter = 0.726
P04 = P03 LC2 (P03-P3)- (reference 1)
= 418062.4516

Pa

C4 = C3/DF2 (initial guess)

h04 = h03
= 467329.6959

J/kg

c4 =
= 82.1176 (iterated)

m/s

h4 = h04 - c42/2
= 467329.6959 82.11762/2
= 463958.0458
t4 Static temperature at exit for h4 = 453.537

J/kg
K

a4 = ( 4R t4)
=

45

= 453.53

m/s

M4(Absolute flow Mach number) = c4 / a4

= 0.193
P4 =

P04 / (1+ (( 4-1) M42 /2))( 4/

= 407393.91
4

4 -1)

Pa

= P4 / R t4
kg/m3

= 3.12
Area4 =

*D4*W4 nrbd*W4*bt4
m2

= 0.0228
ae4 = Area4* Cd

m2

= 0.02216
Cm4 = mf4 /

ae4

= 4.391 / 3.02 * 0.02216

= 63.4103

m/s

ct4 = (c42 Cm42)

= 52.176

m/s

Table No: 3.9- Vaned Diffuser Calculations

Vaned diffuser exit velocity triangle

Fig: 3.4-Vaned Diffuser Velocity triangle

4 (deg):
50.5

46

3.3.1 Best Combination of No of diffuser blades, blade width, flow angle to get
highest Pressure Ratio.
The table below gives the combination of diffuser blades (nrdb), blade width (w) and flow
angle (alpha) for the maximum rise in static pressure ratio in radial diffuser and the compressor
as a whole.

No of
blades
24
24
24
24
No of blades
18
20
22
24
No of blades
18
18
18
18
No of blades
18

blade width
(m)
0.038
0.038
0.038
0.038
blade width
(m)
0.038
0.038
0.038
0.038
blade width
(inch)
1
1.2
1.4
1.6
blade width
(inch)
1

Alpha
(alp4)
30
40
50
60
alpha
(alp4)
30
30
30
30
alpha
(alp4)
30
30
30
30
alpha
(alp4)
30

P4/P1 (static pressure ratio)

Vaned diffuser exit/impeller inlet
4.20478
4.17654
4.12072
3.99098

P5/P1 (static pressure ratio)

Compressor exit/impeller inlet
4.25476
4.20113
4.0959
3.8569

p4/p1
4.23132
4.22382
4.21508
4.20478

p5/p1
4.28715
4.278
4.26732
4.25476

p4/p1
4.24827
4.24225
4.23517
4.22677

p5/p1
4.30782
4.30051
4.29186
4.28159

p5/p1
4.30782

Best Combination

Table No: 3.10- Best Combination of Blades, Width, and Flow Angle for Maximum
Pressure Rise in a Vaned Diffuser

47

3.3.2 Calculation of Static Pressure recovery coefficient:

Table 1 is used to calculate Static pressure recovery coefficient (Cpr) Based on the Area
Ratio (AR) and Length/width (LWR) ratio which are obtained from geometric details.
Throat Blockage and other details known, Cpr can be found out.
For LWR and AR not matching with the values in column, method of linear
interpolation is used.
Loss Coefficient (LC) = Cpi (Ideal Pressure Recovery)-Cp (static Pressure Recovery)
For calculation of static Pressure recovery coefficient Reference [1]

Table No: 3.11-Estimating the Value of Vaned Diffuser Static Pressure Rise Coefficient
by interpolation and Approximation.

48

3.4 Bend Design Validation:

Primary flow equations:
The flow in the bend is assumed to be free vortex. An initial guess of the absolute
velocity (c5) is made on the basis of diffusion factor (DF3). Iterations are performed on
the absolute velocity (c5) and bend exit conditions are found. A 25% pressure loss
coefficient is assumed.
Inputs:
Name
blade width (w5)
diameter (d5)
inlet radius
outlet radius
mass flow (mf5)
stagnation enthalpy
mach no

location
bend
bend exit
bend
bend
bend exit
bend
bend exit

value
0.01673606 m
d4+0.05*2+w5/2*2 m =0.809 m
0.389001
0.40432736
mf4=mf3=mf2
h05=ho2
0.147

Table No: 3.12-Inputs to Bend Design

Assumptions:
Name
loss coefficient (LC3)-initial
coefficient of discharge
(cd5)

location
bend exit

value
0.25

bend exit

0.97

Table No: 3.13-Assumptions made in Bend Design

Calculations:
DF3=Diffusion factor = 1.1
W5=Diffuser width=W4 = 0.01666
mf5=mass flow rate = 4.390
D5=vane less diffuser exit diameter = D5 out-0.809, D5 in-0.778

mm
kg/s
m

LC3=Loss coefficient = 0.25

h05=h04 = 467329.6959
P05=P04 LC3*(P04 - P4) = 414862

J/kg
Pa

ct5 = ct4D4 / D5
= 50.3971

m/s

49

c5 = c4 / DF3 (initial assumption)

