Академический Документы
Профессиональный Документы
Культура Документы
Proceedings of ICFD11:
Eleventh International
Eleventh International ConferenceConference of
of Fluid Dynamics
Fluid Dynamics
December 19-21, 2013, Alexandria, Egypt
December 19-21, 2013, Alexandria, Egypt
ICFD11-EG-4XXX
ICFD11-EG-4103
Design and Optimization of a Multi-Stage Axial-Flow Compressor
Prof. Dr. Atef M. Alm-Eldien Prof. Dr. Ahmed F. Abdel Gawad Prof. Dr. Gamal Eng. Mohamed G. Abd El Kreim
Mech. Power Eng. Dept. Mech. Power Eng. Dept., Faculty of Eng., Hafaz M.Sc. in Mech. Power Eng.
Faculty of Eng. Zagazig Univ., Egypt Mech. Power Eng. Dept.
Currently: Faculty of Eng. Steam Turbine Maintenance Engineer
Vice-president Mech. Eng. Dept. Abou Sultan Power Plant
for Students Affairs College of Eng. & Islamic Architecture Port-Said University
Umm Al-Qura Univ., Saudi Arabia Egypt East Delta Electricity Production Company
Port-Said University, Egypt Egyptian Electricity Holding Company
Fellow IEF, Assoc. Fellow AIAA
Egypt
atef_alameldin@yahoo.com Member ASME, ACS, SIAM, AAAS
mgaber_eg@hotmail.com
afaroukg@yahoo.com
I. INTRODUCTION
ABSTRACT I.1 Previous Investigations
The objective of this paper is to define a methodology There is a large volume of literature on compressor
for the design and analysis of multistage axial-flow design and its design parameters. However, the design
compressors. A numerical methodology is adopted for method is different from one research to another. The
optimizing the efficiency at the design point of a fifteen- intention of this review is to show the reader the related
stage axial-flow compressor with inlet guide vanes (IGV). work and to orient the reader where the current work stands
The calculations are carried out along the mean streamline in relation to the literature.
using the principals of thermodynamics and aerodynamics. Barbosa [1] used a streamline curvature for the flow
A computer program was developed that simulates the calculation along a multi-stage axial compressor of known
compressor model. By specifying the geometry geometry. In his work, many streamlines are adopted from
specifications (tip clearance, aspect ratio, thickness-chord hub to tip of the blades and divided in sections applied at
ratio, blockage factor, etc.) and design parameters (mass the inlet and outlet of each row. Hence, all flow properties
flow, rotational speed, number of stages, pressure ratio, can be determined at each point of the intersection among
etc.), an accurate numerical model can be generated. This the sections of the cascade and the streamlines forming
modeling technique is much simpler than the usual control surfaces. Teinke [13] used a mean-line stage
computational methods that need much more stacking method for axial-compressor prediction. In his
modeling/programming effort and computer run-time. method, the calculations are based on the mean streamline
Starting from a newly-designed axial-flow compressor, an of the axial-compressor channel. The flow properties such
optimized version is obtained with improved design-point as temperature, pressure, velocity and the dimensions of the
efficiency. So, once we get the optimized geometry of the equipments including the balding angles are determined at
compressor, the original geometry is altered to maximize half blade. He stated that his method, when simulated in
the efficiency at the design-point. Concerning the optimized computer, presented fast numeric convergence and
version, analytical relation between the isentropic sufficient accurate results for a first analysis of the
efficiency of the compressor and the flow coefficient, the compressors performance. Casey [3] presented a
work coefficient, the flow angles and the degree of reaction computational program to calculate the efficiency of a
are obtained. single-stage axial compressor to analyze the one-
dimensional condition with a pressure ratio of 1.2. The
author described the importance of analyzing the incidence,
KEYWORDS deviation, profile losses, secondary losses and boundary-
Axial-flow compressor, Efficiency, Performance layer limits from hub tip related with the tip clearance in
optimization. order to predict the axial-compressor performance. The
deviation angle was obtained through Carter's rule [2]. He
used Lieblein's model [10] to calculate the profile losses.
The effect of the relative Reynolds number corrections to
the friction losses was calculated by Koch's model [8] and
1
the Mach number correction was done by Jansen, and
Moffatt procedure [7]. Seyb [12] presented a program for
the design and prediction of an eight-stage, constant outer-
diameter, axial compressor.
As can be seen from the above literature survey, there is
a real need for a better direct method for the design and
optimization of the multi-stage axial compressor. The
method should consider all the parameters that affect the
compressor performance and be reliable for all the
compressor stages.
Therefore, a new method is introduced in this paper to
achieve these objectives. The method is fully explained in
the following sections.
