THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

345 E. 47 St., New York. N Y 10017 85-GT-97

The Society shall not be responsible for statements or opinions advanced in papers or In
discussion at meetings of the Society or of its Divisions or Sections, or printed in its
publications. Discussion is printed only if the paper is published in an ASME Journal.
Released for general publication upon presentation. Full credit should be given to ASME,
the Technical Division, and the authods). Papers are available from ASME for nine months
after the meeting.
Printed in USA. Copyright © 1985 by ASME

A Study Concerning Performance Characteristics of

Centrifugal Compressors
SHIMPEI MIZUKI
Associate Professor
Department of Mechanical Engineering
College of Engineering
Hosei University
Koganei, Tokyo
Japan

HIKARU IMAI
Engineer
Fuji Heavy Industry
Mitaka, Tokyo
Japan
ABSTRACT stalled flow at the impeller and the diffuser inlet at the low flow
A method for estimations of performance characteristics of rate. It would be impossible to predict analytically the losses within
centrifugal compressors was examined by comparing the estimated the channels caused by these phenomena with enough accuracies.
results with those by experiments. Lots of empirical and semi-empiri- Thus, the design techniques which depend on the experiences of the
cal factors which were introduced to the losses within compressor designers play an important role for the estimations of the losses. If
channels in the estimations, and the experimental results for the the predicted performance characteristics for a family of compressors
compressors with a wide variety of configurations, operating condi- give good agreements with the measured results and the method,
tions and design techniques were investigated. After the extensive however, results in the disagreements to the other new families,
comparisons, the most appropriate combinations of the factors for the above-mentioned empirical or semi-empirical factors are modified
the present method were given. or replaced by more developed ones to get the similar performance
By employing this method, the major factors which are directly characteristics as those by experiments.
related to the performance characteristics were computed and the From the above-mentioned viewpoint, a simplified method
relationships among them were discussed. Thus, the ratio of the commonly used to estimate the performance characteristics of
relative velocity at the impeller exit to that at the inlet, the non- centrifugal compressors was applied to a wide variety of compressors
dimensional relative velocity at the impeller inlet and that at the designed by the different techniques [3, 11-17]. By changing many
exit were obtained as the parameters of the slip factor, the flow semi-empirical and empirical factors for this method which were
coefficient and the specific speed. already reported in the literatures [1-3, 8-10], the computations
were made in order to compare the results with those by experiments.
INTRODUCTION After the extensive comparisons, the most appropriate combination
Centrifugal compressors have been used as industrial, aeronautical, of the factors for the present estimation method was obtained. Based
vehicular applications and so on under a wide variety of operating on this method, the major relationship between non-dimensional as
conditions and with many different configurations. well as dimensional parameters. which are directly related to the
At the first stage of the design of the compressors, the major performance characteristics. such as, all components of the relative
configurations are estimated according to their specifications. Based and the absolute velocities, the Mach numbers, the total and the
on the leading dimensions, the performance characteristics are static pressure, the specific speeds. the efficiencies and so on, were
analyzed to check and modify the configurations in detail [ 1-3] closely examined.
under the required conditions. After that, the two-, quasi-three or
three-dimensional flow analysis within the compressor channel with NOMENCLATURE
or without the losses [4-7] is made in order to obtain the appropriate A = area or constant in Eqs. (5.1) and (5.2)
flow pattern with the highest efficiency. When the modifications of B2 = blade height at impeller exit
the configurations are again necessary, the above-mentioned design Cf coefficient of skin friction
procedures will be repeated. Cp = specific heat
Although these design procedures employ the sophisticated D = diameter
predictive methods both for the estimation of the performance Df diffusion factor
characteristics and the flow analysis, the empirical and the semi- G = weight flow rate
empirical formulae, parameters and constants related to the losses G o corrected weight flow rate
within the compressor channels, as well as the precise configurations AG Gchoke — Gsurge
of the compressors used for the predictions, are not extensively H = specific enthalpy for loss or form factor
presented in the published reports. Therefore, it is quite difficult to • Mach number
know all of the design procedures for a wide variety of compressors. • rotational speed
Concerning the internal flow patterns within the compressor N s = specific speed
channels, lots of complex flow phenomena which belong to the impeller tip speed
critical problems in the fluids engineering field appear, for example, ✓ absolute velocity
the development and the separation of the three-dimensional W = relative velocity
boundary layer under the effect of the Coriolis force, the unsteady • number of blades

Presented at the Gas Turbine Conference and Exhibit

Houston, Texas — March 18-21, 1985

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms


= incidence angle ad adiabatic

1 = length along mean streamline aero aerodynamic

= radius av = averaged

= tip clearance b = blade
to = tangential thickness of blade curvature in meridional plane
choke condition
a = angle between absolute velocity vector and radial d rec on choke

