You are on page 1of 6

ISSN 00406015, Thermal Engineering, 2015, Vol. 62, No. 5, pp. 329334. Pleiades Publishing, Inc., 2015.

Original Russian Text E.V. Mayorskiy, B.I. Mamaev, 2015, published in Teploenergetika.

STEAMTURBINE, GASTURBINE, AND COMBINEDCYCLE


PLANTS AND THEIR AUXILIARY EQUIPMENT

Aerodynamic Development and Investigation of Turbine Transonic


Rotor Blade Cascades
E. V. Mayorskiya and B. I. Mamaevb
a

Moscow Power Engineering Institute National Research University, Krasnokazarmennaya ul. 14, Moscow, 111250 Russia
b
Siemens, Bolshaya Tatarskaya ul. 9, Moscow, 115184 Russia
email: boris.mamaev@siemens.com
AbstractAn intricate nature of the pattern in which working fluid flows over transonic blade cascades gen
erates the need for experimentally studying their characteristics in designing them. Three cascades having
identical main geometrical parameters and differing from one another only in the suction side curvature in
the outlet area between the throat and trailing edge were tested in optimizing the rotor blade cascade for the
reduced flow outlet velocity 2 1. In initial cascade 1, its curvature decreased monotonically toward the
trailing edge. In cascade 2, the suction side curvature near the trailing edge was decreased, but the section near
the throat had a larger curvature. In cascade 3, a profile with inverse concavity near the trailing edge was used.
The cascades were blown at 2 = 0.71.2 and at different incidence angles. The distribution of pressure over
the profiles, profile losses, and the outlet angle were measured. Cascade 1 showed efficient performance in
the design mode and under the conditions of noticeable deviations from it with respect to the values of 2 and
incidence angle. In cascade 2, flow separation zones were observed at the trailing edge, as well as an increased
level of losses. Cascade 3 was found to be the best one: it had reduced positive pressure gradients as compared
with cascade 1, and the relative reduction of losses in the design mode was equal to 24%. The profiles with
inverse concavity on the suction side near the trailing edge were recommended for being used in heavily
loaded turbine stages.
Keywords: blade cascade, profile, suction side, curvature, concavity, transonic velocity, streamlining pattern,
efficiency
DOI: 10.1134/S0040601515050080

Proven methods are presently available for aero


dynamic designing of turbine blade cascades with
subsonic and supersonic velocities ([13] and some
other works). However, no methods are known using
which cascades with transonic flow velocity at 2
0.91.1 could be designed in an efficient manner.
With such outlet velocities, a crisislike growth of
losses is observed, which is due to the occurrence of
compression shocks and flow separation zones.
These phenomena are caused by an extremely intri
cate streamlining pattern, when there is a mixed flow
mode on the profile suction side, i.e., when there are
alternating sub and supersonic zones with a complex
wave structure at their boundaries. Under these con
ditions, viscosity has a great effect on the current in
the external flow, and it should be noted that the
structure of shocks and rarefaction waves depends on
the interaction between the boundary layers and
external flow [1, 2].
For reducing excessive flow expansion and com
pression shock intensity, it is usually recommended to
decrease the suction side outline curvature in the out
let area between the throat and trailing edge (to
decrease its skew angle) and even to make the suction
side outline rectilinear in the entire in the outlet area

between the throat and trailing edge or in its small sec


tion at the trailing edge [1, 4]. It is also known that the
abovementioned flow worsening phenomena can be
minimized by imparting an inverse (negative) curva
ture to the suction side outline in the outlet area
between the throat and trailing edge [1].
Nonetheless, in designing a cascade the efficiency
of which does not degrade under transonic streamlin
ing conditions, problems are encountered that have
not been resolved in principle. The intricate structure
of such flow is poorly amenable to a numerical analy
sis. Therefore, the use of an experimental approach
can be helpful in solving the problem of perfecting
transonic turbine blade cascades.
It is exactly such an approach that was used for
aerodynamically elaborating an aircraft turbines first
stage rotor blade cascade (the full pressure ratio in the
stage is approximately 2.6). The flow parameters for
the blade middle section along the radius are as follows:
the inlet angle 1 = 41, the outlet angle 2 = 28.9, and
2 1. Initial cascade 1, which was designed according
to the method suggested in [5] (Fig. 1a) had the follow
ing geometrical parameters: the chord b = 36.4 mm, the
thickness c = 6.4 mm, the leading and trailing edge

329

330

MAYORSKIY, MAMAEV
, %

()
x

r1

2, deg

1
1

c
1

1
2

34

32

30

a2

s
10, 1/mm 1

r2

0 1

(b)
x

3
2

30

38

42

28
50 1, deg

46

Fig. 2. Influence of flow inlet angle on the characteristics


of cascades. (1) Cascade 1, (2) cascade 2, (3) (1 ), and
(4) 2(1).