3
To make the velocity triangles dimensionless, we divide ⎛1⎛ 1 ⎞ ⎞ Inflow angle (14)
all velocities by the blade speed U. The outcome of this is α =α1 3
= arctan ⎜⎜ ⎜ 1 − R − ψ ⎟ ⎟⎟
⎝φ ⎝ 2 ⎠⎠
shown in Fig. 4 from which important results is obtained. ⎛1⎛ 1 ⎞ ⎞ Outflow angle (15)
α = arctan ⎜⎜ ⎜ 1 − R + ψ ⎟ ⎟⎟
2
⎝φ ⎝ 2 ⎠⎠
⎛1⎛ 1 ⎞ ⎞ Relative inflow angle (16)
β 1 = arctan ⎜⎜ ⎜ R + ψ ⎟ ⎟⎟
⎝φ ⎝ 2 ⎠⎠
⎛1⎛ 1 ⎞⎞ Relative outflow angle (17)
β 2 = arctan ⎜⎜ ⎜ R − ψ ⎟ ⎟⎟
φ
⎝ ⎝ 2 ⎠⎠
2
w1 = φ 2 + ⎛ R + ψ ⎞ Rotor relative inflow velocity (18)
⎜ ⎟
U ⎝ 2 ⎠
2
w2 = ⎛ ψ⎞ Rotor relative outflow velocity (19)
φ 2 + ⎜ R− ⎟
U ⎝ 2⎠
2
c 1 = c 3 = φ 2 + ⎛ 1− R − ψ ⎞ Rotor entry and stage exit
⎜ ⎟
U U ⎝ 2 ⎠
2
Swirl velocities Wθ2 and Cθ1 can be related to φ ,ψ and c 2 = φ 2 + ⎛ 1− R + ψ ⎞ Stator inflow velocity (21)
⎜ ⎟
U ⎝ 2 ⎠
R as follows
−
R = h 2 h1 (3) • Lift and drag coefficients in terms of duty coefficients
−
h 3 h1 The dimensionless parameters that indicate profile
Since the stages are repeating for which entry and aerodynamic quality are the lift and drag coefficients CL
leaving velocities are identical, C3 = C1, the denominator of and CD. It is important therefore to express CL and CD in
Eq. 3 may be simplified to terms of the duty coefficients which have a total control
⎛
⎜ c 3 ⎞⎟ ⎛⎜
2
c 1 ⎞⎟
2
(4) over the shape of the velocity triangles. Lift coefficient for
h 3
− h 1
=
⎜ h 03
−
2 ⎟
−
⎜ h 01
−
2 ⎟
= h 03
− h 01
= Δ h 0
a cascade can be expressed in terms of the relative inflow
⎝ ⎠ ⎝ ⎠
Then, the numerator is written as and outflow angles β1 and β2, the vector mean of them β∞
⎛ 2
⎞ ⎛ 2
⎞ and the pitch to chord ratio, t/l as follows:
= ⎜ h 02 − c 2 ⎟ − ⎜ h 01 − c 1 ⎟ =
(5)
h h h h C = 2 (tan β − tan β )cos β − C tan β
− − t (22)
2 1 ⎜ 2 ⎟ ⎜ 2 ⎟ 02 01 L
l 1 2 m D m
⎝ ⎠ ⎝ ⎠
But since there is no work or heat input through the But we have
stator, h02 = h03 and thus h02 - h01 = h03 - h01 = Δh0. Also, 1⎛ 1 ⎞, 1⎛ 1 ⎞ (23)
tan β = ⎜R + ψ ⎟ tanβ = ⎜ R − ψ ⎟
since the axial velocity is assumed to be constant, we have φ⎝ 2 ⎠ 2 φ
⎝ 2 ⎠
( ) ( )
1
2 2 2 2 2 2 Hence,
c 2 − c1 = c x + c θ 2 − c x + c θ 1 =
(6)
c θ 2 − c θ 1 = (c θ 2 − c θ 1 )(c θ 2 + c θ 1 ) ( β β )= φ
2 2 1 R (24)
tan β m
=
2
tan
1
+ tan
2
cθ 2 + cθ 1 = 2U (1 − R ) (9)
Thus, we have Considering the drag coefficient, defined by
cθ 1 = 1 − R − 1 ψ (10) (26)
U 2 D ⎛t⎞
CD = 1 = ζ ∞ ⎜ ⎟ cos β ∞
2
ρ w∞ l ⎝l⎠
cθ 2 = 1 − R + 1 ψ (11) 2
U 2
wθ 1 = 1 − cθ 1 = R + 1 ψ (12) Where, the cascade loss coefficient is based on the vector
U 2 U mean velocity w∞
wθ 2 = 1 − cθ 2 = R − 1 ψ (13)
U U 2 (Δ p0)loss (27)
ζ∞=
We note that the dimensionless velocity triangles and 1 2
ρW ∞
hence the blade shapes required to achieve them are totally 2
determined by the stage duty coefficients φ ,ψ and R. Since, Δ poR & Δ p os (28)
ζR= ζs=
It follows that all angles and velocities may be expressed 1
ρ W 12
1
ρ C 22
as functions of φ andψ as follows: 2 2
4
Where, ζR,ζs are the rotor and stator loss coefficients
expressed in terms of the exit velocities C2 and w3 relative
to the blade rows.