= angle from meridional plane m meridional component

43, = flow coefficient (= Vz, /U2 ) max maximum value

rl = efficiency mean mean value
minimum value
= slip factor min

v = kinematic viscosity opt optimum value
p = density = relative velocity

rr = total pressure ratio surge surge condition

Subscripts th = vaned-diffuser throat or theoretical
z axial direction
1 = inducer inlet
2 = impeller exit
3 = vaned-diffuser leading edge COMPRESSORS EMPLOYED FOR THE PRESENT STUDY
4 = vaned-diffuser exit The measured performance characteristics and the major dimen-
5 compressor exit sions of the compressors employed for the present study are shown
inducer hub in Tables 1 and 2, respectively. Forty-six operating conditions of
IT = inducer tip eighteen compressors designed by nine different manufacturers for
I rms = area-mean radius of inducer leading edge the industrial, aeronautical, vehicular and experimental purposes were
BL = blade loading investigated [3. 1 1-17] . The compressors were chosen to include both
DF = disk friction the primitive and the sophisticated design techniques with the known
HYD = hydrodynamic configurations and the experimental performance characteristics.
INC incidence No swirl condition at the inducer inlet was employed. Since some
LEAK = leakage of the values in Table 2 were estimated from the published reports,
MIX = mixing errors might be involved in them. However, the computed results
RC = recirculation which will be shown in the following sections did not exhibit critical
SEC = secondary flow errors for the major parameters of the performance characteristics.
SF = skin friction The compressors for which D5 /D2. As /A4 and ce 3 b are not shown
VLD vaneless diffuser in Table 2 have vaneless diffusers. The Type E2 is the first stage of
VD = vaned-diffuser the El. and the J 1 and the J2 possessed the conical impellers with

Table 1 Operating Conditions of Compressors

Type A B C D El
N rpr 43200 57600 64800 72000 27920 31410 34900 27920 31410 35000 40000 50000 75000 31800 37100 42400 47700
428 481 535 428 481 426 487 608 626 331 387 442 497
2
M 0.9625 1.2833 1.4438 1.6042 1.2908 1.4521 1.6134 1.2908 1.4521 1.2510 1.4297 1.7870 1.8407 0.9727 1.1373 1.2989 1.4605
u2
R 2.6441 3.7492 4.9763 2.6207 3.3000 4.1667 2.8400 3.6069 3.2100 4.5000 8.8400 8.0000 1.7800 2.1900 2.9300 3.5400
mut
31max 2.2797 3.7373 5.9492 6.3447 3.2621 4.2276 5.3231 3.2200 4.1552 3.3100 4.6000 9.0400 8.8560 2.0500 2.5600 3.2100 4.2000
It 1.7563 3.1907 4.8492 5.6605 2.9414 3.7638 4.7449 3.0300 3.8811 3.2600 4.5500 8.9400 8.4280 1.9150 2.3750 1 3.0700 3.8700
C 0.2345 0.3837 0.5026 0.6200 2.4249 3.1594 4.0353 2.0705 3.4504 0.2858 0.4400 0.8392 0.8050 0.5950 0.805011.4750 1.81501
min
C, 0.3340 0.5263 0.6400 0.7054 2.8548 3.6232 4.2813 3.2142 3.9314 0.3180 0.4944 0.8754 0.9220 1.1500 1.4350 1.7700 2.145(11
MaX

I\Te gl 02 E3 G 111 H2
N rim 53000 28300 31850 35030 40280 23000 18000 22000 25040 10000 14000 16000 18000 14000 160001 220001 26510
552 277 312 343 394 337 332 405 328 209 293 335 377 293 335TE 207 250
11 2

u2 1.6221 0.8140 0.9169 1.0080 1.1578 0.9909 0.9512 1.1603 0.9397 0.6142 0.8610 0.9845 1.1079 0.8610 0.9845 0.6083 0.734 7
II 4.7200 1.3350 1.4250 1.6000 1.9400 1.4030 1.4600 2.1800 1.6761 1.4395 1.9185 2.2593 2.6000 1.6321E1.9333 1.4333 1.6551
min
11 5.3400 1.7250 1.9600 2.2100 2.7500 2.2500 2.3000 3.1600 1.9689 1.4864 2.0247 2.4537 2.9630 1.9481 2.3111 1.4503,1.6948
ma x
II 5.0300 1.5300 1.6925 1.9050 2.3450 1.8265 1.8800 2.6700 1.8225 1.4630 1.9716 2.3565 2.7815 1.7901 2.1222 1.4418 1.6750
G. 2.0700 0.2650 0.3200 0.3960 0.5430 1.5400 2.2600 3.1200 1.6589 2.7805 4.0849 5.3415 6.2289 3.3855 4.0000 0.2370 0.3181
min
2.3400 0.4720 0.5330 0.5860 0.6890 2.2800 3.2800 4.2000 2.1010 6.0000 5.0000 8.7561 9.1975 6.7590 7.1463 0.8027 1.0000
max
Type 71 12 J1 J2 J3 K
N rpm 31000 35500 40000 22000 35500 40000 21540 15610 20740 16234 21650 33680
2 292 335 377 207 335 377 203 147 196 153 204 365
u2 0.8581 0.9845 1.1079 0.6083 0.9845 1.1079 0.5966 0.4320 0.5760 0.4496 0.5995 1.0458
IT E 1.9569 2.3369 2.8497 1.4045 2.3427 2.8594 1.4150 1.2310 1.4100 1.2340 1.4120 1.5129
min
H 2.0028 2.4146 2.9432 1.4056 2.4218 2.9248 1.4910 1.2590 1.4910 1.2580 1.4790 2.2808
max
II 1.9799 2.3758 2.8965 1.4501 2.3823 2.8921 1.4530 1 2450 1.4505 1.2460 1.4455 1.8969
.