= 62.7661

m/s

h5 = h05 c52/2
= 463390.1126
T5 Static temperature = 454.817
corresponding to h5

CP5 = 1020.55
5

J/kg

J/kg K

= 1.3933

a5 = ( 5R t5)
=
= 426.14

m/s

M5 = c5/ a5
= 0.147
P5 =

P05 / (1+ (( 5-1) M52 /2))( 5/ 5-1)

= 414861.891
5

Pa

= P5 / R t5
kg/m3

= 3.178
Area5 =

*D5*W54
m2

= 0.0386
ae5 = Area5* Cd
= 0.0386 * 0.97

m2

= 0.037
cm5 = mf5 /

ae5

= 37.4127
c5new

62.7661 (iterated)

m/s
m/s

50

4W4

/ tan (ALPHA4) =

5W5/

tan (ALPHA5)

Since W4 =W5
ALPHA5 = tan-1(

4*

tan (ALPHA4))

= 39.0501
Table No: 3.14- Bend Design Calculations
Velocity triangle at bend

Fig: 3.5-Bend Velocity Triangle

5 (deg):
50.94

51

CHAPTER-4
Secondary Flow analysis:
Secondary flows are undesirable in centrifugal compressor as they are a direct cause for
flow (head) losses, create non uniform meridional flow profiles, potentially induce flow
separation/stall, and contribute to impeller flow slip; that is, secondary flows negatively
affect compressor performance.
Strong Circulatory secondary flows (Vortex flows) are observed in mixed-flow impellers
such as Centrifugal pumps and Compressors. These vortex flows are undesirable as they
are responsible for head losses, flow no uniformity and slip.
To reduce secondary flows and slip, turbo machinery designers often employ flow
guiding/disruptive elements like splitter vanes and other hardware modifications (which
themselves negatively effect the efficiency of the machine)
4.1 Secondary Flow Regime (Two Step Approximation)
The general characteristics of core flow diffusion process have been described for the
HPCC compressor. The complete description of the flow process , on an one dimensional
basis, in terms of basic governing equations involve a conceptual representation of the
tow zone model, or two step approximation.
It shows a regime designated as primary flow and another regime designated as
Secondary flow. The modeling indicates that there is an isentropic core flow and
that there is an average of all non isentropic stream tubes which is represented as
a single state , known as Secondary flow regime.
This gives a two step approximation, in strong contrast to the single step approximation
represented in the equations above. With a shrouded compressor rotor, it is easy to
distribute losses realistically (even though quantitative estimation may be difficult).In
that situation the rotor flow passages are fully enclosed by rotating walls, on which
boundary layers can be deemed to develop according to the local surface relative velocity
distribution. Between the rotating shroud and stationary casing there is a clearance gap,
leading to two effects. First of these is the type of friction associated with zero through
flow, which is similar to that on the rear face of the rotor, the friction loss of a fully
enclosed two-sided disc is then representative.

52

Second is the leakage loss referred to as "recirculation", which exists where there
is throughflow, whereby a small proportion of the fluid returns from rotor outlet
to inlet. When there is no rotating shroud, cataloguing of losses becomes much
More confusing. A similar recirculation path exists as mentioned above and also
similar "disc-friction on the rear face (only) of the rotor. But it is no longer
obvious how to treat the remaining effects. Growth of a boundary layer on the
outer side of the flow passage combines in some manner with the zero-through
flow friction on the casing, and additional loss takes place due to peripheral
leakage around the blade tips. The latter is primarily associated with blade loading
and hence with through-flow, so that as noted earlier, tip leakage and
recirculation losses could not readily be distinguished in any experiment.

4.2 Modeling of Recirculation and Disk Friction Loss

For want of a better approach, the method adopted here is to assume that the losses from
boundary layer growth and friction together are the same for an unshrouded rotor as for a
shrouded one. Hence boundary layer growth on the passage outer side is treated in the
same way as on the other sides, additional friction is included corresponding to a doublesided disc, and a clearance loss is added according to the best available correlation. The
question then arises how to treat the work represented by parasitic losses.
In the case of a shrouded rotor, "recirculation loss" can be regarded as involving an
unchanging mass of fluid equal to a small proportion of the through flow, which is
continuously following a cycle of compression in the rotor, expansion along the leakage
path with some heat loss to the casing and rotor, and final cooling by giving up heat to
the through flowing air.
"Disc friction" also heats the casing and rotor. Much of the heat passing to the rotor from
these sources is then absorbed by the through-flowing air progressively throughout its
passage within the rotor. This violates the normal adiabatic state during compression
upon which conservation of rothalpy is based, and leads to an some heat loss to the
casing and rotor, and final cooling by giving up heat to the through flowing air.
"Disc friction" also heats the casing and rotor. Much of the heat passing to the rotor from
these sources is then absorbed by the through-flowing air progressively throughout its

