JKAU: Eng. Sci., Vol. 24 No. 1, pp: 73-88 (2013 A.D. /1434 A.H.) DOI: 10.4197 / Eng. 24-1.4 73 Centrifugal Compressor Map Prediction and Modification N.N. Bayomi *,** , R.M. Abdel-Maksoud ** and M.I.F. Rezk *** * King Abdulaziz University, Jeddah, Saudi Arabia, and ** Mech. Power Dept., Faculty of Eng., Mataria, Helwan University, and *** Elsewedy for wind energy generation-Elsewedy Electric, Cairo, Egypt [email protected]Abstract. Centrifugal compressors are utilized in various fields and are used in vast applications. Their operational performance maps are significant to be studied, modified and enhanced. Unfortunately, such maps that describe experimental results do not cover each condition. This is due to expenses as well as the uncovering operational zones. Therefore, map prediction is important, however, it is very complex because of its nonlinearity as well as unstable region that are not easily to be assigned practically. Consequently, the present paper introduced a methodology that predicts the centrifugal compressors performance maps specified at stable and unstable conditions. Enhancement and modification of the compressor performance map is performed using the closed coupled valve and variable drive speed where the later method was more preferable based on shifting of the compressor map towards lower flow rate with less pressure drop. Keywords: Centrifugal compressor, Compressor map, Rotating stall, surge, Choke line. 1. Introduction Centrifugal compressors have been used in vast industrial applications. Knowledge of their operational performance maps is significant; however such maps do not cover all the conditions due to expenses. Therefore, map prediction is important, however, it is very complex because of its nonlinearity as well as unstable regions that are not easily to be assigned practically. From this insight, many researches adopte to predict its performance. As a result of these studies the empirical loss correlation method had been persistently developed by several
16
Embed
Centrifugal Compressor Map Prediction and Modification
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Centrifugal compressors have been used in vast industrial applications. Knowledge of their operational performance maps is significant; however such maps do not cover all the conditions due to expenses. Therefore, map prediction is important, however, it is very complex because of its nonlinearity as well as unstable regions that are not easily
to be assigned practically. From this insight, many researches adopte to predict its performance. As a result of these studies the empirical loss correlation method had been persistently developed by several
N.N. Bayomi et al. 74
researchers[1-2]
. On the other side, map performance is restricted by high flow rate limits denoted by choke line. Choke line was determined by Dixon
[3]. Furthermore, compressor performance is limited by small flow rates where operational instability occurs that are rotating stall and surge that are vastly studied by researchers[4-6]
. Compressor surge control was introduced by other researchers
[7-13].
The present work aimed to predict the stable and unstable compressor performance map and accounts for compressor losses. This is achieved by introducing a methodology. This is determined by pre-matching of the simulated actual with the experimental results to account for different losses represented by previous empirical formulas. Therefore, the uncovered zones in compressor map can be predicted. To estimate the choke line, the present model utilizes the formula of Dixon
[3] for chocking at the diffuser. In addition, the stall line and surge line are determined using the local stability method. The present work is extended to avoid compressor instability by close coupled valve and variable drive speed methods.
2. Methdology for Compressor Map Prediction
In this section, performance map prediction and modifications are demonstrated. Performance map prediction is determined by the actual Euler head at different speeds, the choke line and the instability lines. On the other side, performance modification is attained using two types of controllers that are the closed coupled valve and variable speed drive.
2.1 Performance Map Prediction
Foremost, theoretical Euler’s head should be determined. The theoretical Euler head can be written as:
th 2 u2 1 u1H U C UC= − (1)
The velocity triangles at the inlet and exit of typical centrifugal compressor impeller impeller is shown in Fig. 1.
Fig. 1. Velocity triangles for compressor impeller: a) Inlet velocity triangle, b) Outlet
velocity triangle.