1b

A
2
10, 1/mm 1

34

0 1

2b

Fig. 1. Investigated cascades and curvature distributions


(x) on their profiles. (a) Cascade 1, (b) cascade 2,
(1) profile suction side, and (2) pressure side. A is the
throat section point.

radiuses r1 = 2.4 mm and r2 = 0.35 mm, the relative


pitch t = t b = 0.664, the blade inlet and outlet angles
1b = 47 and 2b = 28, the effective angle 2eff =
arcsin(a2/t) = 28.9, the skew angle = 17, and the
wedge angle 2 = 14. The profile suction side in the
outlet area between the throat and trailing edge is made
with monotonically decreasing the curvature toward the
trailing edge, just as is recommended in [1, 6].
Cascade 1 made on a scale of 1.4 : 1 was investi
gated in the big wind tunnel available in the Moscow
Power Engineering Institute (MPEI) Department of
Steam and Gas Turbines. The blade height is equal to
70 mm. With 2 increased from 0.7 to 1.2, the Re num
ber varied in the range (0.51.2) 106. In the experi
ment carried out at 2 = 1, the average relative error of
the loss coefficient = 3% (or at = 0.03, the abso
lute error 0.001). The angle 2 measurement
error is 0.5.
The obtained dependences for the profile loss coef
ficient (1 ) and 2 (1 ) at 2 = 1 are shown in Fig. 2,
and the dependences (2) and 2(2) at 1 = 41 are
shown in Fig. 3. It can be seen that at 2 = 1 the cas
cade (see Fig. 1) streamlining pattern remains stable in
a wide range of angle 1: a significant growth of losses
begins at 1 < 32 (at the incidence angle larger than
15), and it should be noted that a simultaneous

growth of angle 2 takes place. Less efficient perfor


mance is due to an unfavorable suction side streamlin
ing pattern (Fig. 4): it can be seen that the flow distri
bution curves ( s ) , where s is the relative curvilinear
coordinate along the profile outlines, are character
ized by essential stratification only at the inlet, and at
1 = 29 the value of is significantly greater than
unity ( = 1.27 at s = 0.47). However, the unfavorable
streamlining region is not very large, and at s < 0.35
the streamlining pattern is approximately the same for
different values of angle 1.
At 1 = 41, the decrease of with increasing 2 in
the range from 0.7 to 0.9 (see Fig. 2) can be attributed
to a growth of positive gradients of on both sides of
the profile [1]. This follows from an analysis of the
dependence ( s ) shown in Fig. 5a. It should be
pointed out that at all studied values of 2 there are two
divergent sections on the suction side in the outlet area
between the throat and trailing edge. At 2 < 1 the first
section locates near the cascade throat ( s = 0.22), and
the second one is in the trailing edge zone ( s < 0.08).
With increasing 2 at 2 > 1 the first divergent section
shifts along the flow toward the trailing edge, and both
these sections are ended with compression shocks.
This is clearly seen from the velocity distributions in
Fig. 5a. Here, the streamlining pattern is essentially
more intricate than that obtained from calculations car
ried out according to the pseudoviscosity method [7]
(see Fig. 4a).
Thus, the suction side profile streamlining structure
in the outlet area between the throat and trailing edge at
2 > 0.9 consisting of two systems of shocks and of at
least two rarefaction wave systems, and transforming as
2 increases determines the pattern of dependence
(2) [1]. The compression shocks appearing at 2 > 0.9
cause a crisislike growth of the loss coefficient from
0.02 to 0.035 at 2 = 1 (see Fig. 3). Flow separation in
the trailing edge zone ( s < 0.05) is not observed in this
THERMAL ENGINEERING

Vol. 62

No. 5

2015

AERODYNAMIC DEVELOPMENT AND INVESTIGATION


, %

2, deg

38

3
4
5

36
1

34

32
2

0
0.7

0.8

0.9

1.0

30

1.1

1.2

Fig. 3. Dependences of the coefficient and angle 2 on


the flow outlet velocity 2. (1) , (2) 2, (3) cascade 1,
(4) cascade 2, and (5) modernized cascade 3.

case. Here, at 2 = 1, the pressure ratio in the shock


p = 1.4, which is considerably smaller than its critical
value. The results of flow visualization by painting the
blade surface also confirmed that the flow streamlines
the profile without separation.
No essential changes occur in the outlet area
between the throat and trailing edge in the range
2 = 1.001.11 (see Fig. 5a). The flow becomes more
convergent in nature, and the compression shock at
the trailing edge becomes even less pronounced (up
to p = 1.3). Therefore, the value of first decreases
to some extent and then approaches a stable value. As
2 increases in this range, the angle 2 shows a
smooth growth (see Fig. 3).

