Deterioration of Compressor

Y. Senoo Performance Due to Tip Clearance

Professor,
Research Institute of Industrial Science,
Kyushu University,
of Centrifugal Impellers
Kasugashi, Fukuoka816, Japan
Fellow ASME The authors' theory on the tip-clearance loss of centrifugal impellers is modified to
include the variation of slip coefficient of the impeller due to the tip clearance, by
deriving a rational relationship between two empirical parameters in the theory. In
IV]. Ishida order to compare experimental data in the literature with prediction, examination
Professor, was made regarding accuracy of available data and the way to select corresponding
Faculty of Engineering, flow rates of a compressor with different values of tip clearance. Good agreement
Nagasaki University, between data and prediction was observed. These examples demonstrate the follow-
Nagasaki, 852, Japan ing tendency regarding effects of various parameters on the tip clearance loss. Effi-
ciency drop due to the tip clearance of high-pressure-ratio compressors is less than
that of low-pressure-ratio compressors if the tip clearance ratio at the impeller exit is
equal. The magnitude of clearance loss becomes smaller as the flow rate is reduced
and also at a reduced shaft speed in cases of high-pressure-ratio compressors. The
equations in the theory clearly show these tendencies.

Introduction O GTCP 1 8 5 . 1
a TFE 2ND STAGE TEST 30
ONASA BRU
In many turbomachines impellers are not shrouded and the VTPE331.301 2ND STG TEST 184
ATPE331.301 2ND STG t e s t 186
performance is deteriorated by the pressure loss due to the O TPE 2ND STAGE TEST 135
clearance between the shroud casing and the tips of impeller
blades. For the design of turbomachines, therefore, it is im-
portant to estimate effects of tip clearance correctly, and
several empirical equations have been published in the
literature. Pampreen [1] collected data of six different cen-
trifugal impellers and correlated the efficiency drop to the tip-
clearance ratio at the exit of impeller, and he drew an average
correlation line with an inclination of 0.3 in Fig. 1. In the jj 0 0.05 0.10 0.15

literature usually the relationship has been published as an em- w clearance/rotor exit width variation

pirical formula Fig. 1 Effect of tip clearance on efficiency drop in the literature [1]
A?; lac
(1)
t\ b1+b2 ficiency at zero clearance.1 Since it is not possible to test an
Eckert and Schnell [2] chose a = 0.9 while Pfleiderer [3] recom- impeller at zero clearance, the reference efficiency at zero
mended a = 1.5 — 3.0. Pampreen's average line roughly agrees clearance is dubious, but the inclination of each broken line is
with Eckert and Schnell's equation providing that b{/b2=4 not influenced by one particular datum of reference, therefore
and 57 = 0.8. the inclination is reliable. Since the inclination of broken lines
In Fig. 1 four broken lines are drawn to connect data which is different from one impeller to another, it is clear that the ef-
belong to respective impellers. Inclination of these lines is dif- ficiency drop due to tip clearance can not be decided only by
ferent from each other and all of them are considerably larger the parameter in the abscissa or the clearance ratio at the im-
than that of Pampreen's full line. It is also noticed that none peller exit.
of these broken lines passes through the origin. In Fig. 1 the In [5], the present authors have proposed a new approach
ordinate is the decrement of efficiency in comparison to the ef- regarding the tip-clearance loss. It consists of the pressure loss
induced by the leakage flow through the clearance and the
1
It is experimentally demonstrated that the performance of an unshrouded
impeller with small tip clearance was superior to that of a shrouded impeller
with identical geometry in a wide flow range except at a large flow rate [4]. It is
generally understood that motion of the stationary shroud relative to the blade
Contributed by the Gas Turbine Division of T H E AMERICAN SOCIETY OF prevents boundary layer fluid from accumulating at the blade suction surface
MECHANICAL ENGINEERS and presented at the 31st International Gas Turbine and consequently reduces pressure loss. Therefore, the performance of a
Conference and Exhibit, Diisseldorf, Federal Republic of Germany, June 8-12, shrouded impeller is different from the performance of an unshrouded impeller
1986. Manuscript received at ASME Headquarters January 24, 1986. Paper No. with zero tip clearance and the latter can be estimated only by extrapolating data
86-GT-123. of the impeller with different values of the tip clearance.