53

passage within the rotor. This violates the normal adiabatic state during compression
upon which conservation of rothalpy is based, and leads to a stagnation temperature rise
unmatched by any increase in Euler work additionally there is a quasi-instantaneous
increase in rotor inlet enthalpy due to cooling of recirculation fluid.
4.3 Impeller Boundary Layer Control:
Analysis of the impeller boundary layer was accomplished through the use of a
calculation method. A program computed the characteristics parameters of a two
dimensional boundary layer including compressibility effects and heat transfer. Both
laminar and turbulent flow was handled by the program as well as prediction of
transitions from laminar to turbulent boundary layer conditions. The basic program
required specification of free stream velocity as a function of flow path length. Flow
separation was predicted by a rapid drop off in friction coefficient with length. A wide
variety of test data was used to check program results and agreement in all cases was
excellent.
When examining the impeller, several boundary layer conditions had to be considered.
In the neighborhood of the shroud, a shroud boundary layer was calculated based on the
mean absolute velocity component and corresponding path length. Relative velocities
along the suction and pressure surface at impeller tip were used to define tow additional
boundary-layer conditions along the shroud. In the hub region, suction and pressure
surfaces as well as the mean relative velocities were employed in boundary layer analysis.
The boundary layer on the impeller tended to migrate to the tip and then to the shroud
wall as a result of centrifugal forces. Therefore flow separation on the pressure surface of
the impeller tip would extend into the shroud region and possibly trigger flow separation
across the entire area between blades.

54

4.4 Slip Factor:

The slip factor helps in estimating the energy transfer between the impeller and fluid.
Models of slip factor have been suggested in the past as a function of blade numbers,
flow coefficients, blade angles,etc. The flow pattern in an impeller is very complex due to
the development of secondary flow from the impeller. Eisner slip factor correlation is
most often used.

Weisner Slip Factor Correlation: =sqrt (1-(cos 2b)) /z

0.7

reference []

The slip factor decreases with flow rate. However, the measured and calculated values a
show a contrasting effects which was explained by Eckardt(1980) on account of inertia
effect. i.e. blade design of the impeller, guides the flow nearly to the radial direction until
the backsweep starts downstream between the planes. The increased inertia of a high flow
rate keeps the flow to maintain the strength of the secondary flow up to the impeller exit.
Therefore, the effects of the backsweep on the flow at high flow rate become small. The
deviation angle reduces at impeller exit and the slip factor increases with flow rate.

55

4.5 Mixed Flow Calculation Flow Chart:

Input geometry, N, m

Solve primary zone calculations

considering isentropic core flow

Solve Secondary Zone equations

considering regions of Secondary flow
losses

Solve Mixing Calculations

Perform Design Optimization and verify Aerodynamic

inputs/Assumptions

Perform Inverse Mean line optimization where in

aerodynamic parameters are fixed and geometry is obtained.

Once Geometry and aerodynamic Inputs constrained create Blade

Model using BladeGen Tool.

Fig: 4.1- Mixed Flow Calculation Flow Chart

56

4.5.1 Impeller Secondary flow calculations:

These calculations are performed on a Test Compressor- HPCC-Reference-[]
= ms/m (20% of core flow)
row2s = row2*1.2 (initial)
= 1-(m(1- )/(cos2b*Ageo*w2*row2))
= 0.679
Cm2s = [m/ae2-row2*cm2(1-)]/row2s*
=
Ct2s =

87.5684
Ct2s=u2-(cm2s/tan(cos2b)-(1-sf)*u2
364.65
Sqrt((u2-ct2s)^2*cm2s^2)
155.25
hT+ disk friction loss
hT2s-(2s^2/2)(2^2/2)
406919.964
h2s/Cp
402.29
p2/R/T2s (iterated)

=
W2s =
=
hT2s =
hs2 =
=
T2s =
=
row2snew =
=
= 2.332
2s = cos(cm2s/w2s)

C2s

m/s

m/s
m/s

Joules
m/s
m/s
Kg/m3
J/kg K

= -55.665

deg

= sqrt(cm2s^2+ct2s^2)

=
To2s =
=
po2s =
=

375.01
T2s+(c2s^2)/(2*cp2))
471.81
p2*(To2s/T2)^(g2/(g2-1))
496215.63

m/s
J/kg K
m/s
pascal

Table No: 4.1-Secondary Flow Calculations

57

4.6 Mixing Calculations:

Row2m

= Row2*1.2 (initial)

Ct2m = (((m-ms)*ct2)+(ms*ct2s))/m;

= 366.2681

m/s

cm2m = m/(ae2*row2m)
= 86.44

m/s

p2m = p2+((((m-ms)*cm2)+(ms*cm2s)(m*cm2m))/(area2))
c2m=
sqrt (pow(ct2m,2)+pow(cm2m,2));
=
To2m

376.33

m/s

(((m- ms)*cp2*To2)+(ms*cp2*To2s)+
((diskloss+recir)*ms))/(cp2*array[11]);

T2m

= 469.44
= To2m-(pow(c2m,2)/(2*cp2))

