THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

345 E. 47 St., New York, N.Y. 10017

83-GT-3
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.
^^L Released for general publication upon presentation Full credit should be given to ASME,
the Technical Division, and the author(s). Papers are available from ASME for nine months
after the meeting.
Printed in USA.
Copyright © 1983 by ASME

LOW-SOLIDITY TANDEM-CASCADE DIFFUSERS

FOR WIDE-FLOW-RANGE CENTRIFUGAL BLOWERS

Yasutoshi Senoo
Hiroshi Hayami
Hironobu Ueki
Research Institute of Industrial Science.
Kyushu University, Kasugashi, Fukuoka, 816, JAPAN

ABSTRACT NOMENCLATURE

The pressure recovery of a low-solidity circular b = width of diffuser or impeller exit

cascade diffuser of a centrifugal blower was predicted CL = lift coefficient
by a simple method combining a theory of circular Cp = pressure recovery factor
cascade diffusers and that of vaneless diffusers and it G = mass flow rate
was compared with a series of experiments. Furthermore e = chord length
the stall limit of the diffuser was studied. In order to R = radius ratio, R = r/r o
improve the performance further, a series of tandem- r = radius
cascade diffusers were tested. In these diffusers, the rp = radius of impeller
front row of the cascade was designed for a small flow t = blade pitch
rate while the rear row of the cascade was designed for U = tip speed of impeller
a large flow rate so that the tandem cascade would W = velocity
accomplish good pressure recovery in a wide range of Ul = attack angle
flow rate. Experimental results showed that the operat- (3 = flow angle relative to tangential direction
ing range was as wide as that of a vaneless diffuser and fi = stagger angle
the pressure recovery was excellent at a small flow rate AP = pressure rise
while it was somewhat better than that in a vaneless K = correction factor for cascade
diffuser at a large flow rate. P = density
G = cascade solidity. o = e/t
(P = flow coefficient, 0 = G/2rtpbr c U
* = pressure coefficient, ' = 2AP/9U 2.
Subscripts
c = cascade
e = exit
ir^peZ °ati i = inlet or isolated airfoil
1 eaa ^ er exj f^ t s = static
t = total
ng
INTRODUCTION
edge
rW °f cas cade
At the exit of a centrifugal impeller about one
u half of the energy input to the fluid remains as kinetic
to
energy, which must be efficiently recovered to pressure.
There are several types of diffuser which can achieve a
high pressure recovery, where flow is guided by many
d}ff^ser exit vanes or similar structures. Since the direction of flow
out of the impeller varies with the f]ow rate, however,
the flow separates from the vanes at off-design condi-
Flow rate tions, and the available flow range is narrow.
In order to demonstrate the local deceleration rate
in a diffuser, the variation of velocity with respect to
the flow rate at several representative positions is
Fig.l Variation of local velocity with respect to presented in Fig. 1. Since the circumferential component
flow rate is the major part of the velocity at the exit of an im-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 peller and at the leading edge of vanes, the velocity EXPERIMENTAL APPARATUS AND PROCEDURE
does not change the magnitude much depending upon the
flow rate at these positions. The mean velocity at the Circular cascades were tested as a part of the dif-
throat and at the exit of the diffuser varies almost fuser of a centrifugal blower. The centrifugal impeller
proportionally to the flow rate. Therefore, the de- 152 [m,m 3 /min,rpm] in specific speed or 0.0590 in non-
celeration rate from the leading edge to the throat be- dimensional specific speed was run at 3,500 to 5,000
comes very large as the flow rate is reduced • and if rpm. The diameter of the impeller,2r o , was 500 mm and
the deceleration rate exceeds a critical value the dif- the exit width of the impeller, b, was 14 mm. The impel-
fuser stalls. ler had 18 blades with a backward leaning angle of 41.5
On the other hand at a large flow rate the throat deg. At the exit of impeller there was a diffuser which
velocity is larger than the velocity at the leading consisted of two parallel walls 14mm apart from each
edge, as the throat velocity increases in proportion to other and 1,000 mm in diameter. The circular cascade was