A drastic growth of was noted at 2 > 1.15. A more


than twofold growth of with a simultaneous increase
of angle 2 is connected with the occurrence of devel
oped flow separation in the trailing edge zone on the
suction side ( s 0.075): at 2 > 1.2, the ratio p 2.3,
which is greater than the critical value and confirms
that there is flow separation.
Thus, the initial cascade version was found to be a
highly efficient one in the design mode and in case of
noticeable deviations from it with respect to the values
of 2 and 1 at 2 < 1.15 and 1 32.
The first attempt to improve the cascade was
aimed, first, at decreasing the angle 1b to some extent
and, second, at redistributing the suction side outline
curvature, mainly in the outlet area between the throat
and trailing edge. It was expected that the latter mea
sure would yield better streamlining of the profile in
the design mode with 2 = 1.
Cascade 2 and its parameters are shown in Fig. 1b. It
can be seen that the suction curvature value adopted
near the trailing edge is somewhat smaller than it is in
the initial version. However, an extended section near
the throat is available with constant curvature, the value
of which is larger than it is in the initial version. This
change of curvature was made with a view to achieve
more uniform distribution of velocity in the section with
s = 0.200.12 by eliminating divergent flow in it (see
Fig. 4a). The calculations of profile streamlining pat
terns carried out according to the pseudoviscosity
method confirmed that this effect indeed comes into
picture (see Fig. 4b). The parameters of cascade 2 are as
follows: b = 35.8 mm, t = t/b = 0.677, 1b = 45, and
angle = 19. The other geometrical parameters are the
same as in cascade 1.
The experimental dependences (1) and 2(1) at
2 = 1 for cascade 2 are shown in Fig. 2. It can be

()

(b)

1.2

331

1
2
3
4

1.2

1.0

1.0
5

0.8

0.8

0.6

0.6

0.4

0.4

0.2

0.2
0

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

Fig. 4. Velocity distribution over the profile of investigated cascades at 2 = 1. (a) Cascade 1 and (b) cascade 2. Values of 1, deg:
(1) 29, (2) 35, (3) 41, and (4) 51, (5) calculation at 1 = 41, and (6) throat.
THERMAL ENGINEERING

Vol. 62

No. 5

2015

332

MAYORSKIY, MAMAEV
(a)

(b)

1.4

1
2
3
4
5
6

1.2

(c)

1.4

1.4

1.2

1.2

1.0

1.0

1.0

0.8

0.8

0.8

0.6

0.6

0.6

0.4

0.4

0.4

0.2

0.2
0

0.2

0.4

0.6

0.8

0.2
0

0.2

0.4

0.6

0.8

0.2

0.4

0.6

0.8

Fig. 5. Velocity distribution over the profile at 1 = 41. (a) Cascade 1, (b) cascade 2, and (c) cascade 3. Values of 2: (1) 0.7,
(2) 0.9, (3) 1.0, (4) 1.07, (5) 1.11, (6) 1.2, (7) calculation at 2 = 1.0, and (8) throat.

pointed out in reviewing these results that, first, the


effectiveness of cascade 2 is close to the maximal
value, and, second, that at incidence angles larger than
15 (i.e., at 1 < 30) the level of losses in this cascade
is lower than it is in cascade 1. Thus, the changes made
in the profile inlet section turned to be rational.
At 1 = 41, both of the blades have approximately
the same efficiency in the zone of subsonic velocities;
however, at the design value of 2, the level of was
found to be by approximately 0.004 higher than it was
in cascade 1 (see Fig. 3). This growth of can be
attributed to the distribution ( s ) (see Fig. 4b): the
velocity variation pattern on the suction side in the
outlet area between the throat and trailing edge indeed
became smoother; however, the outlet divergent sec
tion has a large extension s = 00.1 (in cascade 1 such
section has the extension s = 00.08). In addition, a
divergent section has emerged in the interprofile chan
nel on the suction side ( s = 0.250.30), which did not
exist in cascade 1 with a small pressure gradient.
However, the main drawback of version 2 lies in the
general pattern of the dependence (2) in the zone of
transonic velocities (see Fig. 3) rather than in a some
what increased level of . At 2 1, the cascade stream
lining pattern is mixed in nature: sub and supersonic
velocities alternate near the throat and in the cascade
outlet area between the throat and trailing edge (diver
gent and convergent flow regions, respectively). As a
rule, with such flow modes, the compression shocks
locate on the profile suction side near the throat and
edge, a factor that just determines the growth of energy
losses [1]. It can be seen that losses grow intensely in
modes very close to the design one (at 2 > 1.03). An
analysis of the distributions (s ) obtained in the experi
ments (see Fig. 5b) shows that, for example, at 2 =