Centrifugal Compressor Map Prediction and Modification 75
All symbol definitions in the different equations are listed in the nomenclature. The velocity triangles at the inlet and exit can be identified using the data of the impeller dimension, the rotational speed. Air density at the compressor inlet and exit is estimated using equation of state, the ambient conditions and the blade dimensions. The Slip factor used for velocity triangle calculation is specified by the following equation introduced by Stanitz
[14]:
( )0.631
z
⎛ ⎞× πσ = −⎜ ⎟
⎝ ⎠ (2)
Consequently, the theoretical Euler’s Head can be calculated. The actual Euler’s head at different condition can be specified and is given by the following equation:
act th lossH H L= − (3)
In order to assign actual Euler head, eight common different head losses, Lloss, will be estimated from the Table 1 by using the selection losses equations from Oh et al.[2]. The ranges of the coefficients of these equations are specified in Table 2. Substituting the certain values of these coefficients is accomplished using trial and error till matching between the actual Euler head and the experimental results will be performed. Consequently, the uncovered zones in compressor map can be assigned. Since the compressor operational condition is characterized principally
by the efficiency therefore, it is necessary to estimate the efficiency at
different conditions. The efficiency of the compressor can be defined by:
th
th loss
H
H Lη =
+ (4)
Using actual Euler’s head to get pressure ratio
( )1th tot
p 1
H L1
C T
γ
γ−⎡ ⎤⎛ ⎞−π = +⎢ ⎥⎜ ⎟⎜ ⎟η⎢ ⎥⎝ ⎠⎣ ⎦
(5)
N.N. Bayomi et al. 76
Table 1. Losses description equations for centrifugal compressor.
Compressor losses Loss model
Blade Loading Loss
( )2
2 1
2
22 2
2
1 1 11
2 2 2
11
21 2
⎛ ⎞−⎜ ⎟⎜ ⎟
× − + ×⎜ ⎟⎡ ⎤⎡ ⎤⎜ ⎟− +⎢ ⎥⎢ ⎥⎜ ⎟⎣ ⎦⎣ ⎦⎝ ⎠
p
BL
t t
C T TK
W UU
W D DW z
U D Dπ
Incidence Loss 2
1
2
u
mc
WF
Impeller Disk Friction
Loss
3
52
2 2
2
0.2
2 2 2
2
0.0402
4
Ur
r
U r
m
ρ
ρ
μ
⎛ ⎞⎜ ⎟⎝ ⎠
⎛ ⎞⎜ ⎟⎝ ⎠
Skin Friction Loss
2
2
1 2
22
2
f
Dw
C WD D
⎛ ⎞⎛ ⎞+⎜ ⎟⎜ ⎟⎝ ⎠⎝ ⎠+⎛ ⎞
⎜ ⎟⎝ ⎠
Clearance Loss
( )
2
2 2
1 1
2 2
22 22 1
1
40.6
1
t h
t
r rW W
b b Zr r
ε π
ρ
ρ
⎡ ⎤⎡ ⎤⎢ ⎥⎢ ⎥−⎢ ⎥⎢ ⎥
⎛ ⎞⎢ ⎥⎢ ⎥− −⎜ ⎟⎢ ⎥⎢ ⎥⎝ ⎠⎣ ⎦⎣ ⎦
Leakage Loss 2
2
cl clm U U
m
Recirculation Loss
( )2
2 1
2
22 2
2 2
1 1 11
2 2 2
0.02tan 1
1 2
⎛ ⎞−⎜ ⎟⎜ ⎟
− +⎜ ⎟⎡ ⎤⎡ ⎤⎜ ⎟− +⎢ ⎥⎢ ⎥⎜ ⎟⎣ ⎦⎣ ⎦⎝ ⎠
p
BL
t t
C T TK
W UU
W D DW z
U D D
α
π
Table 2. Coefficients used in present methodology.