Amin 0.3812 0.5000 0.8568 0.2795 0.5000 0.8573 0.0750 0.0570 0.0750 0.0580 0.0800 0.9896
Amax 1.2038 1.4267 1.5677 0.8205 1.4411 1.5886 0.7640 0.5450 0.7790 0.5280 0.7580 1.5460

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 the cone angles of 45 and 60 degrees, respectively. If the blade angles The relationships between the total pressure and the peripheral
of the inducer and the vaned-diffuser were not shown in the litera- speed of the compressors are shown in Fig. 2. The dotted line is that
tures, they were estimated under the assumption that the shockless reported by J. Hawthorne in 1946 [19]. The remarkable increase in
inlet condition for the mean-camber-line was maintained at the the total pressure to the peripheral speed in recent years can be clearly
design point. Therefore, a few degrees of error might be involved in seen.
these cases.
The meridional configurations of the impellers are shown in Fig. 1. THE METHOD FOR ESTIMATION OF PERFORMANCE
As can be seen from Tables 1, 2 and Fig. 1, the compressors have a CHARACTERISTICS
wide variety of the major factors which essentially govern the per- The flow chart of the computational process for the estimation is
formance characteristics, for example, the number of the blades, the shown in Fig. 3. The main part of the computer program was similar
inlet-eye diameter, the axial length of the inducer, the impeller outer to that reported by M. R. Galvas[3] and will not be repeated in the
diameter and the blade height, the meridional curvature, the flow present paper. In addition to the Galvas' program, the following
range, the peripheral speed, the length of the mean streamline along formulae, parameters and constants for the losses within the com-
the compressor channel, the inlet as well as the exit diameters of the pressor channels were considered in the present program.
vaneless and the vaned-diffuser and the blade angles of the inducer (1) Inducer Incidence Loss
and the vaned-diffuser. The formula given by M. P. Boyce[2] :
2
AW
H = (1)
I NC I NC 2
where
AW=W • c o t /3
ul rms erns Inns

C
INC
=O. 6-0. 9 0) (1 .2)
=0. 6 - 1. 2 ( i < 0 )

} -A r -D
12
-E3
Typi := ---- El Type
d d
11

10

7
k 6

1 0 200 400 600

U2 (m/Sec)
Fig. 1 Meridional Configuration of Impellers
Fig. 2 Total Pressure Ratio

Table 2 Major Dimensions of Compressors

Ref. (3) (1 ) (12) (13) - (14) 15 - 16 f (17) (14)
J2 J3 K
NM A
EIZn0f) 0
B C
-C14
D El E2
EC
E3 F
•
G
n
Hi
AINVA MI/ MOM
H2 II 12 Ji
1 r • _•
13711111E1211117 17 18 9 18 • 14 20 18 20 15 15 14 14 14 16
11•= 01448 Q2925 0.2925 0.2323 0.1595 0.1990 0.1870 0.2800 04000 0.2500 0.4000 0.1800 0.1800 01800 02070
(Tx
•• gm
01800 0.1800
VERN 04417 0.4417 0.4517 0•.5003 0•.4154 0.601/ 0.3939 0.7000 0.4851 0.7000 0.7000
•• •• :•
07000 06166 06166 06473
06166
A: AD •6 50 3: 95 28 oa 03 21 69
: gozz3. 0607664903 0.0650 0.0618 O•650 0.0650 0.065•00500 00500 00444
Q0500
0.5963 0.8357 05495 05962 05962 06706 05706 05215
06706
3 . 4•• 57.2 57.2 47.0 524 57.8 586 56.0 55.0 53.0 585 600 63.0 63.• 30.5 385 60.0
28.0
250 3•.• 300 00 300 400 400 00 0.0 300 0• 300 0.0 . 00 00 ao 0.0 25.0
D3 • 1.1402 1.2156 1.2156 10600 1.0764 1.1256 1.1176 11071 11080 1,1640 16870 1.6000 16000!16000 1.3300 1.3300 1.3300 17390
COLMI••Z 17366 17366 19356 3.1846 17387 1 8182 i 5429 1.42•5 1.9720
1/31/ESP ZOINFIEEll .•• 28937 41 37 1.70591.8758 1,9844 1.4•36 2,1262 : - -
04 b 780 635 63.3 83.5 75.6 750 75.• 73.• 74.• 69.6 ,