=
row2mnew =
=
po2m =

=
=
=
=
=
sf =
=

399.44
p2m/R/T2m (iterated)
2.3491
p2m*pow(To2m/T2m,(g2/(g2-1)));
sf=pnunew*((lamdanewtan(array[7]*torad))/lamdanew);
475814.99
ct2m/cm2m
4.23
ct2m/u2
0.74
*(( -tan(cos2b)/ );
0.9521

Table No: 4.2-Mixing Flow Calculations

58

The mixing process has been realistically computed in order to deduce the levels of
discharge mixing losses and to have basis upon which one can estimate the loss in kinetic
energy due to formation of secondary flow.
The above two equations give the description of impeller flow process on a two zone one
dimensional basis. Computing the mixing process is important. It provides a closure to
the system of equations. Based on the mass averaged total pressure entering the mixing
control volume and the mixed-out total pressure leaving the control volume gives a
measure of the mixing loss. Secondary flow has been modeled as a low momentum or
wake-flow, and its presence at a relative velocity less than isentropic core flow relative
velocity implies a loss in kinetic energy on formation of secondary flow.

59

CHAPTER-5
DESIGN VALIDATION
5.1 Compressor Design Validation:
The geometric input for the impeller is taken from HPCC input. The incidence angle at
the inlet is kept constant at -5 (deg) from which blade angles are determined.
The Mach no is initially assumed as 0.3 at inlet and 0.6 at exit and iterated to get the
correct value of Mach no.
Impeller efficiency is determined by calculating losses and by subsequent iterations.
Validation Results: Comparison of CFD results with 1D design for HPCC
Reference-[ ]
1
impeller
inlet

2
impeller
exit

101.3

467.253

453.386

418.062

414.862

4.09

(CFD)
Total Temperature
To (K)
30 blades (1D
calculation)

101.129

458.936

445.861

420.949

414.328

4.097024592

288.166

462.115

460.626

456.847

456.747

1.5850

(CFD)
Static Pressure p
(Kpa)
30 blades (1 D
calculation)

288.22

453.91

453.93

454.03

453.99

1.575150926

91.0855

269.154

276.769

407.394

408.661

4.48656

89.767

272.875

293.038

397.862

400.453

4.461026881

Temp Ratio

279.52

395.148

400.46

453.537

454.817

1.627

278.382

392.117

402.448

446.718

449.68

1.615334325

0.392

0.919

0.87124

0.193

0.147

0.421

0.89

0.805

0.273

0.214

Total Pressure po
(Kpa)
30 blades (1D
calculation)

(CFD)
Static Temperature
T (K)
30 blades (1 D
calculation)
(CFD)
Mach No
(M)
30 blades (1 D
calculation)
(CFD)

3
vaneless
diffuser exit

4
vaned
diffuser exit

5
bend

Pr Ratio

Temp Ratio

Pr Ratio

Table No: 5.1-Comparision of CFD results With 1D code for Test Compressor-1

60

Validation carried out on Test Compressor 1:

Mean Line Flow Modeling based on a compressor tested by Eckardt (1980) available in
the book Introduction to Turbo Machinery by David Japikse.
The Compressor Features an impeller with radial blades at the exit.
The design is Validate by comparing results with the measured results obtained from test
compressor tested by Eckardt.

Table No: 5.2-Comparision of CFD results With 1D code for Test Compressor-2

61

Validation carried out on Test Compressor -2:

Compressor design validation by comparing results of design Code with (NASA-TM71719) Test Compressor Developed for Energy Research Centre and Development
Agency Automotive Gas Turbine program NASA

Table No: 5.3-Comparision of Test Data Results with Results from 1D code for
Test Compressor-3

62

CHAPTER -6
Inverse Design for Centrifugal Compressor using Meanline
6.1 Approach:
This approach is used for determining the optimum geometric parameters of centrifugal
compressor stages given specific performance requirements. This is commonly known as
inverse design approach.
The opposite process, which of calculating the performance parameters based on
geometric details is usually called analysis, or direct calculation. An algorithm and
computer code implementing the inverse approach is described.
As an alternative to commercially available inverse design codes, this program
uses a trusted database of loss models for individual stations, such as impellers,
vaned diffusers, etc.
This algorithm extends applicability of the inverse code by ensuring energy
conversation for any working medium, like imperfect gases. The concept of loss
coefficients for rotating impellers is introduced for improved loss modeling.
The governing conservation equations for each component of a stage are
presented and then described in terms of an iterative procedure which calculates
the required one-dimensional geometry.
A graphical user interface which facilitates user input and presentation of results
is also described. The visibility and re-usability of the code is highlighted as a
platform which easily provides for maintainability and future extensions.
The one-dimensional codes for centrifugal compressor stages have typically solved the
problem of analyzing geometry rather than doing inverse design. Furthermore, many of
these codes have no generalized loss models, but have instead past compressor
performance data tabulated to indicate performance of the machine under study. This
technique works for all cases where the new design is similar to an existing one, but
poorly when the new design differs significantly.