Hence, we get
2
Δ poR ⎛ w∞ ⎞
2
⎛ cos β 1 ⎞ (29)
ζR= 1
= ζ ∞ ⎜⎜ ⎟⎟ = ζ ∞ ⎜⎜ cos β ⎟⎟
ρ W 12 ⎝ 1⎠
w ⎝ ∞⎠
2
So that CD becomes
⎛ ⎞
C D = ζ 1⎜ l ⎟
t cos β ∞
3
⎛ t ⎞ 2φ 4 φ + (1 + φ )
= ζ 1⎜ ⎟
2 2
( ) (30)
⎝ ⎠ cos 2 β 1
( )
3
⎝l⎠
4φ 2 +1 2
β1 β1⎛t ⎞
DF = 1 −
cos
+
cos
(
⎜ ⎟ tan β 1 − tan β 2 ) (31)
cos β2 2 ⎝l⎠
The rotor and stator blade rows will have different profile
geometry. In order to select suitable values of pitch/chord
ratio t/l to control aerodynamic loading, we have for the
rotor
4φ + (1−ψ ) (32)
2 2
t ψ ⎛ ⎞
DF = 1− +⎜ ⎟
R
4φ +
2
(1−ψ ) 2
φ
⎝l⎠ 4 2+
(1−ψ )2
5
• Calculation of static pressure and temperature at • Calculation of the equivalent diffusion ratio (Deq*)
rotor inlet (P1, T1)
1. From EES, find h01, S01, K01 and Cp01 using (P01, T01). cos(β 2) ⎡ cos (β1)
2 ⎤
2. Find h1 = h01 - C12/2. Deq* = cos( ) ⎢1.12 + 0.61 (tan(β 2) − tan(β1))⎥⎥ (35)
β1 ⎢ ⎣
σ
⎦
3. From EES, find ρ1, Cp1, k1 and μ1 using (h1, S1 = S01).
4. Find T1 = T01 - (C12/2 × Cp1).
⎛ k ⎞
⎜ 1 ⎟
⎛ ⎞⎜ k −1 ⎟ • Calculation of the compressor losses
5. Find p = p ⎜ T 1 ⎟⎝ 1 ⎠
1 01 ⎜ ⎟ • Profile loss model
⎝ T 01 ⎠
• Calculation of static pressure and temperature at The profile loss model used is a modified version of the
rotor outlet stator inlet (P2, T2) two-dimensional low speed correlation of Lieblein
et al. [10], Fig. 9.
1. Find compressor exit temperature
⎛ 1 ⎞⎛ k −1 ⎞
⎜ ⎟⎜ 01 ⎟
Te =T
01
(π ) ⎜ η ⎟⎜ k
⎝ p ⎠⎝ 01 ⎠
⎟
10. Find p = p ⎛⎜ T 2 ⎞⎟ ⎝ k 2 ⎠
⎜ −1 ⎟
2 02 ⎜ ⎟
⎝ T 02 ⎠ Fig. 9 Profile loss parameter with variation of Mach
• Calculation of static pressure and temperature at number [10].
stator outlet (P3, T3)
1. From EES find h03, S03 using P03, T03
2. Find h3 = h03 - C32/2 The profile loss parameter is expressed as
3. From EES find ρ3, Cp3, k3, μ3 using h3, S3 = S03
4. Find T3 = T03 - (C32/2 × Cp3) 2
⎛ k ⎞ ζ p
0.5 v12 cs(α 2) = f ( M 1 , D eq ) (36)
⎜ 3 ⎟
v2
⎛ ⎞ ⎜ k −1 ⎟
5. Find p = p ⎜ T 3 ⎟⎝ 3 ⎠
3 03 ⎜T ⎟ A fourth-order polynomial fitting method has been used to
⎝ 03 ⎠ interpret the graph that has the form
The above calculation process is repeated for each stage
f ( x) = a 0 + a1 x + .... + a n −1 x n −1 + a n x n
noting that:
The starting rotor inlet-conditions will have the same
velocity and radius outlet of the previous stage and the • The polynomial coefficients are listed below
stagnation properties is taken from the previous stage.