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 The specific speed is expressed by
If the slip factor and the specific speed were assumed to be constant,
the predicted curves as in Fig. 12 could be obtained from these simple
equations. The predicted curves qualitatively resemble the computed
Ns - (15)
H°. 75 ones given by the estimations of the performance characteristics. The
estimated lines were given by the following equations:

Assuming H = H a d = pU 2 2 /g, under the non-swirl condition, the next

equation is approximately derived as follows: 0.473 (N s S 100)
WI
C = 0.470-0 1 + 0.403 (N s 80) (17)
= (01 2 + r.5 (16) U2
U2 01
0.273 (N s 50)

C = 1.9 x 10 -6 . I -5 • N s 2 (16.1)
The estimated W 2 /U 2 did not directly correlate to the flow coeffi-
cient and the peripheral speed. Only the qualitative correlation was
obtained for the meridional component of the velocity as in Fig. 13.
The constant meridional component of the impeller inlet and exit ex-
hibited the average of the results for all of the compressors. It may
confirm the result already reported by the present authors for the im-
peller with the constant meridional velocity [23, 24] . The estimated
W2/U2 showed the clear correlation to a2 as in Fig. 14. This could
also be predicted from the simple considerations about the velocity
triangle at the impeller exit. As it can be expressed by the following
equation:

W2
= cot2a2 + (p - 1) 2 } 0.5 (18)
U2

the curve obtained by the constant slip factor were easily drawn. The
empirical formula given by the estimated results could be obtained as
O follows:

W2 'a2
+ 1.4587 ± 0.0503 (19)
U 2 64.6

The effects of the backward-leaning angles of the impeller blades on

the above relation were little for the present compressors.

4.0 6.0 8.0

71:, 2
Variation of Flow Angle at Impeller Exit

0.1 0.2 Q3 0.4 01 0.2 0.3 0.4

91
Fig. 13 Meridional Component of Non-dimensional Relative
Fig. 12 Non-dimensional Relative Veloc ty at Impeller Inlet Velocity

-8-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 and by M. P. Boyce[2] :
COMPARISON BETWEEN COMPUTED AND EXPERIMENTAL
H = 0.03 u • V • U 2 (10) RESULTS
UH u2th The adiabatic efficiency 'lad was employed for the comparisons
between the computed and the experimental results. However, it was
were employed.
inappropriate and it made the comparisons complex. The variations of
(6) Secondary Flow Loss
the disk friction and the recirculation losses affected the denominator,
As regards the secondary flow loss [10] , whether each or the
both of the skin friction and the blade loading losses could be while the other losses were included in the numerator as shown below:
replaced by the secondary flow loss or not was examined:
sinS2 0.3 lad = Haero — (HINC + HBL + HSF + HVLD HVD) (13)
?
H SEC C SEC [ 2 W 2 COS fi
D2 ).( W2 ) 2 2 •
(
U 222 e
„ ) ( Haero + HRC + HDF
2 ) 2
W2 2
Thus, the total pressure ratio was employed here. The exchange of the
2D formula for the disk friction loss did not affect the other losses.
UD
where The comparisons were made qualitatively rather than quantitative-
ly. The considerations about the total amount of the losses led one to
51 employ the qualitative comparisions, i.e., predicted performance
SEC R characteristics given by the total pressure ratios already included all of
e
the losses accounted within the compressor channels and the measured
W 2 • D IIYD values were also given by the flow patterns after the generations of the
R (11.2) losses. It would be impossible to know the influences of a loss on the
e1111) V
3 other losses accurately. Therefore, in the following, the combinations
(7) Mixing Loss which apparently derived the unreasonable results in the present meth-
The possibility of the application of the mixing loss[8] to the od will be discussed as the differences between the estimated and the
present estimation was examined: measured performance characteristics were clarified.
6 With regard to the secondary flow loss, the variation for the flow
W t 2 rate exhibited the similar tendency as those of the skin friction and
1 ( 2 7r r 2 /Z2 ) 2 the mixing losses as in Figs. 4 (a) and 4 (b). In the range of high pres-
H V
MIX V +t 2 sure ratios, the total amount of the losses was always overestimated
u2 2 192
2 (1 + ) 1 ( ,, w /7 ) as in Fig. 5 for any combinations of the other losses. Thus, the results
Vm2 r 2 ‘, 2
(12) containing the secondary flow loss derived the larger losses than the
measured values without the skin friction and the blade loading losses.
where Therefore, the employment of the secondary flow loss was not always
suitable for the present method.
2W2 + is i r1,8 ( r2 W,2 ) 2 0.4
= C • (
2 7 r 2
t ) U2 — 2 re U2
w 2 02
Co S132 0.12
(12.1) H 5F
N rpm MR Galvas
C =0
'
3 (12.2) 40280
35030
As to the recirculation loss, the same formula as that proposed by 3185 0
F. Dallenbach[ I ] was used without discussions here, because it was 010 28300
also used by M. R. Galvas[3] and M. P. Boyce[2]. Eq (3) -