63

6.2 Methodology
These codes are used for the design in an iterative manner, that is, the designer inputs the
geometry and sees if the code predicts the desired performance. A final design is thus
arrived at when the desired performance is arrived at. A more direct route to the final
design was thus desired.
These codes use a generalized set of loss models; it was desired to incorporate the basic
idea of these codes with the specific loss models capabilities.

6.2.1 Variation in Aerodynamic Parameters with Impeller Inlet Tip

Radius for an In-house Test Compressor
R1tip
(m)

Po2
Kpa

P2 Kpa

To2 K

T2 K

alpha2 deg

rotor
Efficiency

M2

M1t

0.0265

368199

224121

444.544

386.048

65.89

0.8755

0.82745

0.822

0.027

376191

228825

445.463

386.774

66.4131

0.876255

0.8391

0.722

0.0275

379951

231036

445.891

387.111

66.65

0.876615

0.844575

0.658

0.028

381841

232144

446.133

387.286

66.7694

0.876

0.8472

0.607

0.0285

382593

232586

446.21

387.63

66.8164

0.87688

0.8483

0.567

0.02886

382645

232619

446.228

387.379

66.82

0.876885

0.84845

0.541

0.02886

396248

240552

447.644

388.461

67.6323

0.8783

0.8676

0.541

Table No: 6.1-Variations in Impeller Exit Conditions as a Function of Impeller Inlet Tip
Mach number.

64

6.2.2 The plot below shows the variation of impeller inlet tip radius Vs impeller
efficiency
Based on the plot below we can make a reasonable estimate of how much we can twitch
the inlet radius (reduce space) without causing much reduction in efficiency

rotor efficiency

impeller inlet tip radius Vs rotor efficiency

0.855
0.85
0.845
0.84
0.835
0.83
0.825
0.82
0.815
0.0265

0.027

0.0275

0.028

0.0285

0.02886

impeller inlet tip radius

Fig: 6.1-Variation of Rotor Efficiency with Impeller Inlet Tip Mach number
6.2.3 The plot below shows the variation of Impeller inlet tip radius as a function of
impeller inlet tip mach no.
The radius at tip should be varied in such a way that the tip mach no at inlet does not go
beyond a critical value of 0.7

impeller inlet tip mach no

impeller inlet tip radius Vs inlet tip mach no

0.900
0.800
0.700
0.600
0.500
0.400
0.300
0.200
0.100
0.000
0.0265

0.027

0.0275

0.028

0.0285

0.02886

impeller inlet tip radius

Fig: 6.2- Variation of Inlet Tip Mach number with Impeller Inlet Tip Radius

65

6.2.4 The Table Below shows the variation of losses as a function of Imepeller Inlet
Tip Radius.
skin loss
(joules)

R1 tip

Recirculation
(joules)

blade loss
(joules)

Disk friction
loss (joules)

incidence loss
(joules)

0.0265

19036.4

4151.99

5959.91

2828.17

524.681

0.027

17512.1

4254.82

6116.82

2882.39

145.212

0.0275

16641.1

4344.69

6274.48

2907.88

22.9929

0.028

16072.7

4429.19

6440.35

2920.69

0.69

0.0285

15690.2

4510.3

6612.34

2925.82

30.1134

0.02886

15495.8

4567.22

6739.52

2926.2

72.3906

0.02886

11838

4608.42

7013.33

3017.84

533.209

Table No: 6.2-Various Losses Tabulated as function of Impeller Inlet Tip Radius.
6.2.5 The plot below shows the variation of impeller losses as a function of impeller
tip radius
The loss distribution can be seen and its variation and dependence on the geometry is
clearly visible

Impeller Losses

Impeller Loss Vs Impeller Tip Radius

20000
18000
16000
14000
12000
10000
8000
6000
4000
2000
0

Skin Loss
Recirculation
Blade Loss
Incidence loss

0.0265 0.027 0.0275 0.028 0.0285 0.0289

Impeller Tip Radius

Fig: 6.3 Plots of Impeller Losses as a Function of Impeller Tip Radius.

66

6.3 Other Modes of Operation:

The fundamental purpose of the algorithm presented here is to perform a design
calculation that is to find the geometry given performance.
However, the code can also be run in the analysis mode where performance parameters
are found given geometry.
A mixed mode of calculation, where a combination of some geometry and some
performance are given resulting in a mixture of geometry and performance, is also
possible.
This mode is particularly useful for designers who are trying to modify an existing
machine and have access to some of the geometric information, and some of the
performance conditions. The output is always the full set of geometry, velocities and
thermodynamic properties.
The following describes how each component offers the user the choice of design or
performance parameters.
To begin, for the overall stage the user may input either the power or outlet total
pressure. The user may input either the power or outlet total pressure. The user
may also enter both, but if this is done it is equivalent to setting the stage
efficiency, so efficiency will not be calculated independently through loss models.
For the inlet (station 00), the user may input, in addition to the total conditions,
either the static pressure, cross sectional flow area, or velocity (meridional). The
same is true for the outlet when it is a discharge cone.
For the inlet guide vane, the user may choose either the angle of the vane, or the
preswirl coefficient. Of course the user may also choose a vaneless inlet.