rm,1 = rm,3(i-1), Cm,1 = C m,3(i-1), α1 = α3(i-1)
M1 a0 a1 a2 a3 a4
0.3 -8.26097e-02 2.62982e-01 -2.66675e-01 1.14774e-01 -1.61839e-02
P01 = P03 (i-1), T01 = T03 (i-1), h01 = h03 (i-1), S01 = S03 (i-1)
0.7 -1.30107e-01 3.68490e-01 -3.56939e-01 1.48500e-01 -2.08264e-02
• Calculation of the pitch-chord ratio (s/c) 1.0 -1.36535e-01 3.78126e-01 -3.66336e-01 1.52219e-01 -2.13465e-02
2 2
C + 2 .5 C min + 0 .5 U 2 (40)
F ef =
4C 2
2
C min =
sin (α + β ), if (α + β ) ≤ 90 and β ≥ 0
2
2
C
2
C min = 1, if (α + β ) > 90
2
C
2 2
C min = U 2 if β < 0
2 2
C C
Tip
Clearance a0 a1 a2 a3 a4
0.0 3.23881e00 -3.66895e01 1.60855e02 -3.14825e02 2.32625e02 Figure.11 Diagram giving definition of Fef.
0.02 2.86933e00 -3.18679e01 1.36001e02 -2.58533e02 1.85224e02
Figure 12 shows a correlation of stalling pressure-rise
coefficient (CpD) and of diffusion-length to exit passage-
0.04 -2.00381e-01 1.04984e00 3.12191e00 -2.0345e01 2.48570e01
width (L / g2).
0.07 8.18792e-01 -8.62635e00 3.57996e01 -6.61454e01 4.61697e01
L σ (41)
=
0.1 2.38135e-01 -2.36201e00 1.01622e01 -1.92343e01 11.36794e01 g2 ( ) ⎛θ ⎞
cos β b 2 cos⎜ ⎟
⎝2⎠
⎛ Cp ⎞ ⎛ Cp ⎞ ⎛ Cp ⎞ (39)
Cp,max= CpDFef ⎜⎜ ⎟⎟ ⎜⎜ ⎟⎟ ⎜⎜ ⎟
⎟
⎝ CpD⎠Re⎝ CpD⎠ε ⎝ CpD⎠Δz Fig.12 Correlation of stalling pressure-rise
coefficient (CpD).
Where, Fef is the effective dynamic-pressure coefficient that
is represented by a weighted-average of the free-stream
7
A fifth-order polynomial fitting method has been used to
interpret the graph that has the form
f ( x) = a 0 + a1 x + .... + a n −1 x n −1 + a n x n
a0 a1 a2 a3 a4 a5
7.6431e-02 4.916e-01 -2.5166e-01 9.1688e-02 -1.9627e-02 1.7779e-03
g=
( ( )
π r m cos β b1 + cos β b2 ( )) (42) Fig. 14 Effect of axial-spacing on stalling pressure-rise
Z coefficient (CpD).
Where, Z denotes the number of blades in one row.
Figure 15 shows the effect of Reynolds number on stalling
pressure-rise coefficient (CpD).
• The polynomial coefficients are listed below
a0 a1 a2 a3 a4 a5
1.1191e00 -6.1567e-01 9.6073e-01 -2.2107e-01 -7.4519e-01 5.1421e-01
8
Ksh Blade Type
0.7 DCA
1.0 65-SERIES
1.1 C- SERIES
2 3
⎛t⎞ ⎛t⎞ ⎛t⎞ (45)
k it = −0.0214 + 19.17⎜ c ⎟ − 122.3 ⎜ ⎟ + 312.5 ⎜ ⎟
⎝ ⎠ ⎝c⎠ ⎝c⎠
9
II.2 Design Optimization of Axial-Flow Compressor By applying Buckingham's π-theorem and applying
dimensional analysis, Eq. 50 may be simplified to the
Once we get the geometry supplied by the program of
following dimensional form:
the axial-flow compressor design-point efficiency, its
geometry is altered to maximize efficiency at the design- ⎛ ⎞
η tt = f ⎜⎜ φ ,ψ , R , w1 , c 2 , M 1 , M 2 , R em , ζ , ζ s ⎟⎟ (51)
point. The structure of the optimizing program is shown in ⎝ U U R
⎠
Fig. 17. Where M1 and M2 are the rotor and stator exit mach
numbers that are defined as:
W1
M1=
a1
C2 (52)
M2 =
a2
Rem is the stage Reynolds number based on mean radius
U rm (53)
Rem = υ
ζ R ζs
, are the rotor and stator loss coefficients expressed
in terms of the exit velocities C2 and w3 relative to the blade
rows.