Journal of Turbomachinery JANUARY 1987, Vol. 1 0 9 / 5 5

Copyright © 1987 by ASME
Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo
 pressure loss for supporting fluid against the pressure gradient pressure force along the wall is equal to (2a - X)c times the un-
in the blade channels and agamst the pressure gradient in the disturbed pressure difference across the wall where a is the
thin annular clearance space generated by the rotating blade contraction coefficient of jet through the slit.2 Similarly in
tip and the shroud. Predicted values agreed well with ex- cases of leakage through a tip clearance of c, it is proper to
perimental data of seven axial impellers as well as data of two assume that the blade loading or the pressure difference across
types of centrifugal blowers in the literature. More com- a blade is not effective on a width of (2a- l)c from the blade
parisons with experimental data of high-pressure-ratio cen- tip. That is, the effective loading width of the blade is
trifugal compressors are described. b-(2a-\)c.
The performance of a compressor is influenced by the tip Near the tip of blade where the blade loading is smaller the
clearance in two ways: change of the input power, and in- mean through-flow velocity between a pitch of blades is also
crease of the pressure loss, but the former was not discussed in smaller in proportion to the blade loading; as a result the flow
[5]. In this paper an attempt is made to correlate the change of rate through the channel of impeller is also reduced or the ef-
input power to the leakage flow, and the flow model in [5] is fective width of flow passage is less than the blade width,
slightly modified so that the leakage flow, the pressure loss, although the physical passage width is made larger by the
and the change of input power due to tip clearance are clearance. That is, for a given flow rate the mean velocity be-
interrelated. tween blades is increased as follows
Regarding experimental study it should be noted that varia- Aw b c
tion of performance due to tip clearance is not large and an er-
ror of a fraction of one percent in head H or efficiency ij may w0 b-(2a-l)c b
mask the correct relationship between AH or Arj and the tip- The input head of a centrifugal impeller is usually expressed
clearance ratio X2. On the other hand it is very difficult to as
evaluate the total pressure correctly at the exit of impeller
where the velocity is severely distorted and periodic. If the i/-,/2=l-A:-0cot(3w (3)
total pressure is measured at the exit of the diffuser, data are <f> in the above equation is the dimensionless meridian compo-
influenced by the performance of the diffuser; therefore, the nent of velocity, and variation of x//,- due to tip clearance
difference of stage efficiencies of a compressor with two dif- c2/b2 = X2 is expressed as
ferent values of tip clearance is not always equal to the dif- Ai/-/2AX2 = - (2a - l)</>cot /362 (4)
ference of impeller efficiencies unless the pressure loss of the
diffuser is identical. Since the diffuser performance is sensitive It has been experimentally recognized [6] that A^,/AX2 varies
to the inlet condition to the diffuser, it is important that cor- almost in proportion to cot /3 M , but Ai//, was not exactly zero
responding flow conditions of a compressor with different for a radial impeller. It is presumed that equation (4) does not
values of tip clearance are chosen so that the pressure loss of perfectly express the mechanics of flow, but in practice equa-
the diffuser is identical. tion (4) is acceptable for most cases.
The contraction coefficient a varies depending upon the
geometry of the slit, and a is usually between 0.6 and 0.8. In
Modification of the Previous Theory on Tip-Clearance
some literature Ai/',/AX2 is experimentally measured, and we
Loss can estimate the contraction coefficient a using the data and
If there is a clearance between the edge of a loaded blade equation (4). If the value is between 0.6 and 0.8, the experi-
and a wall, near the edge the pressure difference across the ment supports equation (4) and the hypothesis behind it.
blade is less than the ordinary value without clearance, and the However, equation (4) should not be used to estimate a for
defect of pressure difference is related to contraction of cases of very small value of <> / cot pb2, because a small uncer-
leakage flow through the clearance. At the same time the tainty in Ai/-,/AX2 results in a large error in a.
pressure distribution is related to the input head of the im- Flow Model. It is experimentally known that variation of
peller and also to the pressure loss due to the tip clearance. blade loading with respect to flow rate is easily predictable
The relationship among these variables is discussed here and based on equation (3) or one-dimensional flow analysis, i.e., it
consequently the previous theory on the tip-clearance loss is is not influenced much by blockage in the impeller. In cases of
slightly modified.
Contraction Coefficient of Leakage Flow and Blade
Loading of Impeller. In cases of flow through an orifice or a This relation is easily derived using a momentum equation and a Bernoulli
equation. In many textbooks it is assumed for simplicity that the wall pressure is
slit of width c on a big wall, the pressure along the wall falls not changed by the flow, then ( 2 a - l)c = 0 and a— 0.5. In reality a = 0.61 for an
toward the slit, and it is easily proved that the defect of the orifice with a sharp edge.