67

For the impeller the user chooses either the diameter or the work input coefficient,
either the diffusion factor or outlet width, and either the mass flow rate or the
outlet blade angle. Of the latter, the user may choose both, but doing so is
equivalent to setting the slip so it will not be computed by a slip model.
For a vaneless diffuser, as discussed, the user may choose any tow of outlet
diameter, width, or diffusion factor. For a vaned diffuser any three of diameter,
width, diffusion factor, or outlet angle are chosen.
The basic conservation equations shown above naturally do not change when
code is run in various modes. The iterative algorithm does not change, however,
and must take into account the great many combinations of inputs the user may
wish to run.

68

CHAPTER-7
ANSYS-BladeGen
7.1 Introduction to ANSYS BladeGen
BladeGen is a component of ANSYS Blade Modeler. The Blade Modeler software is a
specialized, easy to use tool for rapid 3D design of rotating machinery components. The
software is used to Design mixed flow and radial blade components of Compressor and
Diffusers.
BladGen with its rich set of tools and functionalities for designing a turbo machinery
blade from scratch, using industry-specific tools, workflow, and language that the blade
designer expects.
With BladeGen, the user can re-design existing blades to achieve new design goals or
create completely new blade designs from scratch. Re-designing or evaluating existing
blade design, BladeGen facilitates the import of blade geometry interactively or through
user supplied files.
Blades were created using Angle-Thickness view and by choosing a radial blade
impeller. Input geometric details were provided in a radial vs. axial plot.
BladeGen allows sculpted or ruled element blades with linear or compound lean
leading or trailing edges.
Over/Under-Filing can be applied and leading and trailing edge shapes are easily
specified as a full radius, an ellipse ratio, or a simple cutoff. BladeGen allows
sculpted or ruled element blades with linear or compound lean leading or trailing
edges. Over/under-Filling was applied and trailing edge shapes are easily
specified at full radius, and ellipse ratio, or a simple cutoff.
With its provision for seamless path to both structural and fluid analysis, which enables
the user to efficiently transition from preliminary blade design, to full 3D viscous flow
analysis, and finally to users native CAD system.

69

7.2 Using BladeGen

This BladeGen introduces a method of designing blades that is interactive, allows for
blade section development familiar to the aero-designer.
All the available inputs are geometric in nature but are represented as familiar
aerodynamic parameters. Any preprocessing, separate analysis, or designer experience is
easily communicated to blade model through a robust interface. General stream surfaces
are used to allow for stream surface generation and editing. These curves can be read in
from a file to allow for compatibility with existing flow analysis codes.
7.3 ANSYS BladeGen Blade Pictures

Three Views are made available of the blade generated using BladeGen:
Figure 1
The left most section shows the blade profile and in the blade to blade view. The
Shape and the NACA thickness distribution is clearly visible.
The middle section shows the meridional view of the impeller blade in a
R (radial) Vs (Axial Plot).
The last section shows the Grid View of the Compressor- impeller. The hub and
shroud are clearly visible In this view.

Figure 2
This view shows the impeller in the Grid Form.

Figure 3
This gives the shaded view of the Impeller Blades.

70

Fig: 7.1-Blade-Blade View, Meridional View and Meshed View

71

Fig: 7.2- Meshed View of Impeller Blade

72

Fig: 7.3- Shaded View of Impeller Blade

73

CHAPTER-8
Compressor Performance Based on impeller Geometry
8.1 Plots showing Loss Distributions as a function of Geometry.

impeller exit
backsweep angle
(deg)

skin
Loss

Recirculation
loss

(joules)

disk friction
loss

(joules)

blade loading
loss

(joules)

incidence
loss

(joules)

(joules)

20

13245.6

5093.72

3075.67

9114.72

72.3906

25

13829.3

4932.69

3036.26

8370.47

72.3906

30

14604.1

4753.58

2985.6

7560.05

72.3906

35

15621.8

4541.79

2917.67

6630.04

72.3906

40

16962

4274.74

2822.77

5519.8

72.3906

45

18761.7

3899.83

2675.54

4097.11

72.3906

50

21311.9

3244.59

2392.66

2057.2

72.3906

Table No: 8.1- Variations of Impeller Losses with Impeller Blade Backsweep
The plot below shows the Variations of individual losses as a function of impeller blade
backswept angle.

Impeller Losses
(Enthalpy loss)

Impeller Blade Exit angle VS Impeller Losses

25000
20000
15000
10000
5000
0
20

25

30

35

40

45

50

Impeller Blade Exit angle

Skin Friction Loss

Recirculation Loss

Blade Loading Loss

Incidence Loss

Disk Fricition Loss

Fig: 8.1- Plot of Impeller Blade Exit Angle Vs Impeller Losses

74

8.2 Plots showing Variation of Impeller Exit Pressure as a function of

Geometry.