Although the other calculation methods for the vaneless and

the vaned-diffuser losses exist[ 1, 2. 20] , those employed by M. R.
Galvas[3] were used. They were considered to possess the higher
generality than the others.
0.08
•
The slip factor is the critical parameter for the estimations of the
H SEC
performance characteristics. The formula by F. J. Wiesner[ 21] was
employed here as done by M. R. Galvas[3].
Ea.(11)
The tip leakage loss was not accounted by M. R. Galvas[3 [ .
Although this loss was considered here, the formulae are proportional
to the square of the peripheral speed and were less affected by the
variation of the flow. rate. In addition, the tip clearances were usually ons -

not shown for the compressors. Therefore, only the influence of

this loss on the total amount of the losses was taken into account
as it caused the equal decrease of the efficiency or the total pressure
ratio for the variation of the flow rate.
In this way, the numbers of the combinations of the above-
mentioned formulae, parameters and constants including those in the 004
M. R. Galvas' program became as follows: (1) three for the inducer
incidence loss, (2) three for the skin friction loss including the
exchange of this loss with the secondary flow loss, (3) nine for the
blade loading loss by combining this loss with the secondary flow loss,
(4) three for the disk friction loss. The half of the results for the 0.02 02 63 0:4 0.5 66
forty-six operating conditions, 3000, were carefully examined by 0.1
comparing those with the experimental ones. The maximum and the
minimum flow rates for the respective operating conditions were
determined by the preliminary computations in order to involve
the surge and the choke flow rates given by the criteria of M. R. Fig. 4 (a) Skin Friction and Secondary Flow Losses
Galvas[3] and C. Rodgers[22] with enough allowances.

—5—

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms



The inducer incidence loss given by M. R. Galvas [3] showed the sons with the experimental results at the off-design flow rates. The dif-
values of nearly zero. In his paper, the cases where the compressors ferences between Eq. (1) and Eq. (2) are shown in Fig. 6. Eq. (2) given
were operated at the small variations of the flow rate near the design by the present authors exhibited the best agreements in the compari-
point condition in the range of the high pressure ratio were discussed sons.
at almost zero incidence angle. However, at low pressure ratios with Two essentially different types of the curves appeared for the
the wider variations of the flow rates, this loss behaves in the more blade loading loss as in Fig. 7. The formula of M. R. Galvas [3] and
critical manner toward the off-design point conditions. As to Eq. (1), that with Eq. (4) showed the opposite tendencies of the curves for the
it showed clearly the overestimation of this loss during the compari- variation of the flow rates to those by Eqs. (5) and (6). When the inci-

0.10 Type H1 0.08

0.08 0.06

0.06 f
0.04

0.04 0.02

0.02

/ A i
Fig. 6
0.1 0.2 03

Inducer Incidence Loss

04

....7
= i
--
= ,

0.1 0.2 03 0.4 05

T
Fig. 4 (b) Mixing Loss
0.08
7.0 Type A N=72000 rpm

6.5 0.06

pi,60
0.04

5.5
- without Secondary
Flow Loss
0.02
--- with Secondary
5.0 Flow Loss

Q55 0.60 0.65 0.70 0.75 0.1 0.2 0.3 04 0.5 06

Go(Kg/Sec)
Fig. 5 Comparison between Estimated and Measured Results Fig. 7 Blade Loading Loss

-6-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms



dente, the skin friction, the secondary flow and the mixing losses were More qualitative comparisons than those mentioned above were
examined in Figs. 4 and 5, these losses increased with the increase in also made for the respective curves given by the combinations of the
the flow rate except for the incidence loss. If the blade loading loss formulae, parameters and constants. The mixing loss was considered
varied in the same manner, the result shown in Fig. 8 could not be ob- necessary because the results containing this gave better agreements.
tained. Therefore, Eqs. (5) and (6) were considered inappropriate for Eq. (7) showed the better results than those by Eq. (8) and so, Eq. (7)
the present method. will be used in the following. Only the thirteen cases resulted in the
The results employing the newly proposed diffusion factor of Eq. good agreements of the predicted results with those by experiments as
(4) exhibited less influences on the blade loading loss than those by in Fig. 8. However, the secondary flow loss was involved in Fig. 8.
M. R. Galvas [3] and F. Dallenbach [1]. The essential difference be- This fact also implied that the different assumptions for the losses
tween these two diffusion factors could not be shown for the present could give one of the best agreements with the experimental results.
compressors.
Concerning the skin friction and the mixing losses, it was quite MAJOR PARAMETERS DIRECTLY RELATED TO
difficult to make the comparisons as in Fig. 9. According to Fig. 4 (a), PERFORMANCE CHARACTERISTICS
the skin friction loss by Eq. (3) showed almost the twice of that by The relationship between the variation, between the surge and the
M. R. Galvas [3]. Although this loss was not as large as the secondary choke flow rates, of the relative velocity ratio to that of the total pres-
flow loss at high pressure ratios, the values of the coefficient of skin sure ratio and the total pressure ratio is shown in Fig. 10, where the
friction was about 0.008, which was twice of the value used by M. R. relative velocity ratio was defined as the ratio of the relative velocity at
Galvas [3]. Thus, the constant value of M. R. Galvas [3] was con- the impeller inlet to that at the exit. The remarkable decreases of the
sidered more reasonable. variation of the flow range from the surge to the choke point occured
as the total pressure ratio increased. Thus, the variation of the
d(W2/W1)/dif, _ 2 exhibited the steep curve and approached to a con-
Type 12 N=40000 rpm stant value toward the higher pressure ratio. On the other hand, at the
3.0 low pressure ratio, the variation of the flow rates becomes much wider
than that at the higher pressure ratio. Thus, the variations of the rela-
tive velocity ratio also became wider compared to those of the total
pressure ratio. The similar parameter as d (W2/W1 )(iv - 1_2 was ex-
amined in Fig. 11, where a 2 is the flow angle at the impeller exit. This
parameter was defined in the same manner as mentioned in Fig. 10.
k' 2.9 The similar curve as in Fig. 10 was also obtained. The strong correla-
tion between the cv 2 and the W 2 /W 1 appeared. Therefore, the non-
dimensional forms of each of W 1 and W2 will be examined separately
in the following.
As the flow coefficient was defined as (i51= Wiz/U2 and the non-
dimensional area-mean eye-diameter at the impeller inlet can be di-
2.8 fined by E = DIrms/D2, the following equation is obtained.