Δp
ζ R = 1 oR
ρ W 12
2
Δ pos (54)
ζs= 1
ρ C 22
2
Ψ= Δ ho (55)
U
Fig.17 Structure of the optimizing program of the axial- • Independent design variables
flow compressor.
The designer is free to select the design duty
By applying dimensional analysis for a single stage and coefficients ( φ , ψ). As these duty coefficients have a
making the following assumptions: profound effect upon the stage efficiency ηtt even with
optimum aerodynamic design. φ and ψ control the shape of
1. Constant axial velocity Cx.
2. Constant mean radius rm = 1/2(rh + rt). the velocity triangles and thus the flow environment within
3. Identical velocity vectors C1 and C3 at entry to and exit which the blades operate. Also, the degree of reaction (R)
from the stage at the mean radius rm. has a direct control over velocity triangle shape and hence
efficiency.
The efficiency ηtt of this stage is dependent upon the
following variables, [9]: • Dependent design variables affecting (ηtt)
⎛ Δ h o , h1, h 2 , h 3 ω , r m , c x , w 3 ,⎞ (50)
η = f⎜ ⎜c2,μ,ρ,a
⎟
Experimental cascade tests show that the loss
tt
⎝ 2 ,a3,Δ p or , Δ p os ⎟⎠
coefficients ζ R , ζ s are themselves dependent upon blade
• Thermodynamic variables. The stage stagnation enthalpy-
row Reynolds number and inlet Mach number. We would
rise Δho determines the specific work input and signifies
also expect that loss levels to be directly influenced by the
stage loading. The specific enthalpies h1, h2 and h3 typify
velocity triangle environment within which the blades have
the progression in energy transfer through the stage. All
to operate and hence to depend upon φ , ψ and R. We can
four are independent variables.
• Speed and size. Both are independent variables. express this through [9],
• Velocity triangles. Four velocities are required to = f 1 (φ ,ψ , R , Re R , M 1)
ζ
determine the shape of the velocity triangles these are the R
10
Equation 51now is simplified into
η tt =1−
1 ⎛ 2 1
⎜φ + (1+ψ )2 ⎞⎟ (ζ R + ζ s ) (64)
η tt = f (φ ,ψ , R , , ζ R , ζ s )
2ψ ⎝ 4 ⎠
(58)
Equation 64 is equivalent to the parametric Eq. 58
Thus, the efficiency of an axial-compressor stage
derived from the dimensional analysis for a 50% reaction.
depends upon five dimensionless parameters which are
But it is in the much more useful explicit form of an
sufficient to account for all the 15 items listed in Eq. 50Of
analytical relationship which shows how ηtt depends upon
these parameters, just three may be independently selected
the various dimensionless groups. From this, we can deduce
by the designer, namely φ , ψ and R. The loss coefficients
that the efficiency of a 50% axial-compressor stage is
themselves are also dependent upon the duty parameters φ , dependent upon two main factors:
ψ and R but in addition are influenced by Reynolds number
1. The stage duty coefficients ( φ , ψ).
and Mach number.
2. The blade-row loss-coefficients ζ R and ζ s (i.e., blade-
• Simple analytical formulation for the total to total row aerodynamic).
efficiency of a compressor stage The initial selection of the stage duty coefficients ( φ ,
Equation 58 can be converted into a more useful ψ) is crucial. Thus, we could rewrite Eq. 64in the form
η tt = 1 − f c (φ ,ψ )(ζ R + ζ s )
analytical form. By assuming for the moment a fixed
(65)
reaction value R = 0.5. From h0 – S diagram, Fig. 18, by
defining the stagnation enthalpy loss due to irreversibility
into Where, the loss-weighting coefficient ( fc) is given by
1 ⎛ 2 1
f c (φ ,ψ ) = ⎜φ + 1+ψ
2ψ ⎝ 4
( )2 ⎞⎟ (66)
⎠
(Δ p0)loss = 1 2 2
⎛ w1 ⎞ 1 ⎛ C2 ⎞ ⎛⎜ ζ R + ζ s ⎞⎟⎛ 2 1 2⎞ (63) Eq. 68 is consistent with Eq. 61 as a result of linking stage
ζ R⎜ ⎟ + ζ s⎜ ⎟ = ⎜ ⎟⎜⎝φ 4 (1+ψ ) ⎟⎠
+
ρU2 2 U
⎝ ⎠ 2 U
⎝ ⎠ ⎝ 2 ⎠ duty ( φ , ψ) and reaction R to velocity triangles and thus to
stage aerodynamics and thermodynamics.