Table No: 8.2- Values of Stagnation Pressure are tabulated for Variations in Blade Backsweep

Variation in stagnation pressure rise across impeller plotted as a function of blade

backsweep angle

Stagnation Pressure Ratio

Blade Backsweep Angle Vs Stagnation Pressure Ratio

4.5
4
3.5
3
2.5
2
1.5
1
0.5
0

Impeller exit
Stagnation Pressure
Ratio

25

30

35

40

45

50

Blade Backsweep

Fig: 8.2- Blade Backsweep Vs Impeller Stagnation Pressure Ratio

75

Impeller Exit Stagnation

Pressure

Blade Backsweep Angle Vs Stagnation Pressure Rise

500000
400000
300000

Impeller Exit
Stagnation Pressure

200000
100000
0
25

30

35

40

45

50

Impeller Exit Backsweep Angle

Fig: 8.3- Blade Backsweep Vs Impeller Stagnation Pressure Rise

76

CYCLS CODES
Gas Turbines, Compressors

Operational Database

Synthesis-(8)

Compressor
Meanline
Design
Optimization
-(7)

3D Geometry
Generation and
Through Flow
Analysis-(6)

CFD
Preprocessor-(5)
(9)
ComponentDevelopment-1
Product Development-2
Electronic Laboratory-3

Designers

Rapid Prototyping
-(10)
Numerical M/c
Casting/Molding

3D flow
Modeling-(4)

CFD Post
Processor-(3)
Computer
aided
Drafting(1)

Rotor
Dynamic
Seals
Bearing-(2)

Planning for Product

life, Maintenance and
Support
Design Principles
and Educational Aid
Fig: 8.1
From Aerodynamic Design of Blades to Blade Fabrication

77

CHAPTER-9
9.1 CONCLUSIONS:
The present 1D code correctly predicts the pressure ratio, total temperature
increase and efficiency of centrifugal compressor impeller over a wide range of
operating conditions.
The loss models used are generalized and can be applied to a number of
compressors with out modifications.
The flow in the Vaneless, Vaned Diffuser and Bend has been modeled based on
Certain Values of Loss Coefficients which have been obtained from Experimental
Plots and Empirical Correlations and hence may not be very accurate.
The Design Validation successfully suggests that the design process has reached
reasonable levels of accuracy but has scope for improvements like the
implementation of Shock Losses when Mach No reaches values > 1.2.
The predicted flow development through the impeller is in good agreement with
the measured data.
The code reasonably simulates the evolution of secondary flow in the impeller
which affects the jet-wake formation and location. The core of wake region at the
impeller exits near the suction side for the near stall flow rate, and the
shroud/suction side corner for the design and near choke flow rates.
The predicted values of slip factor increase with the flow rate for backswept
Eckardt impeller and are in good agreement with measured values.
The mixed out velocity vectors are in good agreement with the values obtained
from code except for small differences in the calculated total conditions at the
vaned and Bend exit.
The Generalizations of the correlations imposes restriction on the robustness of
the code, as the generalizations make reasonable assumptions and hence the
accuracy of the code may be affected slightly.
The exhaustive Testing of the design using Test Compressors from Open
literature implies that the code is robust and generalized and can be used for a
wide range of compressors.

78

9.2 Towards a Better Design:

A plot of various parameters like clearance, incidence angle, Back sweep angle, Blade
number should be plotted against rotor efficiency to determine the best value of the
parameter to give highest rotor efficiency.
The accuracy of the design depends on the how effectively the losses are
modeled.
Knowing the blade profile and thickness variations of blades will help produce a
more efficient design.
The angle of incidence (i=b-) be so chosen that the incidence loss is minimum.
The blade backflow angle effects the rotor efficiency, hence a optimum value of
back flow should be used.
The secondary flow and mixing should be considered for more efficient design.
Considerable area blockage would occur in impeller and diffuser blades, hence
blockage factor should be considered in the design.
The number of blades and splitters cause variation in impeller pressure rise and
rotor efficiency hence should be appropriately chosen.
The pressure loss coefficient is a function of length/width (L/W) ratio, blockage
and Area Ratio hence appropriate plots or contours be chosen in its determination.
Slip factor depends on the blade geometry and no of blades and is a critical
component in determining rotor efficiency and stage pressure rise hence for the
determinations of slip factor the best formula should be chosen.
The loss models should be as exhaustive as possible considering all the losses like
blade loading, skin friction, disk friction, recirculation and leakage losses.
There will be considerable loss in the dynamic head during energy transfer
processes like impeller and diffuser, hence the pressure loss coefficients should be
appropriately chosen.
There will be variation in the values of constants like K and Cp; hence appropriate
equations should be used to study their variations.
When the Mach no is >1.2 shock losses will occur and the loss models for shock
waves should be incorporated in the design.
The mass flow is assumed constant but there will be slight variations in the mass
flow rate across the compressor.