0.8 11.2 14 16
G. (Kg/Sec) (01 2 ± E 2 )0.5 (14)
Fig. 8 Comparison between Estimated and Measured Results U2

Type E 3 N =23000 rpm

1.8
2.5
- 1.6
2.3
1.4

1.2
2.1
1.0
0
1.9
. 4∎■ , Eq(3)
1/4 Neglect
08
Eq.(12) - -
0.6
1.7 Negtect -- 0
-- -

0.4
Li 11
1.5 02•
•
0

14 1.8 20 22 24 0 6.0 8.0

1.6 2.0 4.0
G o (Kg /Sec) 7t1-2
Fig. 9 Comparison between Estimated and Measured Results Fig. 10 Variation of Relative Velocity Ratio

-7-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 The specific speed is expressed by
If the slip factor and the specific speed were assumed to be constant,
the predicted curves as in Fig. 12 could be obtained from these simple
equations. The predicted curves qualitatively resemble the computed
Ns - (15)
H°. 75 ones given by the estimations of the performance characteristics. The
estimated lines were given by the following equations:

Assuming H = H a d = pU 2 2 /g, under the non-swirl condition, the next

equation is approximately derived as follows: 0.473 (N s S 100)
WI
C = 0.470-0 1 + 0.403 (N s 80) (17)
= (01 2 + r.5 (16) U2
U2 01
0.273 (N s 50)

C = 1.9 x 10 -6 . I -5 • N s 2 (16.1)
The estimated W 2 /U 2 did not directly correlate to the flow coeffi-
cient and the peripheral speed. Only the qualitative correlation was
obtained for the meridional component of the velocity as in Fig. 13.
The constant meridional component of the impeller inlet and exit ex-
hibited the average of the results for all of the compressors. It may
confirm the result already reported by the present authors for the im-
peller with the constant meridional velocity [23, 24] . The estimated
W2/U2 showed the clear correlation to a2 as in Fig. 14. This could
also be predicted from the simple considerations about the velocity
triangle at the impeller exit. As it can be expressed by the following
equation:

W2
= cot2a2 + (p - 1) 2 } 0.5 (18)
U2

the curve obtained by the constant slip factor were easily drawn. The
empirical formula given by the estimated results could be obtained as
O follows:

W2 'a2
+ 1.4587 ± 0.0503 (19)
U 2 64.6

The effects of the backward-leaning angles of the impeller blades on

the above relation were little for the present compressors.

4.0 6.0 8.0

71:, 2
Variation of Flow Angle at Impeller Exit

0.1 0.2 Q3 0.4 01 0.2 0.3 0.4

91
Fig. 13 Meridional Component of Non-dimensional Relative
Fig. 12 Non-dimensional Relative Veloc ty at Impeller Inlet Velocity