the flow rate. The high velocity at the throat not only located between the two parallel walls so that the lead-
increases the friction loss but also increases the dif- ing edges of the blades were about 270 mm from the axis.
fusion loss downstream and the pressure recovery is re- A U.S.A.35-B airfoil was adopted as the original airfoil
duced considerably. At a high inlet Mach number, section to design the circular cascade, because this
obviously the throat limits the choke flow rate. That airfoil has a high maximum lift coefficient and also has
is, both the minimum flow rate and the maximum flow rate a mild characteristic for blade stall (3) at low Mach
are decided by the flow behavior at the throat, or the number. The characteristics of the airfoil are shown in
throat is the cause of difficulty with respect to the Fig.2.
flow range of a diffuser. The airfoil was used to form two linear cascades
If it is desired to design a vaned diffuser with a where the stagger angle was 71 deg and the solidity was
wide flow range, an obvious conclusion is to eliminate 0.69 or 0.35, then the linear cascades were conformally
the throat or to design a diffuser of low-solidity transformed to circular cascades with eleven blades.
cascade. It looks as if a diffuser of low-solidity They are identified as AO type and BO type respectively.
cascade cannot achieve a high pressure recovery, but During the performance test of the blower the stagger
obviously the pressure recovery is higher than that of a angle of the A-type cascade was reduced by one or three
vaneless diffuser, which is widely used. deg. They are identified as Al or A3 type. Similarly the
In order to design a circular cascade for a dif- stagger angles of El type and B1 type cascade are re-
fuser of centrifugal blower, there are many problems to spectively one degree smaller and larger than that of
overcome. First of all the aspect ratio of blade is very the BO type cascade. Later two blades were added to the
small and the side-wall boundary layer may significantly A3 cascade to increase the solidity to 0.82. The cascade
change the performance of a two-dimensional circular is identified as A3'.
cascade. The geometry of a circular cascade and the In cases of tandem cascade, B-type cascade was used
velocity distribution are conformally transformed to as the front row and the eleven blades of A-type cascade
those of a linear cascade for inviscid flow. However, were arranged closely behind each blade of the front
since the exit/inlet velocity ratio of the circular row. Therefore, the blade geometry of the rear row is
cascade varies in reciprocal proportion to the not identical to the conformally transformed geometry of
exit/inlet radius ratio, the pressure loss and the stall U.S.A.35-B airfoil.
limit of a two-dimensional circular cascade are differ-
ent from those of the conformally transformed linear
cascade. That is, the exit/inlet radius ratio of a
circular cascade is an additional parameter to the many
parameters of a linear cascade, and there are no re-
liable experimental data which may be applicable for de-
sign of a specific cascade diffuser.
In the literature tandem cascade diffusers are re-
ported to have achieved a high pressure recovery and de-
monstrated an operating range somewhat wider than that
of conventional vaned diffusers, but the solidity of
those cascade is large and the operating range is by no
means comparable to that of vaneless diffusers (1)(2).
Although there are no rational design method on a tandem
cascade diffuser in the literature, the authors intend
to include a tandem cascade as a part of their study on
low-solidity circular cascades for wide-flow-range
blowers.
In cases of centrifugal compressors where the Mach
number at the exit of impeller is high, the diffuser
must be designed considering the influences of com-
pressibility. It is presumed that the leading edge of
blades must be thin to secure good pressure recovery at
the design condition, while thin blades are easy to
stall as the incidence angle is increased. As a result
the flow range of the cascade for stable operation would
be narrower as the Mach number is larger. The authors
hope that the present study serves as a first step for
development of wide-flow-range diffusers for subsonic
flow.
Fig.2 Characteristics of U.S.A. 35-B airfoil (3)