9.3 Scope for Future Work

The Design can be made more robust with provision to include Shock Losses.
The accuracy of the design could be improved by using Integration Methods and
Numerical Methods.
The number of Stations could be increased to average the flow at these stations to
gather more comprehensive information about the flow field.
CFD analysis could be carried out in a later stage to ascertain the validity of the
design.
Inverse Meanline design could be employed to get rough estimates of geometry
by aerodynamically constraining it with critical parameters.
79

APPENDIX-A
SNAPSHOTS:
The following figures show how input is fed on a text file which is read by the Impeller
and Compressor Code:
A.1 Impeller Inputs
The input details to the Impeller are fed in Impeller_inputs.txt. The impeller code reads
the details from this file and performs iterations. The results of the impeller code are
written into a file impeller_diffuser_micro.txt.

Fig: A.1-Impeller Input Text File

80

A.2 Vaneless ,Vaned and Bend Inputs

The inputs for diffuser and bend calculation are appended on the file created by
the impeller code which has impeller exit conditions already written on the file.
The Diffuser (Vaneless & Vaned), Bend geometry are furnished in this file and
the file is read by the compressor code, which performs design iterations to get
the complete compressor details.

Fig:A.2-Vaneless, Vaned Diffuser and Bend Input File

81

A.3 VIEW SOURCE CODE:

Fig: A.3-Impeller Source Code

82

A.4 Figure below shows the code written of hub-tip and mean conditions at inlet
Impeller inlet.

Fig: A.4- Source Code For Calculating Conditions at Hub, Tip and Mean

83

A.5 Figure below shows the code written for Loss Calculations:

Fig: A.5- Source Code Showing Correlations of Loss Models

84

A.6 Figure Below shows the Code for Vaneless diffuser Design:

Fig: A.6- Source Code Showing Correlations For Vaneless Diffuser Design

85

A.7 Figure below shows the code for Vaned Diffuser Design:

Fig: A.7- Source Code Showing Correlations For Vaned Diffuser Design

86

SNAPSHOTS OF OUTPUT:
A.8 Impeller Output

Fig: A.8- Output of Impeller Design Code

87

A.9 Impeller, Vaneless Diffuser, Vaned Diffuser and Bend Output

Fig: A.9- Output of Vaneless Diffuser, Vaned Diffuser and Bend Design Code

88

Reference:
1) Analytical correlation of centrifugal compressor design geometry for maximum
efficiency with specific speed by Michael R Galvas, Lewis Research Centre ClevelandOhio, March 1972, NASA
2)Coordinates for High Performance 4:1 Pressure Ratio Centrifugal Compressor Ted F
McKain, Greg J Holbrook, Detroit Diesel Allison, Lewis Research Centre NASA, July
1997
3) Microjet_design_document, compressor_code-Honeywell
4) Introduction to Turbo machinery- David Japikse, Nicholas C.Baines.
5) Centrifugal Compressor Surge and Speed Control- Jan Tommy Gravdahl, Member,
IEEE, and Olav Egeland, Member, IEEE, 5th sept, 1999
6) Method of Performance Prediction for Centrifugal Compressors-M.V HerbertNational Gas Turbine Establishment.
8) Laser Anemometer Measurements of the Flow Field in a 4:1 Pressure Ratio
Centrifugal Impeller-G.J. Skoch U.S. Army Research Laboratory Lewis Research
Center Cleveland, Ohio, June 2-5, 1997
9) Michael R Galvas , A Fortron Program for predicting the Off Design Charactersitics of
a Centrifugal Compressor, November 1973,Lewis Research Centre and US Army Air
Mobility R& D Laboratory Cleaveland, Ohio
10) Dr. C.A Bryce, J.R. Ervin, Dr E.L. Nelson, Dr. R. K .Tu, Small High Pressure Ratio
Compressor Aerodynamic and Mechanical Design, NASA Lewis Research Centre
NAS3-14306
11) David Japikse (1994). Centrifugal Compressor Design and Performance, Concepts
ETI, Inc
12) David Japikse (1994). Introduction to Turbo Machinery, Concepts ETI, Inc
13) Dr . R. K. Bansal (Edition-2002). Fluid Mechanics and Hydraulic Machines, Laxmi
Publication (P) LTD.
14) SM Yahya. Turbines Compressor and Fans, Tata McGraw-Hill Publishing Company
Ltd.
15) Yuri Biba and Peter Menegay, Inverse Design of Centrifugal Compressor Stages
Using a Meanline Approach., June 2002, Aerothermo Design Group,Olean, New
York,USA.

89

16) Yasootoshi Senu, Researches On Centrifugal Compressors, 2005, Communicated by

Hidesato Ito, M.J.A
17) Michael R. Galvas, A Compressor Designed For the Energy Research and
Development Agency Automotive Gas Turbine Program, May-1975, Technical
Memorandum, (NASA_TM_X-71719),

90