-8-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 SURGE AND CHOKE CRITERIA
The choke flow rate for the vaned-diffuser was predicted accurate-
ly as in Table 3 except for the compressors with the large divergent
angles of the channels.
With regard to the inducer surge and the choke flow rates, the
same relation as that proposed by C. Rodgers[22] was examined in
Fig. 15, which was less accurate both for the surge and the choke.
0.8 • However, the backward-leaning blades were involved here.
SI
/
The criterion for the diffuser surge condition given by
Y. Senoo[25] showed the reasonable results qualitatively as in
Fig. 16. However, the shifts of the lines for the respective compressors
are seen. Although the more detailed examinations would be necessary
A
by taking into considerations the configurations, the loadings and
0.6 -az so on, they are beyond the scope of the present study because of the
employment of the generalized pressure recovery coefficient in the
estimations.
4: In addition to these criteria, the surge-to-choke margin in Fig. 17
‘Iv ■ exhibited the large scatter of the computed results. This scatter is
0 Predicted ■;' .0 rather large for the accurate prediction of the surge flow rate.
Thus, the accurate criteria for the surge and the choke were not
• Eq(8) • clarified except the vaned-diffuser choke flow rate.
AI= 0.90
0.80 -.-
Table 3 Choke Flow Rate of Vaned-Diffuser
Q2 C): Some of the Data were
7., at the Surge Choke Flow Rate (kg/Sec)
LW; All of the Data were Type A 4 /A 11
1-5 Predicted
3 Measured
at the Surge
1) 0.1905 1.756 0.3340 0.3447
A 4.849 0.6400 0.6362
B 0.1698 3.030 2.8548 2.8554
3.881 3.6232 3.6008
40 50 60 70 80 D 0.1400 8.428 0.9220 0.8964
0(2 (Degree)
El 0.2263 1.915 1.1500 1.1689
3.870 2.1450 2.1590
E3 0.2734 1.827 2.2800 2.5276
Fig. 14 Non-dimensional Relative Velocity at Impeller Exit F 0.2874 1.880 3.2800 4.3037

1.3 (25)
1.2 Blb
40 ° od d: Vaned -
50° 1.2 Diffuser
60° 4:1
0 1.1
1.0 •0 ,
OC

1.0
co .
▪- 09 0
! Q8 X '
A (.) oec \
0 0.8 0
\0 0.7
\
Q6
At 13 MI
0 60
A
e,
\ 0.6 e
0
A 0
0 e
04 0.5

-10 0 +10 +20 04 5 10 15 20 25

iirrns (Degree) hr (Degree)
Fig. 15 Surge and Choke Criteria by Inducer Incidence Angle Fig. 16 Surge Criterion by Vaned-Diffuser Incidence Angle

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 CRITERIA FOR RELATIVE VELOCITY RATIO The relationships among the parameters in this figure will be able to
There are many papers for the criterion of the value of the relative explain the fundamental trends for the designs of the compressors
velocity ratio under the optimum or the surge condition[25-27] . operated from the high to the low peripheral speeds as well as the
The variation of this ratio from the surge to the choke are shown in flow rates. The compressors operated at the high pressure ratio
Fig. 18. As was shown in Fig. 10, the W2 /WI converged to a certain attained the lower flow rate with the higher slip factor and those with
value with the increase of the total pressure ratio. When the increase the low pressure ratio will not be able to be operated in the extremely
of the total pressure ratio is considered to be equal to the increase of lower flow rate. Furthermore, the impellers with the constant meridi-
the relative Mach number at the root-mean-square radius of the in- onal velocity from the inlet to the exit were preferable at the higher
ducer inlet, the relative velocity ratio will converge to a certain value. specific speed, while those designed for the lower specific speed
Therefore, the criteria of this parameter only exhibit the value attain- had the lower meridional velocity at the impeller exit than that at
able at the highest pressure ratio which is assumed for the modern the inlet.
compressors. As denoted in Fig. 14, the same relationship between
the W2 /U 2 and the a 2 is valid at the surge condition. The ultimate CONCLUSIONS
value of the a 2 was 81 degrees. The following conclusions are obtained from the present study.
By using Eqs. (16) and (18), the relative velocity ratio, W2 /WI (1) A method is proposed for the estimation of the performance
was examined for a 2 as in Fig. 19. All of the ordinates and abscissas characteristics of centrifugal compressors, which are used for the
were arranged to involve the surge and the choke conditions. It is compressors with both the high and the low pressure ratios as well
apparent that the optimum value as well as the criterion for the as the large and the small flow rates.
surge which employ this relative velocity ratio must be determined as (2) The non-dimensional relative velocity at the impeller inlet is
the parameters of the slip factor, the flow rate and the specific speed. expressed by the empirical formula Eq. (17) with the flow coefficient,
the slip factor and the specific speed as parameters. It is also predicted
from the simple theoretical considerations in Eq. (16).
(3) The non-dimensional relative velocity at the impeller exit is
0.7 related to the flow angle at the impeller exit. The empirical formula
is given in Eq. (19), while the theoretical one is expressed by Eq. (18).
(4) The ratio of the relative velocity at the impeller exit to that at
0.5 0
the inlet is expressed by the flow coefficient, the slip factor and
the specific speed for the flow angle at the impeller exit. Thus, the
constant values for this ratio defined as the surge and the optimum
Q3 conditions can not be determined without employing these para-
0
coo meters.
0.1 g0 eP
e REFERENCE
1. Dallenbach, F. et al., "Study of Supersonic Radial Compressors
04 0.5 0.6 0.7 0.8 0.9 1.0 1.1 for Refrigeration and Pressurization Systems", WADC TR 55-
Mtn 257, (1956)
2. Boyce, M. P., "New Developments in Compressors", Proceedings
Fig. 17 Surge-to-Choke Margin of Texas A&M University, (1972)
3. Galvas, M. R., "Fortran Program for Predicting Off-Design
Performance of Centrifugal Compressors", NASA TN D-7487,
(1972)
— C. Rodgers(2/
-- Y. Senoo (25)
— H.Mishina 14
Gacke