- 2 -

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 in Fig.4. The ordinate is the pressure coefficient, that
is, the ratio of the pressure rise to the mean dynamic
115 pressure of flow at the diffuser inlet or at the impel-
ler exit. The experimental data of the vaneless diffuser
are indicated in Fig.4 as open circles and the predicted
pressure recovery factor based on one-dimensional an-
Al alysis using a skin friction coefficient of 0.0075 is
also plotted in Fig.4 as a full line. The agreement is
satisfactory in both cases 0 =0.12 in Fig.4 (a) and
p=0.24 in Fig.4(b). This value of 0.0075 was adopted as
the skin friction coefficient for the side wall in cases
of cascade diffuser.
For prediction of the performance of a cascade dif-
fuser, the diffuser is divided into three zones. They
115 are the front vaneless zone, the circular cascade zone
and the rear vaneless zone. The flow at the front and
rear vaneless zones was analysed by a conventional one-
dimensional method on vaneless diffuser (4). The flow at
the circular cascade zone was analysed based on a
B1E0

I'll
Cascade
Fig.3 Geometry of test circular cascades A
a
U E B1 E3
The solidity of the rear blade row was 0.63. and
the blade row was identified as EO when stagger angle .••- Al
was 64 deg. Three kinds of tandem cascade of
BIEO, BIE3 and B E3 were tested where E3 means that the 05
stagger angle is 67 deg, or three deg larger than E0.
The circumferential position of the rear blade row rela- Vanel ess
tive to the front row was varied to achieve the maximum
pressure recovery. The geometry of the Al cascade and
the best arrangement of B1EO cascade are shown in Fig.3.
The exit of the diffuser was opened to the
atmosphere. The flow rate was adjusted by a valve at the
inlet of a long straight suction pipe and it was
measured with an orifice flow-meter in the pipe. The
0 1.0 1.5 R 2.0
suction pipe was followed by a plenum chamber and air (a) pQl2
was accelerated by a factor of 8.77 from the plenum
chamber to the intake of the blower.
The static and total pressure at the impeller exit 1.01 --
varied circumferentially due to the diffuser blades.
Therefore, the blower with a vaneless diffuser was
Cascade
n.
tested for various flow rates in advance, and the char- U A
acteristics of the impeller were estimated using the to-
tal pressure of main flow at 1.09r o and the wall static fF E B1E
-a—a
pressure at 1.1r o . It was assumed that the characteris-
tics of the impeller were not influenced by the dif- —_ - -i Al
fusers. In this assumption if the pressure loss of the 0.5
impeller is varied by changing the diffuser, the incre- „'• Vaneless
ment of pressure loss is attributed to the pressure loss
of the diffuser, providing that the input power to the
impeller is not modified by the diffuser blades. In the
present experiments, no measurement on the shaft power
was made. However, the above assumption was regarded re-
asonable since the mean value of the circumferentially
distributed wall static pressure at the impeller exit O v
1.0
was hardly changed by changing the diffusers. 1.5 R 20
(b) *=0,24
EXPERIMENTAL RESULTS AND DISCUSSION

Radial Distribution of Wall Static Pressure

The radial distribution of wall static pressure be-
tween the impeller exit and the diffuser exit is shown Fig.4 Radial distribution of pressure

-3-
3-

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 cascade correction method' of the characteristics of
isolated airfoil 2 and then it was modified to include
the effect of blade drag and the side-wall skin fric-
tion. 20
For the case of the Al cascade diffuser, the pres- Exp. Pred,
sure distribution was calculated and is plotted in Fig.4 Vaneless o
as a dashed line, while the closed circles are the ex-
perimental data. The predicted pressure is somewhat Al ---
higher than the experimental data at 4=0.24 and it is CL ^^ A3 A3 A ..........
lower at 0=0.12. A better agreement is possible if a A3' o ___
little modification is allowed for the cascade correc- ;-A1
tion factors.
1.5
For the case of the tandem cascade B1E3, the ex- Surg e limit
perimental data are shown in Fig.4 as open squares. The A3'
pressure predicted by the method mentioned above was
much lower than the experimental data. Therefore, each
A3^ `^
of the cascade correction factors was multiplied by a
y
factor of 1.4 to incorporate the mutual interference
effect of the two blade rows (see Appendix). The pre-
dicted pressure distribution is plotted in Fig.4 as a
chain line.