Nomenclature

a = coefficient a = contraction coefficient of

b = blade width leakage flow
c = tip clearance /3 = flow angle, from circumference Subscripts
G = mass flow rate & = blade angle p - pressure surface
k = slip coefficient e = pressure recovery coefficient s = suction surface
m = meridional distance along V = efficiency 0 = values for zero clearance
shroud X = tip clearance ratio = c/b2 1 = impeller inlet
P = pressure P = density of fluid 2 = impeller exit
r = radial distance from axis a = effectiveness of flow area
U = tip speed at impeller exit * = flow coefficient at impeller exit Superscripts
V = component of leakage velocity tf = pressure coefficient of impeller ' = due to leakage
normal to blade +, == pressure
work input coefficient " = due to pressure gradient on an-
w = mean through-flow velocity be- loss coefficient due to
+, tip clearance nular clearance
tween a blade pitch '" = due to pressure gradient on
z = number of blades Q = angular velocity of impeller blockage in channel

56/ Vol. 109, JANUARY 1987 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo

 shrouded impellers or open impellers with very small tip replaced by unity, in equation (7) w0 is changed to w0/a, and
clearance, the effective area of blade to support the blade in equation (8) the first term u/2U in the parentheses is
loading is n o t influenced by the blockage, a n d the pressure dif- multiplied by a, while in equation (9) the term da/dm is added.
ference across a blade is n o t influenced by the blockage either. They are reproduced as follows
Usually the pressure difference across a blade is expressed as
follows v = ~J2(ws — wp)w0/a (7)