Coefficient Value Unit
KBL 0.75 or 0.60 for conventional or splitter
impellers, respectively
-----
Fmc 0.7 -----
Cf 0.004 -----
ε 0.0038 m
lc 1 -----
The choking of mass flow can be expressed by the famous equation
of Dixon[3]
:
Centrifugal Compressor Map Prediction and Modification 77
( )( )
( )
( )
2 1
2 12
2 1o1
o1 o1 12 2
22
o1
U1 1
am 2a
A 1U
1 1a
γ
γ−
γ+•
γ−
⎡ ⎤ησ+ γ −⎢ ⎥
⎛ ⎞⎣ ⎦= ρ ⎜ ⎟γ +⎝ ⎠⎡ ⎤σ+ γ −⎢ ⎥
⎣ ⎦
(6)
Since the above equation represents a theoretical relationship
between the choke line flow rate and the different parameters, a new
treatment is herein presented to suit the actual prediction. This is
performed where γ is replaced by the polytropic index that equals to 1.2
as a correction in order to be suitable for precise prediction.
In order to determine the stall and surge line local stability analysis is
used. This method is used extensively by previous researchers such as
Abed El-Maksoud[8]
. The local stability analysis method is herein used to
assess the system whether the system is stable (rotating stall) or unstable
(surge) at the left of the peak. Regarding the stability condition, stall
point is defined as stable condition, since the characteristics will be
asymptotically stable in low-flow small pressure rise region. Stall line
can be predicted just on the left the characteristic peaks. In case of surge,
the flow coefficient and pressure fluctuates with certain amplitudes and
such phenomenon is unstable. The local stability analysis method
depends on the roots of the Jacobean matrix of Moore and Greitzer
model[4]
system description state equations. The two state equations of
the Moore - Greitzer model are:
( )( )c
c
d 1
d l
ϕ= ψ ϕ −ψ
ζ and ( )2
c
d 1
d 4B l
ψ= ϕ − Γ ψ
ζ (7)
The following equation determines the compressor map that is
defined by a fifth order polynomial of the flow coefficient, ϕ:
( ) 4 3 2
c 1 2 3 4 5C C C C CΨ ϕ = ϕ + ϕ + ϕ + ϕ + (8)
Applying stability condition on Moore and Greitzer model, hence
system stalls or surges could be assigned. Stability analysis has been
implemented by several researchers to analyze the compression system.
The following equations present the treatment of stability analysis
method. Therefore, the Jacobean matrix of the Moore and Greitzer model
will be:
N.N. Bayomi et al. 78
( )c
c c
2 2
c c
d1 1
l d l
1
4 l 2 l
⎛ ⎞Ψ ϕ−⎜ ⎟
ϕ⎜ ⎟⎜ ⎟Γ
−⎜ ⎟⎜ ⎟β β ψ⎝ ⎠
(9)
For stalling condition:
( )c
2
c c
d10
l d 2 l
⎛ ⎞Ψ ϕ Γ− <⎜ ⎟
ϕ β ψ⎝ ⎠ and
( )
( )
2
c
2
c c
c
2 22
c cc
d1
l d 2 l
d1 14
l d 4 l2 l
⎛ ⎞Ψ ϕ Γ−⎜ ⎟⎜ ⎟ϕ β ψ⎝ ⎠
⎛ ⎞⎛ ⎞Ψ ϕ Γ< − +⎜ ⎟⎜ ⎟⎜ ⎟⎜ ⎟ϕ ββ ψ⎝ ⎠⎝ ⎠
(10)
Otherwise, the system surges.
2.2 Performance Map Modification
The first controller introduced herein is the closed coupled valve,
CCV. It is named so as the valve is installed very close and coupled to
the compressor. According to Gravdahl[6]
, close coupled valve is to be
understood that the distance between the compressor outlet and the valve
is so small that no significant mass storage can take place. The
assumption of no mass storage between the compressor and the valve
allows for the definition of an equivalent compressor. The closed coupled
valve is introduced into the compression system to achieve a pressure
drop on the compressor pressure rise. Thus, reduces the characteristic
peak and shifts it towers lower flow rates, thus avoiding falling into
surge. Consequently the second state equation of model Eq. 7 can be
rewritten as:
( )ccv2
c
d 1
d 4B l
ψ= ϕ − Γ ψ − Γ ψ
ζ (11)
The second controller introduced in this sub-section is the variable
speed drive. From the findings of Gravdahl[6]
, Variable speed drive is
herein used as a second method to achieve compression system
stabilization. This is approach is used as active surge control. The use of
Centrifugal Compressor Map Prediction and Modification 79
the rotational speed as a control variable renders the equilibrium globally
exponentially stable and the use of the drive speed as control ensure
exponential convergence. The control manipulates the compressor map in
such a way that the compressor map is shifted to the left with lower flow
rates. The equation that describes this type of controller can be written as:
( ) ( ) ( ) ( ) ( ) ( )4 3 2
c 1 2 3 4 5N C N C N C N C N C NΨ = ϕ + ϕ + ϕ + ϕ + (12)
In the above equation the pressure coefficient, Ψc , is dependent on
the speed, N, that varies according to the onset of instability. The
controller is used to reduce the speed of the compressor so that the peak
of the performance characteristic is lowered and shifted towards the
lower flow rates similarly to the closed coupled valve. This behavior
avoids falling into surge. The following section deals with the results of
the present model and the assessment of the two controllers.