Characteristics of Blower
Single Cascade Diffuser. The characteristic curves
of the blower with a single-row circular cascade dif-
fuser are presented in Fig.6. The ordinate is the pres-
sure coefficient. =2dP/PU 2 , and the abscissa is the
flow coefficient, m =G/2npbr 0 U. The mean flow angle at
the diffuser inlet, p , is also indicated on the ab- 0.5
W `A 3\` '
Wi A3'
2.0

80
(3 deg 6 8 10 12 14 16 18 2 0 \\
1.5 70
0.1 0.2 0 0.3
60 °
SO ° Fig.6 Characteristic curves of blowers with single
1.0 a° row cascade diffusers
° 30°

0.5 scissa. The full lines of '' and 4't in the figure are
respectively the static pressure coefficient and the to-
tal pressure coefficient measured at the impeller exit
for the blower with a vaneless diffuser, and the other
n V'-lines are the static pressure coefficient at the exit
0 1 2 3 of various kinds of diffuser.
t/l Open circles in Fig.6 are the experimental data for
the case of the vaneless diffuser, while the predicted
k = (dC L /da) C /(dC T /da) i characteristic curve is drawn as a full line. Agreement
is satisfactory. Furthermore, closed circles, open tri-
Fig.5 Cascade correction factor of flat plate linear angles and open squares are the experimental data for
cascade (5) the cases of Al, A3 and A3' cascades, while the pre-
dicted characteristic curves are shown as a dashed line,
a dotted line and a chain line respectively. Good agree-
1 The correction factor for a linear cascade of flat ment is observed in the small-flow-rate range, but the
plate shown in Fig.5 (5) was adopted for the correction predicted pressure coefficients are a little higher than
of the gradient of lift coefficient vs. attack angle, the experimental data in the large flow rate range.
and the zero-lift attack angle of a cascade,doc, was The discrepancy would be diminished if the zero-
evaluated from the zero-lift attack angle of an isolated lift attack angle is reduced by one deg. Alternatively,
airfoil, dot, as follows (6): the discrepancy may be attributed to the decrease of the
duc - dut = 15n.(Q/t) 2 [-(h/Q)cos2F+ O.425(s/e )sin26'1 deg effective passage width due to development of the side-
where e/t is the solidity. F is the stagger angle and h/Q wall boundary layer. That is, for the case of Fig.6, be-
and s/Q represent the camber ratio and the thickness tter agreement would be achieved by reducing the scale
ratio of the airfoil respectively. of abscissa considering a blockage factor of 0.05 for
the prediction.
2 In the analysis it is assumed that CL=0.10(t(+5.2) At the left end of the characteristic curve, the
for all values of d in deg. According to Fig.2, this system of the test blower was too unstable to take data,
equation over-estimates C L for C L >1.3 for the isolated and it was assumed as the critical flow rate for surge.
airfoil. The critical flow rate of the Al cascade is almost equal

- 4 -

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 pattern at the inlet section indicates that all incoming
wall-streamlines near the suction side of a blade flow
back near the pressure side of the adjacent blade and
none of the wall streamline passes through the blade row
to the exit. The velocity vector in the side-wall
boundary layer continuously varies from the wall-
streamline to the main flow and the major part of fluid
in the boundary layer flows toward the pressure side of
the adjacent blade. If the pressure rise along the blade
is not very large, the accumulated wall boundary layer
on the blade pressure surface can flow downstream with-
out separation, thus the cascade does not stall. The
exit/inlet velocity ratio of 0.37 may correspond to the
maximum pressure rise along the blade pressure surface.
Such lateral movement of side-wall boundary layer is a
feature of flow in a cascade with a large stagger angle.
and the stall limit of airfoil is not penalized by the
side-wall boundary layer as demonstrated in Fig.6.