Introducing Eq. 63 into Eq. 61,we have
11
• Optimum reaction III. Results and Discussions
III.1 Results of Design Program
For any prescribed ( φ , ψ) duty, we may estimate the
stage reaction R, which will produce maximum efficiency. At first we summarize the main steps in the design
Eq. 68 can be written as. procedure described in the analysis section. Having made
appropriate assumption about the axial velocity, it is
η tt = 1 − L (69)
possible to calculate the annulus area at the inlet and outlet
Where, L is given by of the compressor and calculate the air angles required for
each stage at the mean diameter. Then, by the use of vortex
⎧ ⎡
ψ ⎞ ⎤⎥ ⎡
ψ ⎞ ⎤⎥ ⎫⎪
2 2
1 ⎪ ⎢ 2 ⎛ 2 ⎛ (70) theory, the air angles can be calculated at various radiuses
L= ⎨ζ φ ⎜+ R + + ⎢
⎟ ζ s ⎢φ ⎜+ 1− R + ⎟ ⎬
2ψ ⎪ R ⎢ 2 ⎠ ⎥⎦ 2 ⎠ ⎥⎦ ⎪ from root to tip. Throughout this work, there was a
⎩ ⎣
⎝ ⎣ ⎝ ⎭ limitation on blade stresses; rates of diffusion and Mach
number that were not exceeded. The results of the program
The minimum loss and therefore maximum efficiency were validated using the data of Ref. [6]. The compressor
with respect to reaction R follows from of the present study is a 15-stage axial compressor with
⎧ ∂L ⎫ (71) 122 kg/s of air at ambient pressure of 1.013 bar and
=0 ⎨ ⎬
⎩ ∂R ⎭φ ,ψ temperature 288 K, pressure ratio 20 and polytropic
efficiency of 90%. Tables 1-5 show the present calculated
Where φ and ψ are kept constant. If we assume that the values and the relative differences in comparison to the data
of Ref. [6]. There is a good agreement as far as dimensions
loss coefficients are weak function of R and may be are concerned and a reasonable agreement in the other
assumed constant also, then Eq. 71 yields to parameters. This may be attributed to some difference in
design assumptions. The biggest differences between the
ψ
(ζ s − ζ R )+ ζ s present results and those of Ref. [6] are noticed in Table 5.
R optimum = 2 (72) The difference in the inlet stator angle may reach about 7%.
ζ s +ζ R Figures 19 and 20 show the rise of both the static
One possible solution to this which is true for all values pressure and temperature through the compressor stages,
of ψ is R = 0.5 and ζ s = ζ R . Although the stator and rotor respectively. Figures 21-23 show the variation of air angles,
degree of reaction, rotor/stator exit Mach number from root
velocity triangles are identical for this condition of 50%
to tip for a selected stage (stage 10). In Fig. 21, the radial
reaction, in reality there will be a difference in the two loss
variation of air angles of the rotor shows a change in fluid
coefficients. Even so the strong indication is that 50%
deflection for a considerable twist along the blade height to
reaction will be close to optimum [9].
ensure that the blade angles are in agreement with the air
• Optimum ψ for a given φ and R angles. In Fig. 22, the stator deflection is less in comparison
to the rotor deflection due to the nature of building-up
Alternatively, we may search for ψ value leading to pressure in stator blades. In Fig. 23, the degree of reaction
minimum loss for given φ and R values by writing increases from root to tip, which indicates a high mass
⎧ ∂L ⎫ (73) flow-rate per unit blade-height and thus plays an important
⎨ ⎬ =0 role in to raising the stage efficiency. Figures 24 and 25
⎩ ∂ψ ⎭φ , R
show the rotor and stator end wall, profile and total losses.
Resulting in Profile losses are contributed to boundary-layer separation
1 (74) while end-wall losses are mainly due to secondary flow
ψ φ + R (R − 1 )
2
= 2 +
optimum 2
effects and mixing for the rotor. The profile and end-wall
Where, we have also assumed that ζ s and ζ R are losses increase through the stages with the result of an
independent of ψ. Two stages of special interest are the increase in total losses due to the increase in the work that
50% reaction stages which we have already considered and is required to accomplish fluid turning and raising the
the 0% reaction or "impulse stages". For these two pressure through different stages as well as the generation
reactions Eq. 74 becomes of entropy. At the stator, the end-wall and profile losses
2 (75) decrease, resulting in a decrease in total losses due to the
ψ = 4φ + 1 optimum
diffusing working nature of the stator blades.
2
ψ optimum = 4φ + 2 Table 1: Comparison of the present (AFCP) results and
the data of Ref. [6] for the compressor tip radius.