(PP -PSVP = (w? - w2p)/2 = (ws - wp)(ws + wp)/2

(""2 pa\ / v y
= (w, •wp)w (5) fr- irr24> J mi | 0 2 s m Pb ^ U / (—
\2U
where w is the mean velocity between a blade pitch; it is
2ir wn
evaluated disregarding blockage in the impeller. -*-) dm (8)
In cases of open impellers with a finite tip clearance, the ef- all dm /
dm
fective width of a blade becomes narrower due to leakage a n d
2 f'"2 P r\w0 . r w0 / 1 dw. o
the flow model is equivalent to an impeller with a narrower
M+ fi = — 77-sin 0 - 2 « e — — ( —
width by a factor of a, where 1 — a —(2a— \)c/b. As a result, /2V "I Pi U dm
r24> Jmj p2 au <- ail \ol
the through-flow velocity is w=wQ/a, and the effect of
blockage on the blade loading is different from that in or- w0 da
da \ 2(3 w n dr Wo dr
cos /3- cos P- [dm
dinary cases without leakage.
2
aU dm aU 2
dm a2 (fir dm
The blade loading of a n impeller is related to the change of (9)
angular m o m e n t u m of flow as follows
The change of pressure coefficient \j/ and the change of im-
(Pp -ps)rabzdm = pw(ws — wp)robzdm peller efficiency r\ due to a change of tip clearance X2 are
related to these parameters in the above equations and the
= 2%rbapw sin /3 (r2Q-rw cos (3)dm relations are presented in equations (10) and (11) in [5]. They
dm are
Replacing w = w0/a the above equation becomes (10)
2irr . r „ dr w0 „ da
w
s-wP= dm + —f- cos 0 - dm
sin /3120—7—
1 w
vo =i-(^o/^/o). •n = i-Wh + i'iyi'i
— If ° cos p- a dr - + cos ,dwt - w 0 sm /3- tin ~ V'i
a \ r dm dm dm •)} <» - + (l-i?o> (11)
^,-\ 2
1^2 ^/^2
This equation is identical to equation (12) of [5].
where \j/h is the hydraulic pressure loss other than the tip
In [5] it was assumed that the effective width of the blade
clearance loss, and the subscript 0 indicates the values for the
was b while the m e a n through-flow velocity was w0. In the
case of zero clearance. i/</l0 - \j/h is almost zero but it may be
present flow model the effective width is ab while the mean
negative at off-design conditions.
through-flow velocity is w0/a. T h a t is, according t o equation
(5) in the present flow model the pressure difference across a These parameters are calculated a n d listed in Table 1 for the
blade is larger by a factor of 1/a but the blade force as a whole experimental cases in the following chapter. In the first four
is identical to the model of [5]. Because of the larger pressure lines of Table 1, the effects of flow rate at a constant speed are
difference across a blade, the leakage velocity v is larger by a demonstrated for two compressors, where the parameters
factor of 1/Va. Since a is usually larger t h a n 0.95, the leakage were calculated for the m a x i m u m and the minimum tip
flow may be increased by about 2 percent based on the clearances at the specified identical flow rate and the dif-
modified flow model providing that the contraction coeffi- ference of the parameters was divided by the difference of the
cient is identical. T h a t is, the change of flow model hardly in- tip clearances AX2 at each flow rate. In the last six lines, the ef-
fluences the results of prediction, but in the new model the fects of shaft speed are demonstrated for two compressors,
blockage factor is directly related to the contraction coeffi- where the parameters were calculated for the m a x i m u m a n d
cient of the leakage flow a n d there is less freedom to adjust the minimum tip clearances at slightly different flow rates so
prediction. that the diffuser efficiencies were identical in the experiments
and the difference of the parameters was divided by the dif-
Working Equations. Working equations based on the new ference of the tip clearances AX 2 . The listed flow rates were the
flow model are a little bit different from those in [5]. In [5] one at the minimum clearance. The flow rate was slightly less
equations (1), (2), and the two equations in between, a is at the m a x i m u m clearance.

Table 1 Predicted pressure loss and efficiency drop due to tip clearance

a, G, Mi M'i Ml" (r,0-V)Mi — AJ;

Compressor percent kg/s
AX2 AX2 A\ 2 l/<;AA2 AX2

A 100 0.86 0.166 0.224 0.013 0.024 0.296

Fig. 7 100 0.80 0.162 0.193 0.009 0.020 0.26
C 90 0.86 0.204 0.173 0.012 0.025 0.302
Fig. 13 90 0.82 0.203 0.165 0.011 0.023 0.287
A 100 0.86 0.164 0.211 0.011 0.024 0.285
Fig. 5 80 0.574 0.126 0.132 0.004 0.028 0.209
50 0.249 0.110 0.073 0.000 0.019 0.148
B 90 0.826 0.121 0.120 0.003 0.029 0.221
Fig. 11 70 0.516 0.103 0.082 0.000 0.032 0.179
50 0.280 0.099 0.060 0.000 0.032 0.141

Journal of Turbomachinery JANUARY 1987, Vol. 109/57

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo

 0.72
0.70 0.80 0.90
Equivalent mass flow r a t e , kg/sec
Fig. 2 Change of input head due to tip clearance, Compressor A [7]

0.05
A^2
Fig. 5 Effect of shaft speed on tip-clearance loss, Compressor A

0.82

0.78
^ ^ %
0.2 0.4 0.6 0.8 1.0
° Equivalent mass flow rate, kg/sec dynamic pressure
Fig. 3 Change of stage pressure ratio due to tip clearance, Com- is reevaluated
pressor A [7] 0.74 I I
0.70 0.80 0.90
o 7.6 Clearance. Percent at design Equivalent mass flow rate, kg/sec
D 115 percent equivalent speed
A 16. S
- •' 3.7 <o* 10
°
Fig. 6 Change of impeller efficiency due to tip clearance, Compressor
ncy

s Surg* 80 90 A [71
5 0.75 60
u
M-l
50
J*\/ ^1/y sl^^aTvf curately measured. However, if the pressure loss in the dif-
0)
fuser is not equal for the tests with different values of tip
Dverall compressor