Tandem Cascade Diffuser. The experimental charac-

Fig.7 Oil-film pattern on side wall of single-row teristics of the blower with B1E3, B1E3 or B1EO tandem
cascade cascade diffuser are plotted in Fig.8 as different open
marks, while the predicted characteristics using the
modification factor for interference between blade rows
mentioned before are also plotted as different kinds of
to that of the vaneless diffuser; therefore, the surge
lines. Agreement is satisfactory except the data near
may have been caused by stall of the impeller. On the
the surge limit. All experimental data of 'k are higher
other hand, for the cases of A3 and A3' cascades, the than the full line which represents the case of the Al
critical flow rate is somewhat larger than that of the
cascde by 0.02-0.05 in the whole range of flow rate.
vaneless diffuser; therefore, the surge was induced by
stall of the diffuser.
The diffuser stall may be induced either by stall
of the blades or separation of the side-wall boundary
layer due to a steep pressure gradient. The cascade of Exp. Fred.
diffuser stall may be determined by increasing the num- 2.0 CL
ber of blades, because the loading of each blade is re- B1E3 0 —
duced while the pressure-rise across the cascade becomes
larger for a given inlet flow condition.
B1E3 ° ----
The lift coefficient, C L , of the blade and the B1EO
exit/inlet velocity ratio across the blade row, We/Wi,
are evaluated using a one-dimensional analysis and the
results are presented in Fig.6. In the cases of the Al
B1
or A3 cascades, the lift coefficient at the critical E3
condition of surge is 1.81 which almost coincides with 1.5 Surge limit ,\
the maximum lift coefficient of an isolated airfoil. B1r^
Therefore, the stall of blades may have been the cause B1 1
of diffuser stall. In the case of the A3' cascade the
lift coefficient of blade is only 1.65 at the critical
BTI
flow rate; therefore, it is not likely that the blades
were stalled, and the surge was probably induced by
separation of the side-wall boundary layer due to a
steep pressure gradient. 1.0
The exit/inlet velocity ratio of the cascade was
0.37 at the critical flow rate in the A3' cascade. In Y5
the case of the A3 cascade the velocity ratio, We/Wi, at
the critical flow rate was also 0.37. It is presumed
that this value is the critical velocity ratio for
separation of the side-wall boundary layer. If the
critical lift coefficient of blade in the circular
cascade is 1.8, equal to the maximum lift coefficient of
the isolated airfoil, the A3 cascade has achieved the
0.5
maximum pressure recovery with the minimum solidity. E3
The pattern of oil-film on the side wall in Fig.7 Wi B1E3 .1E3
demonstrates that the side-wall boundary layer moves B1E0
from the suction side of a blade to the pressure side of
the adjacent blade; consequently the boundary layer on
the suction surface of a blade is sucked by the sec- 0 p deg 6 8 10 12 14 16 18 2 0\
ondary flow along the side walls and separation of
boundary layer on the blade suction surface is delayed. 0 0.1 0.2 0.3
Therefore, a very high lift coefficient is possible in
spite of the extremely small aspect ratio of blades.
The oil-film pattern also demonstrates that the
side-wall boundary layer between blades flows downstream Fig.8 Characteristic curves of blowers with tandem
except in the inlet triangular zone. The oil-film cascade diffusers