In practice the stator and rotor loss coefficients ζ s and
ζ R do vary with both φ and ψ and form experimental tests Compressor Tip Radius (mm)
Present
the sensible design for ψ lies between the two values. Stage Row
(AFCP)
Ref. [6] Diff. (%)
2 (76)
ψ = 0 .185 4 φ + 1
opt ., exp 1 0.524 0.528 0.773
ψ max = 0 .32 + 0.2φ 1 2 0.513 0.513 0.024
3 0.505 0.507 0.416
1 0.505 0.507 0.416
2 2 0.496 0.5 0.799
3 0.490 0.494 0.749
3 1 0.490 0.494 0.749
12
2 0.483 0.489 1.187 3 0.362 0.372 2.763
3 0.479 0.484 1.023 1 0.362 0.372 2.763
1 0.479 0.484 1.023 7 2 0.367 0.375 2.247
4 2 0.473 0.480 1.405 3 0.369 0.378 2.545
3 0.470 0.476 1.221 1 0.369 0.378 2.545
1 0.470 0.476 1.221 8 2 0.373 0.380 1.982
5 2 0.471 0.474 0.66 3 0.374 0.383 2.402
3 0.468 0.470 0.357 1 0.374 0.383 2.402
1 0.468 0.470 0.357 9 2 0.377 0.385 2.008
6 2 0.464 0.467 0.665 3 0.378 0.387 2.250
3 0.462 0.464 0.441 1 0.378 0.387 2.250
1 0.462 0.464 0.441 10 2 0.381 0.389 1.986
7 2 0.458 0.462 0.83 3 0.382 0.390 2.032
3 0.457 0.460 0.723 1 0.382 0.390 2.032
1 0.457 0.460 0.723 11 2 0.385 0.392 1.871
8 2 0.453 0.457 0.784 3 0.385 0.393 1.967
3 0.452 0.455 0.599 1 0.385 0.393 1.967
1 0.452 0.455 0.599 12 2 0.388 0.394 1.633
9 2 0.449 0.453 0.79 3 0.388 0.395 1.768
3 0.449 0.452 0.768 1 0.388 0.395 1.768
1 0.449 0.452 0.768 13 2 0.390 0.396 1.501
10 2 0.446 0.450 0.884 3 0.391 0.397 1.664
3 0.445 0.449 0.816 1 0.391 0.397 1.664
1 0.445 0.449 0.816 14 2 0.392 0.397 1.201
11 2 0.443 0.448 1.094 3 0.393 0.398 1.387
3 0.443 0.447 0.991 1 0.393 0.398 1.387
1 0.443 0.447 0.991 15 2 0.394 0.398 0.975
12 2 0.441 0.446 1.215 3 0.394 0.399 1.177
3 0.440 0.445 1.084
1 0.440 0.445 1.084 Table 3: Comparison of the present (AFCP) results and
13 2 0.438 0.444 1.265 the data of Ref. [6] for root-mean-square radius.
3 0.438 0.443 1.111
1 0.438 0.443 1.111 Compressor Root Mean Square Radius (mm)
14 2 0.437 0.443 1.48 Stage Row
Present
Ref. [6] Diff. (%)
3 0.436 0.442 1.308 (AFCP)
1 0.436 0.442 1.308 1 0.415 0.421 1.445
15 2 0.435 0.441 1.644 1 2 0.415 0.421 1.445
3 0.435 0.441 1.457 3 0.415 0.421 1.445
1 0.415 0.421 1.445
Table 2: Comparison of the present (AFCP) results and 2 2 0.415 0.421 1.445
the data of Ref. [6] for the compressor hub radius. 3 0.415 0.421 1.445
1 0.415 0.421 1.445
Compressor Hub Radius (mm) 3 2 0.415 0.421 1.445
Present 3 0.415 0.421 1.445
Stage Row Ref. [6] Diff. (%)
(AFCP) 1 0.415 0.421 1.445
1 0.264 0.274 3.614 4 2 0.415 0.421 1.445
1 2 0.285 0.300 5.142 3 0.415 0.421 1.445
3 0.299 0.310 3.604 1 0.415 0.421 1.445
1 0.299 0.310 3.604 5 2 0.415 0.421 1.445
2 2 0.314 0.321 2.329 3 0.415 0.421 1.445
3 0.323 0.332 2.923 1 0.415 0.421 1.445
1 0.323 0.332 2.923 6 2 0.415 0.421 1.445
3 2 0.333 0.339 1.792 3 0.415 0.421 1.445
3 0.339 0.346 2.065 1 0.415 0.421 1.445
1 0.339 0.346 2.065 7 2 0.415 0.421 1.445
4 2 0.347 0.351 1.159 3 0.415 0.421 1.445
3 0.351 0.357 1.664 1 0.415 0.421 1.445
1 0.351 0.357 1.664 8 2 0.415 0.421 1.445
5 2 0.350 0.360 2.767 3 0.415 0.421 1.445
3 0.354 0.364 2.905 1 0.415 0.421 1.445
1 0.354 0.364 2.905 9 2 0.415 0.421 1.445
6
2 0.359 0.368 2.365 3 0.415 0.421 1.445
13
1 0.415 0.421 1.445
10 2 0.415 0.421 1.445
3 0.415 0.421 1.445 Table 5: Comparison of the present (AFCP) results and
1 0.415 0.421 1.445 the data of Ref. [6] for the compressor inlet blade angle.