1.0
Fig. 19 Predicted Relationship between Relative Velocity Ratio
Fig. 18 Surge Criterion by Relative Velocity Ratio and Flow Angle at Impeller Exit

10

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms

 4. Wu, C. H., "A General Theory of Three-Dimensional Flow in
Subsonic and Supersonic Turbomachines of Axial-, Radial- and
Mixed-Flow Types", NACA TN 2604, (1952)
5. Katsanis, T., "Use of Arbitrary Quasi-Orthogonals for Calculating
Flow Distribution in the Meridional Plane of a Turbomachines",
NASA TN D 2546, (1968)
6. Katsanis, T., "Computer Program for Calculating Velocities
and Streamlines of a Blade-to-Bade Stream Surface of a Turbo-
ma chine", NASA TN D-4525, (1968)
7. Marsh, H., "A Digital Computer Program for the Through-flow
Fluid Mechanics in an Arbitrary Turbomachines Using a Matrix
Method", ARC R&M 3509, (1968)
8. Johnston, J. P., et. al., "Losses in Vaneless Diffusers of Cen-
trifugal Compressors and Pumps", Trans. ASME, Ser. A,
Vol. 88-1, (1966-1), P49
9. Rodgers, C., "Specific Speed and Efficiency of Centrifugal
Impeller", Performance Prediction of Centrifugal Pumps and
Compressors: ASME, New York: (1976), P191
10. Mashimo, T., "Effects of Reynolds Number on Performance
Characteristics of Centrifugal Compressors, with Special Refer-
ence to Configuration of Impellers", ASME Paper 74-GT-59,
(1974)
11. Beard, M. G., Pratt, C. M. and Timmis, P. H., "Recent Experience
on Centrifugal Compressors for Small Gas Turbines", ASME
Paper 78-GT-193, (1978)
12. Runstadler, P. W. and Dean, Jr., R. C., "An Experimental Study
of Surge in Centrifugal Compressor", Centrifugal Compressor
and Pump Stability, Surge and Stall; ASME, New York: (1976),
P.69
13. Osborne, C. Runstadler, P. W. and Stacy, W. D., "Aerodynamic
and Mechanical Design of an 8 :1 Pressure ratio Centrifugal
Compressor", NASA CR-13782, (1975-4)
14. The Answer to the Questionnaire of the Committee for Estima-
tion of Performance Characteristics of Large Scale Turbomachine
in the Japan Society of Mechanical Engineers, (1981)
15. Eckardt, D., "Flow Field Analysis of Radial and Backswept
Centrifugal Compressor Impellers", Performance Prediction
of Centrifugal Pumps and Compressors; ASME. New York:
(1979), P.77
16. Bammert, K. "The Influence of the Meridional Impeller Shape
on the Energy Transfer in Centrifugal Compressors", ASME
Paper 80-GT-48, (1980)
17. Mizuki, S., Ariga, I. and I. Watanabe, I., "Estimation of Jet and
Wake Flow Rates within Centrifugal and Mixed-Flow Impeller",
Performance Predictions of Centrifugal Pumps and Compressors;
New York (1979), P.87
18. Morris, R. E. and Kenny, D. P., "High Pressure Ratio Centrifugal
Compressors for Small Gas Turbine Engines", Advanced Cen-
trifugal Compressor; New York: ASME, (1971)
19. Hawthorne W. R., SAE Annual Meeting. (1946-7), P.7
20. Runstadler, P. W., "Diffuser Data Book", Creare Inc. No. 75-
15017, (1975)
21. Wiesner, F. J., "A Review of Slip Factor for Centrifugal Impel-
lers", Trans. ASME, Ser. A. Vol. 89-4. (1967-10), P.558
22. Rodgers, C., "Typical Performance Characteristics of Gas Tur-
bine Radial Compressors", Trans. ASME, Ser. A, Vol. 86, (1964),
P.161
23. Mizuki, S. et. al., "Investigations Concerning the Blade Loading
of Centrifugal Impeller", ASME Paper 74-GT-143, (1974)
24. Mizuki, S. et. al., "A Study on the Flow Mechanism within
Centrifugal Impeller Channels", AMSE Paper, 75-GT-14, (1975)
25. Senoo, Y., "Pressure at Shroud and Flow in a Supersonic Cen-
trifugal Impeller", Paper No. 42, 1977 Tokyo Joint Gas Turbine
Congress, (1977)
26. Mishina, H., "Effect of Relative Velocity Distribution on Impel-
ler Efficiency", Trans. JSME, Vol. 44-384, p.2689
27. Rodgers, C., "Impeller Stalling as Influenced by Diffusion
Limitations", Centrifugal Compressor and Pump Stability, Surge
and Stall; ASME, New York: (1976), p.37

11

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 07/24/2015 Terms of Use: http://asme.org/terms