- 5 -

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 That is, the tandem cascade diffuser is superior to the
cascade diffusers of single row in Fig.6.
Also included in Fig.8 are the three lines which
show the predicted exit/inlet velocity ratio We/Wi B1EO DIFFUSER
across the tandem cascade of the three diffusers. 0=0.10
Furthermore, for the case of the B1E3 cascade the ..r4 HUB SIDE
exit/inlet velocity ratios across the front BI blade row
and the rear E3 blade row as well as the predicted lift
coefficients, C L . of blades in the each blade row are
also shown. The predicted lift coefficient of the front
blade row exceeds 2.0 at the flow rate near the surge
limit. The value is larger than the maximum lift
coefficient of the isolated airfoil in Fig.2. The dis-
crepancy is based on the simplified assumption mentioned
in the foot-note 2. In Fig.8, the experimental data on
pressure rise 4' deviate from the prediction near the
surge limit. The discrepancy is partly based on the
over-estimation of the lift coefficient at a large
attack angle, but mostly based on deterioration of the
performance of the cascade due to three-dimensional
effect of flow near the surge limit.
The overall exit/inlet velocity ratio across the Fig.9 Oil-film pattern on side wall of tandem cascade
tandem blade row is 0.25 at the critical condition of
surge. This value is much smaller than that of a
single-row cascade and that of ordinary diffusers. That
is, the overall pressure rise across the tandem cascade than a vaneless diffuser does.
is too large for the accumulated side-wall boundary- The performance of the diffuser is well predicted
layer fluid to flow downstream along the pressure side by a simple method combining a conventional theory of a
of the blade. However, because of the slit between the circular cascade and that of a vaneless diffuser.
two blade rows, the accumulated boundary layer along the When two rows of a small-solidity cascade are
pressure side of the front blade is channeled into the arranged to form a tandem cascade, a very good pressure
main flow through the slit, and a new boundary layer is recovery is achieved for a wide range of flow rate.
developed on the blade of the rear blade row. The At the critical condition for stall of a tandem
pattern of oil-film in Fig.9 demonstrates accumulation cascade, each blade row achieves almost as much pressure
and disposition of the side-wall boundary-layer fluid rise as the critical pressure rise of a respective sin-
near the slit. gle blade row; consequently the overall exit/inlet
Because of the transverse flow of the side-wall velocity ratio is very small and a large pressure rise
boundary layer between the blades of the front blade is achieved by a properly designed tandem cascade.
row, the side-wall boundary layer at the inlet to the On the other hand, it is clear that, due to the
rear blade row is thin. Consequently, each blade row of interference between the two blade rows, the relation-
a tandem cascade can achieve a pressure rise almost ship between the lift coefficient and the attack angle
independently and the overall pressure rise is the sum for each blade row of a tandem cascade is considerably
of those two steps of pressure rise, in spite of that different from that for a cascade of single blade row.
the side wall is continuous for the two blade rows. That In order to design a tandem cascade diffuser, quantita-
is, the three-dimensional nature of the flow in the tive data on the interference between two blade rows are
side-wall boundary layer prevents the tandem cascade required and further experimental and/or theoretical
from stalling against a high pressure rise. study is desired.
The exit/inlet velocity ratio across the rear blade
row near the surge limit is 0.45 in Fig.8 which is a REFERENCES
little larger than that of a single-row cascade. For
predicting the curves in Fig.8 the cascade correction 1 Pampreen, R. C., "The Use of Cascade Technology in
factor of the individual cascade was multiplied by an e- Centrifugal Compressor Vaned Diffuser Design," Journal
qual value of 1.4 considering the interference between of Engineering for Power. Trans. ASME, Ser.A, Vol.94,
the two blade rows, so that the overall performance of No.3, July 1972, pp.187-192.
the tandem cascade agrees well with the experimental 2 Pampreen, R. C., "A Cascade Analogy of Vaned Dif-
data. In reality, the multiplication factor may not be fuser Influence on Centrifugal Compressor Stability and
identical for the two blade rows. If it is assumed that a Comparison between Quasi-Static and Transient
at Q>=0.09 near the surge limit the lift coefficient is Stabilities," Centrifugal Compressor and Pump Stability,
1.8 both for the front and the rear blade row, the pre- Stall and Surge, ASME, New York, March 1976. pp.167-177.
dicted performance agrees well with the experimental 3 Jacobs. E. N., and Abbott, I. H., "Airfoil Section
data and the exit/inlet velocity ratio across the rear Data Obtained in the N.A.C.A. Variable-Density Tunnel as
blade row is 0.4. These values are nearly equal to those Affected by Support Interference and Other Corrections, 0
for the case of a single-row cascade. NACA Rep.669, (1939).p.14.
4 Eckert,B., and Schnell, E., "Axial-und
CONCLUSION Radial-Kompressoren," Springer-Verlag, Berlin, (1961).
pp.387-395.
Using a centrifugal impeller with backward-leaning 5 Weinig, F., "Die Stromung um die Schaufeln von
blades, it is demonstrated that a diffuser with a Turbomachines," J. A. Barth, Leipzig, (1935).p.40.
small-solidity cascade has a wide range of stable opera- 6 Eckert,B., and Schnell, E., "Axial-und
tion comparable to that of a vaneless diffuser and the Radial-Kompressoren," Springer-Verlag, Berlin, (1961).
diffuser achieves considerably better pressure recovery p.260.