11 2 0.415 0.421 1.445
3 0.415 0.421 1.445 Inlet Blade Angle
1 0.415 0.421 1.445 Present
Stage Row Ref. [6] Diff. (%)
12 2 0.415 0.421 1.445 (AFCP)
3 0.415 0.421 1.445 Rotor 51.275 54.542 6.372
1 0.415 0.421 1.445 1
Stator 46.29 49.051 5.965
13 2 0.415 0.421 1.445 Rotor 52.625 53.109 0.919
3 0.415 0.421 1.445 2
Stator 48.942 51.052 4.311
1 0.415 0.421 1.445
14 2 0.415 0.421 1.445 Rotor 53.896 54.290 0.732
3
3 0.415 0.421 1.445 Stator 49.367 52.069 5.473
1 0.415 0.421 1.445 Rotor 55.095 55.395 0.544
15 2 0.415 0.421 1.445 4
Stator 49.721 53.026 6.647
3 0.415 0.421 1.445 Rotor 57.325 56.448 -1.531
5
Stator 50.253 53.722 6.903
Table 4: Comparison of the present (AFCP) results and
the data of Ref. [6] for the De Haller Parameter. Rotor 57.325 57.187 -0.242
6
Stator 51.253 54.356 6.054
De Haller Parameter Rotor 58.370 57.864 -0.867
7
Stage Row
Present
Ref. [6] Diff. (%) Stator 52.443 54.932 4.746
(AFCP) Rotor 59.375 58.484 -1.500
Rotor 0.739 0.739 0.000 8
1 Stator 54.589 55.455 1.586
Stator 0.77 0.808 4.935
Rotor 60.342 59.053 -2.136
Rotor 0.734 0.744 1.362 9
2 Stator 54.693 55.925 2.253
Stator 0.765 0.752 -1.699 Rotor 61.275 59.575 -2.774
Rotor 0.73 0.744 1.918 10
3 Stator 55.758 56.686 1.664
Stator 0.761 0.752 -1.183
Rotor 62.174 60.238 -3.114
Rotor 0.726 0.745 2.617 11
4 Stator 55.784 57.396 2.890
Stator 0.756 0.753 -0.397
Rotor 63.042 60.962 -3.300
Rotor 0.721 0.746 3.467 12
5 Stator 55.772 58.155 4.273
Stator 0.751 0.750 -0.133 Rotor 63.880 61.752 -3.331
Rotor 0.717 0.741 3.347 13
6 Stator 55.722 58.962 5.815
Stator 0.746 0.745 -0.134
Rotor 64.688 62.613 -3.208
Rotor 0.717 0.737 2.789 14
7 Stator 58.632 59.821 2.028
Stator 0.746 0.740 -0.804
Rotor 65.468 63.545 -2.937
Rotor 0.713 0.732 2.665 15
8 Stator 58.504 59.314 1.385
Stator 0.741 0.735 -0.810
Rotor 0.709 0.727 2.539
9
Stator 0.736 0.730 -0.815
Rotor 0.705 0.723 2.553
10
Stator 0.73 0.728 -0.274
Rotor 0.701 0.722 2.996
11
Stator 0.725 0.726 0.138
Rotor 0.697 0.721 3.443
12
Stator 0.72 0.723 0.417
Rotor 0.693 0.719 3.752
13
Stator 0.714 0.720 0.840
Rotor 0.69 0.717 3.913
14
Stator 0.709 0.716 0.987
Rotor 0.686 0.715 4.227
15
Stator 0.703 0.700 -0.427
14
Fig.19 Static pressure rise through the compressor.
Pressure Efficiency
Case Ψ
Ratio (%)
Max. 0.45 23.02 86.92
Optimum 0.3 15.35 70.86
Design 0.391 20 81.14
Arbitrary 0.35 17.9 76.73
16
Fig.33 Koch surge limit for the compressor, ψ = 0.45.
Fig.30 Koch surge limit for the compressor, ψ = 0.3.
Fig.32 Koch surge limit for the compressor, ψ = 0.391. Fig.35 Rotor camber variation for different values of
work coefficient.
17
Fig.39 Rotor-deviation variation for different values of
Fig.36 Stator camber variation for different values of the work coefficient.
the work coefficient.
20