3. Results and Discussion
In order to determine the uncovered zones in the compressor map,
the present model results are matched to the experimental results. The
comparison is herein performed using three different ratios of Eckardt[7]
denoted in this finding by rotor A, B, and O and rotor of Bayomi[15]
.
Foremost, in order to conduct simulations, different variables of the
present model should be specified to estimate the different losses in the
present model. The ambient temperature and pressure are 287 K and 1
Bar, respectively. The specific heat at constant pressure, specific heat
ratio and gas constant for air are taken to be equal to 1005 J/kgK and
1.333 and 287 J/kgK, respectively. Using the experimental maps of these
rotors and their data of these rotors to matching these maps with the
present model to find out different uncovered zoned. Tables 3 and 4 show
the three Eckardt rotors and Bayomi rotor geometrical parameters,
respectively. Table 5 illustrates the performance experminal data of
Eckardt and Bayomi rotors.
Figure 2 demonstrates the results of model pre-matching with the
experimental results for Eckardt rotors A for different four speeds.
Consequently, the losses considered in Table 1 are valid at these speeds.
This makes the losses different rotational speeds could be determined.
Hence, the compressor stable operation at different speeds that is not
covered by the experimental results could be predicted.
N.N. Bayomi et al. 80
Table 3. Eckardt rotors geometrical parameter.
Geometrical Parameter Eckardt
rotor A
Eckardt
rotor B
Eckardt
rotor O
Inlet Tip Diameter mm 280 280.3 280
Inlet Hub Diameter mm 120 191.8 90
Discharge Diameter mm 400 400 400
Discharge Width mm 26 26 26
Number of Blades 20 20 20
Length in axial direction mm 130 84.2 130
Blade Thickness mm 3 3 3
Inlet Blade Angle 30 30 30
Exit Blade Angle 30 40 90
Maximum Rotational Speed RPM 16000 16000 16000
Table 4. Bayomi rotor geometrical parameter.
Geometrical Parameters
Impeller outer diameter mm 160
Inducer tip diameter ratio 0.70
Inducer hub diameter ratio 0.2375
Inducer tip diameter mm 112
Inducer hub diameter mm 38
Exit width ratio 0.0766
Blade thickness ratio 0.0163
Impeller discharge width mm 12.256
Impeller blade thickness mm 2.608
Exit blade angle 60
Inducer tip angle 60
Inducer hub angle 40
Number of blades (7 splitter blades) 7
Design speed rpm 55000
After pre-matching is achieved, Fig. 3 illustrates Euler’s head and
total predicted losses for Eckardt rotor B at different operating speed in-
between stalling point the choking point. Total losses plotted in this
figure could be utilized to determine the compressor efficiency. As one
may observe that total losses increase with the rotational speed and mass
flow rate.
After estimating the different losses, it is principally necessary to
estimate the compressor efficiency at different speeds. Efficiency of
Bayomi rotor is assigned by plotting the simulated results in Fig. 4 using
Eq. 4. The global observation is that the efficiency values appear to be
higher with the increase of rotational speed. Furthermore, efficiency
increases with reduction of mass flow rate till or at least near the peak of
map characteristic.
Centrifugal Compressor Map Prediction and Modification 81
Table 5. Data for Eckardt's Bayomi's experimental data.