- clearance of the impeller, the variation of impeller perfor-

H
o

J i? 1 mance due to tip clearance can not be evaluated from the stage
ss 0^ performances. In order to minimize variation of the diffuser
s
II performance in different experiments, the inlet flow angle
- must be far from the angle for stall and far from the angle for
i

1 1 i 1 " I 1 1
choke, and furthermore the flow rate must be adjusted in dif-
1
0.2 0.4 0.6 0.8 1.0 ferent experiments so that the inlet flow angle to the diffuser is
Equivalent mass flow rate, kg/sec identical.
Fig. 4 Change of stage efficiency due to tip clearance, Compressor A As the tip clearance increases, the blade loading and the
[71 tangential velocity component at the impeller exit decrease
while the radial component increases because the gas density is
lower and the blockage is larger, consequently the direction of
Comparison With Experimental Data flow tends to be away from the tangential direction.
Therefore, in cases of high-pressure-ratio compressors, in
Accuracy of Total Pressure. The relative flow pattern at order to keep a constant flow angle at the diffuser inlet, the
the impeller exit is not axisymmetric due to blades and the pat- mass flow rate should be slightly reduced as the tip clearance
tern rotates with the impeller; therefore, the velocity relative increases.
to a fixed probe changes the magnitude and the direction
periodically, and the measured time-mean total pressure is not Test of Compressor A. A centrifugal compressor was
equal to the mass-averaged total pressure. The axial distribu- tested with four different values of tip clearance [7]. It was
tion of velocity is also distorted. That is, the meridian compo- designed for a pressure ratio of 6 and a specific speed of 0.769
nent of velocity is very low in a thin layer near the shroud due at 80,000 rpm. The impeller diameter was 136.5 mm, the exit
to the effect of tip clearance of the impeller, and the layer width was 4.67 mm, and there were, 24 blades with an exit
changes the thickness with the tip clearance; therefore, it is blade angle of 59.2 deg from circumference.
quite difficult to measure the mass-averaged total pressure. Variation of the input head due to change of the tip
The wall static pressure at the exit of the impeller may be the clearance is presented in Fig. 2 where four straight lines are
best parameter to accurately demonstrate the change of im- prediction based on equation (2) assuming that a = 0.655 or
peller performance due to tip clearance, but it is seldom the blockage factor I-a = ( 2 a - 1 ) X2 = 0.31 X2. The as-
presented in the literature. sumed value of k in equation (3) is 0.0258 which is smaller
than the value usually adopted. That is, in this experiment the
Corresponding Flow Rates of a Compressor With Different input power is considerably larger than what is usually ex-
Values of Tip Clearance. At the exit of diffuser, the flow is pected and it may be the cause for the low efficiency of the
almost steady and the dynamic pressure is small; therefore, the impeller.
stage performance including the diffuser can be easily and ac- Figures 3 and 4 are characteristic of the compressor with a

58/Vol. 109, JANUARY 1987 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo

 Percent of design
Clearance,
percent equivalent speed
G(kg/s)Expt Predicted //•/ 100
0.86 O 3.9 Biial tip clearance at design speedl
0.84 A /'/. / 6.4
0.82 •
- ///' 8.9
0.80 O

k=0. 0258
~/0*
/'A'
11.3
90

a=0.655 y*
' ///>
/#' A-' 3 -

- w 80000 rpm
(100 %)
0.2 0.4 0.6 0.8 1.0
7 ' i
° Equivalent mass flow rate, kg/sec
0.05 0.10
AA2 Fig. 9 Change of stage pressure ratio due to tip clearance, Com-
pressor B [8]
Fig. 7 Effect of flow rate on tip-clearance l o s s , Compressor A