- 6 -

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 APPENDIX

Note on the Correction Factor for Tandem Cascade Table 1 Performance of tandem cascades with differnt
Let a tandem cascade be made by cutting each blade of interference factors,
a flat plate cascade 70 deg in stagger angle and 0.98 in 1=70 deg, 6f=0.35, 6r=0.63.
solidity in two parts at the location of about 1/3 from
the leading edge. When the spacing between the two blade
rows is very narrow case 1, the overall performance of
the tandem cascade is not different from that of the
original cascade providing that the effect of boundary B iBs ae CL,(front) CL(rear)
layer is disregarded. On the other hand, if the spacing
Case 1 10 - 18.60 - 1.24
between the two blade rows is very wide case 2, the
Case 2 10 12.06 15.96 1.07 0.89
overall performance of the tandem cascade can be
Case 3 10 13.39 18.57 1.67 1.04
evaluated as simple sum of the performances of respec-
tive blade rows assuming that there is no interference
effect between the two blade rows. In a real tandem Case 1 18 - 19.83 - 0.20
Case 2 18 18.63 19.45 0.20 0.14
cascade, the spacing between the two blade rows is
Case 3 18 18.96 19.83 0.30 0.14
narrow and the performance must be intermediate between
the two extreme cases. The performances of these two ex-
treme cascades will be demonstrated using a potential
flow analysis. as is flow angle between two blade rows.
The zero-lift attack angle for a thin flat-plate
cascade is obviously zero, which is equal to that for an
isolated airfoil. On the other hand, according to Fig.5,
the cascade correction factor, K. on the lift/incidence necessary to incorporate the interference of the two
relation of a single airfoil dCL/dd. is 1.630 for the blade rows.
cascade with a solidity of 6 =0.98. while (1.O83 for the In order that the tandem cascade with two small-
cascade with 6 =0.35 and K=1.303 for the cascade with solidity blade rows in case 2 achieves as much turning
C=0.63. Using these values, the lift coefficient and the as the cascade of single row in case 1 does, it is
flow angle at the inlet and exit of respective blade necessary to multiply the respective cascade correction
rows are calculated for the cases that p1=10 and 18 deg, factor, K, by 1.65 as presented in case 3.
and the results are listed in case 1 and case 2 in Table In reality, as there is a little spacing between the
1. The example calculation indicates that the overall two blade rows, the turning angle become smaller as the
turning angle, p e - pt. is increased remarkably by locat- spacing is larger and it may be reasonable to assume 1.4
ing two blade rows close to each other, and modification as the interference factor for the present tandem
of the lift coefficient of blades in each blade row is cascade.

- 7 -

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use