with a vaned diffuser the flow range is usually limited by stall

0.78T --^.-v
and choke of the diffuser and the flow range is considerably
^^>\
narrower than the flow range of the impeller itself. Therefore,
an impeller with backward-leaning blades is often tested with a
^^^^v. vaneless diffuser in order to examine the characteristics of the
Clearance impeller in the entire range of flow rate. The impeller effi-
£ 0.74 —
- percent P r e d i c t e d
o 3.7 ciency of Compressor A is presented in Fig. 6 where the total
o 5.7 k=0.067 c9 pressure at the exit of impeller was estimated as the sum of the
A 7.9
- a 9.6 a=0. 74 & measured static pressure and the estimated dynamic pressure
assuming that the flow was uniform in the width which
1 1 i 1 1 1 covered the blade and the clearance. In reality, blockage in-
0.70
0.84 0.92 1.00
Equivalent mass flow race, kg/sec
creases with the tip clearance; therefore, in Fig. 6 the total
pressure and the impeller efficiency were underestimated due
Fig. 8 Change of input head due to tip clearance, Compressor B [8] to the assumptions of wider exit width and of no blockage.
If the efficiency of the vaned diffuser is estimated based on
the difference between the maximum stage efficiency in Fig. 4
vaned diffuser. It is clear that the tip clearance influences the and the impeller efficiency at that flow rate in Fig. 6, the dif-
compressor performance a great deal, but variation of the fuser efficiency is higher for the case with a larger tip
pressure ratio at a low speed is not apparent because the clearance. This means that a better pressure recovery is
pressure ratio itself is very small. Therefore, comparison of achieved when the blockage at the diffuser inlet is larger. The
experimental data with prediction is limited to only efficiency. trend is against our knowledge on diffusers and the cause is at-
Regarding corresponding flow rates for different values of tributed to the unsuitable way for estimating the dynamic
tip clearance, they are chosen near the flow rate for the max- pressure at the exit of impeller in Fig. 6.
imum efficiency and furthermore the flow angle at the exit of Regarding the velocity distribution at the exit of impeller it
impeller is made identical regardless of the tip clearance. may be proper to assume that there is no outward flow from
These corresponding flow rates are indicated by arrows in Fig. the annulus of tip clearance and the outward flow is uniform
4. only in the effective width of impeller. Assuming the above
velocity distribution, the total pressure at the exit of impeller
Influence of Shaft Speed. The efficiency of the com- was re-evaluated from the data in Fig. 6, and the modified ef-
pressor with the minimum tip clearance is used as the base3 ficiencies were presented as dotted lines. Using the modified
and the decrement of the maximum efficiency due to the incre- impeller efficiency, the relationship between -Ai? and AX2
ment of tip clearance is experimentally determined at six dif- was examined for four flow rates between 0.80 and 0.86 kg/s
ferent shaft speeds, which are indicated as different symbols in and the results are presented in Fig. 7 together with the
Fig. 5 while the predicted relationships are presented as predicted lines.
straight lines. Some of the experimental data are not on the Although the estimated total pressure may not be accurate,
predicted straight lines, but they are located near the cor- both the experimental data and the prediction demonstrate
responding straight lines. that the influence of tip clearance on the efficiency becomes
According to Fig. 5 for a given increment of tip clearance weaker as the flow rate is reduced. One of the reasons is that
the decrement of measured efficiency becomes less as the shaft the blade loading is reduced with the flow rate and the leakage
speed is lower, and the trend is also predicted. At a reduced flow rate is also reduced. That is, Ai/-//AX2 is smaller. More
speed the ratio p/p2 is relatively large along the shroud, conse- important is that at a reduced flow rate deceleration in the im-
quently w/U is small compared to the value at the design peller is large only near the inducer, but the tip clearance of in-
speed. According to equations (4), (5), and (6) a small value of ducer is not much changed by axial movement of the shroud in
w/U results in a small value of pressure loss similar to the case the test, while deceleration along the rest of shroud is less.
mentioned in the following section. Consequently the tip-clearance loss based on deceleration
(A\p,"+ Ai/</")/AA2 is small at a reduced flow rate.
Influence of Flow Rate. If a centrifugal impeller is tested
Compressor B [8]. This compressor was designed for a
The tip clearance did not remain constant due to flowering of the impeller pressure ratio of about 6 and a specific speed of 0.718 at
toward the shroud as the speed was increased. Since the compressor with the 68,384 rpm. The diameter and the exit width of the impeller
minimum tip clearance is used as the base at each speed, AA2 is the increment of were 161 mm and 5.16 mm, respectively, and it had 19 blades
tip clearance relative to that of the base compressor, and the variation of tip with an exit blade angle of 55.3 deg.
clearance of the base compressor due to the rotative speed is excluded for the
comparison providing that — Ai) is almost proportional to AA2. Symbols in Fig. 8 show the relationship between the tip-

Journal of Turbomachinery JANUARY 1987, Vol. 109 / 59

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo

 Clearance, percent
Q 1 9 taial tip clearance at riesign speed! 31895 rpm
O 6.4 (90 7.)

0.65 3

Running Clearanc
Datum
Datum+0
Datum+0

0.2 0.4 0.6 0.8 1.0 n. 2.0 _1_

0.5 0.75 1.0
Equivalent mass flow race, kg/sec Equivalent mass flow rate, kg/sec
Fig. 10 Change ol stage efficiency due to tip clearance, Compressor B Fig. 12 Change of compressor performance due to tip clearance, Corn-
el pressor C [9]

G(kg/s)Expt.Predicted k=0.072S
0.86 O a=0.8
0.84 A
0.82 O •

A\2 AA 2
Fig. 11 Effect of shaft speed on tip-clearance loss, Compressor B Fig. 13 Effect of flow rate on tip-clearance loss, Compressor C

ciency is predicted at three flow rates in Fig. 13 and compared

clearance ratio and the input head of the impeller which was
with symbols which are deduced from the experimental data
tested with a vaneless diffuser, while the four straight lines are
in Fig. 12. Good agreement is observed.
predicted assuming k = 0.067 and a = 0.74 in equations (3)
and (4). It is puzzling that the input power of compressor with Compressors D and E. In [5] the theory was applied to a
X2 = 0.079 is smaller than that with X2 = 0.096 for the identical radial compressor D and a compressor E with 45 deg
flow rate. backward-leaning blades, and the experimental relationship
Figures 9 and 10 are performances of the compressor with a between the decrement of efficiency and the increment of tip
vaned diffuser. At 100 percent speed the surge flow rate was clearance was predicted fairly well by the theory in the range
close to the choke flow rate and the flow condition at the max- of X2 < 0.1. Those experiments covered a wide range of X2 and
imum efficiency was not identical for cases with different they demonstrated that ~drf/d\2 became smaller as X2 was
values of tip clearance; therefore they are omitted from the larger. In the present paper, the theory was modified, but as it
present discussion. was mentioned before, the predicted inclination - AT;/AX 2 did
It is embarassing that in Fig. 9 the compressor with not change more than a few percent which was smaller than
X 2 = 0.064 has a higher pressure ratio than the compressor the uncertainty of experimental data. In order to predict the
with X 2 = 0.039, while in Fig. 10 the latter has a higher effi- tip clearance loss in a wide range of X2, for estimating 1/7"and
ciency than the former has. The strange relation was not men- 1/7"' it is necessary to include the effect of reduction of the
tioned at all in the original paper and it could be a simple error radial pressure gradient due to the tip clearance.
on symbols. Otherwise, the inconsistent data in Figs. 8, 9, and
10 mean difficulty of reliable measurement which was ac- Conclusions
curate enough to compare the performances of a compressor
with different values of tip clearance. The contraction coefficient of leakage flow through a tip
The efficiency of the compressor with the minimum tip clearance is related to the blade loading and to the blockage of
clearance is used as the base, and the decrement of peak effi- flow in the channels between blades. Using the relationship,
ciency due to the increment of tip clearance is indicated in Fig. the change of input head and the change of pressure loss are
11 using different kinds of symbols for each shaft speed while quantitatively related to a change of tip clearance, and the
the prediction is indicated as different kinds of line for each authors' previous theory on the tip clearance is modified
shaft speed. slightly. The theory not only predicts the relationship between
the tip clearance and the pressure loss at the design condition,
Compressor C [9]. This compressor was designed for a but also correctly predicts how the relationship varies at off-
pressure ratio of about 3.6 and a specific speed of 0.485, and it design flow rates and at off-design shaft speeds. They are sum-
was tested at 31,895 rpm which was 90 percent of the design marized as follows:
speed. The characteristic curves of the compressor with three 1 The tip-clearance loss is almost proportional to the tip-
different values of tip clearance are presented in Fig. 12 clearance ratio in the range of X2 < 0 . 1 .
without indicating experimental data. In the prediction, the 2 The ratio -A?;/AX 2 becomes somewhat smaller as the
contraction coefficient a is assumed as 0.80 based on the ex- flow rate is reduced.
perimental data of the input head, and the relationship be- 3 If the tip-clearance ratio c2/b2 is equal, compressors
tween the increment of tip clearance and the decrement of effi- with a higher design pressure ratio have smaller values of

60/ Vol. 109, JANUARY 1987 Transactions of the ASME

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo

 - Arj/AXj, because near the inlet the blade height is larger and Aerodynamics," ASME JOURNAL OF ENGINEERING FOR POWER, Vol. 95, July
the local tip-clearance ratio c/b is smaller. 1973, p. 251.
2 Eckert, B., and Schnell, E., Axial- und Radial-Kompressoren, 2nd ed.,
4 If a high-pressure-ratio compressor is tested at a reduced Springer-Verlag, 1961, pp. 192, 357.
shaft speed, -Aij/AX2 is smaller than the value at the design 3 Pfleiderer, C , Die Kreisel Pumpen, 5th ed., Springer-Verlag, 1961, p. 99.
speed. 4 Harada, H., "Performance Characteristics of Shrouded and Unshrouded
Impellers of a Centrifugal Compressor," ASME JOURNAL OF ENGINEERING FOR
GAS TURBINES AND POWER, Vol. 107, Apr. 1985, p. 528.
Acknowledgments 5 Senoo, Y., andlshida, M., "Pressure Loss Due to the Tip Clearance of Im-
peller Blades in Centrifugal and Axial Blowers," ASME JOURNAL OF ENGINEER-
When the authors submitted the manuscript of [5] to ING FOR GAS TURBINES AND POWER, Vol. 108, Jan. 1986, pp. 32-37.
ASME, one of the reviewers suggested that the authors should 6 Ishida, M., and Senoo, Y., "On the Pressure Losses Due to the Tip
apply the theory to data in [7, 8, 9]. The titles of these papers Clearance of Centrifugal Blowers," ASME JOURNAL OF ENGINEERING FOR
did not properly indicate the contents in spite of the com- POWER, Vol. 103, Apr. 1981, p. 271.
7 Klassen, H. A., Wood, J. R., and Schumann, L. F., "Experimental Perfor-
prehensive work on tip clearance. Without his suggestion these mance of a 13.65-centimeter-Tip-Diameter Tandem-Bladed Sweptback Cen-
references were unknown to the authors and the present paper trifugal Compressor Designed for a Pressure Ratio of 6," NASA Technical
was not possible. The authors are grateful to the reviewer and Paper 1091, Nov. 1977.
also to Dr. C. Hah who served as the coordinator for review of 8 Klassen, H. A., Wood, J. R., and Schumann, L. F., "Experimental Perfor-
mance of a 16.10-centimeter-Tip-Diameter Sweptback Centrifugal Compressor
the paper at that time. Designed for a 6:1 Pressure Ratio," NASA Technical Memorandum X-3552,
June 1977.
References 9 Beard, M. G., Pratt, C. M., and Timmis, P. H., "Recent Experience on
Centrifugal Compressors for Small Gas Turbines," ASME Paper No.
1 Pampreen, R. C, "Small Turbomachinery Compressor and Fan 78-GT-193, Apr. 1978.

Journal of Turbomachinery JANUARY 1987, Vol. 109/61

Downloaded From: http://turbomachinery.asmedigitalcollection.asme.org/pdfaccess.ashx?url=/data/journals/jotuei/28581/ on 04/06/2017 Terms of Use: http://www.asme.org/abo