1.071.11, transonic velocities appear in the channel


on the profile suction side ( s = 0.250.30), and the
divergent effect becomes more pronounced. Such a
phenomenon was not observed in cascade 1. In the out
let area between the throat and trailing edge, the value
of at s 0.13 has an increased level as compared with
that in the initial cascade, and the divergent region
occupies approximately half of the suction side exten
sion in the outlet area between the throat and trailing
edge, which is unfavorable. In addition, the appearance
of the profile (s ) in the section s = 00.03 confirms
that there is flow separation at 2 > 1.07.
It should be pointed out that two fundamentally
different transonic cascade streamlining patterns are
known [1]. In the first one, there is no flow separation
in the trailing edge zone. Such a streamlining pattern
is typically observed in cascades with a moderate or
small skew angle . Although the value of increases
in cascades like cascade 1 in the range 2 = 0.951.05,
the level of losses in them is on the whole is fairly
small. The profile cascades similar to cascade 1 can be
streamlined efficiently enough in the rotor blade
bucket at 2 1.1 (see Fig. 3). The second streamlining
pattern, which is inherent in cascades with increased
values of angles , is characterized by developed sepa
ration extending along the flow, which emerges in the
trailing edge zone at 2 1. Such a streamlining pat
tern is accompanied by higher values of and by a cri
sislike growth of losses as in cascade 2 (see Fig. 3). In
particular, in using cascade 2, a noticeable drop of tur
bine stage efficiency should be expected at 2 > 1.03.
Thus, the distribution of curvature over the suction
side that was adopted in cascade 2 did not yield a pos
itive effect in the zone of transonic velocities at a
somewhat higher value of . It was also determined
THERMAL ENGINEERING

Vol. 62

No. 5

2015

AERODYNAMIC DEVELOPMENT AND INVESTIGATION


(a)

333

(b)
x

A
x

r2

2
1
t

10, 1/mm 0.5

0 0.5

Fig. 6. Profile rear part (a) and suction side outline curvature in the outlet area between the throat and trailing edge (b).
(1) Cascade 1 and (2) cascade 3.

that in the presence of compression shocks the nega


tive effect could not be predicted or explained by
means of optimization calculations carried out
according to the method [7] that does not take the
influence of viscosity into account.
The results of the performed investigations have
shown once more how much important is the way in
which the curvature is distributed over the suction side
outline [1]. For example, an attempt to increase the
suction side curvature in the throat zone in version 2
resulted in an unfavorable distribution of velocity in
this region. At the same time, some reduction of
energy losses can be achieved in this range of operating
modes only by organizing a smooth distribution of
velocity in the outlet area between the throat and trail
ing edge and decreasing the level of local velocities in
the trailing edge zone. This means that efforts should
be taken to decrease the suction side outline curvature
in the outlet section (up to introducing inverse con
cavity) with trying not to increase, at least consider
ably, the curvature in the throat zone. All the more so,
the suction side curvature in the channel should not be
increased.
In the second attempt of improving the cascade,
the main attention was paid to the design mode, in
which some degradation in the efficiency of versions 1
and 2 was noted. The objective of extending the range
of working modes of operation with respect to the
value of 2 was also set forth. It was recognized that
initial cascade 1 was the best base for modernization.
In so doing, the following constraints were fulfilled:
the profile inlet part remains unchanged, the suction
side outline remains unchanged to the throat point,
and the thickness of the new profile in its rear part
(from the throat line to the edge) remains the same as
in the initial profile.
Thus, the changes in the outline were made only in
the profile outlet part and primarily by altering the
THERMAL ENGINEERING

Vol. 62

No. 5

2015

suction side. In making changes to its outline in the


outlet area between the throat and trailing edge, we
took into account unfavorable velocity distributions in
cascade 1 in the sections s = 0.180.15 and s = 0.08
0.02 with divergent flow (see Fig. 5a). It was also taken
into account that the level of and the velocity distri
bution pattern (s ) in the trailing edge zone are almost
independent on the flow inlet angle (see Figs. 2, 4). In
view of this, the changes of suction side shape favor
able at the design inlet angle give a positive effect also
at other values of angle 1.
In view of what was said above, it was decided to
make the suction side section s = 0.220.15 locating
immediately after the throat with a somewhat increased
curvature in order to eliminate the divergent zone in this
part. However, such a change could yield a negative
effect in the trailing edge zone, in which the compres
sion shocks might become more intensive. Therefore,
inverse concavity was applied on the suction side near
the trailing edge concurrently with increasing the curva
ture near the throat with a smooth transition to inverse
concavity (Fig. 6). Expectedly, such profile shape with
inverse concavity in the suction side outlet section
occupying approximately onethird of its outline in the
outlet area between the throat and trailing edge should
yield decreased positive pressure gradients at the edge
and exclude the possibility of flow separation due to
increased velocities on the suction side.
It should be pointed out that profile cascades with
socalled inverse concavity have been known since
long ago [8]. As a rule, such profiles are used in nozzle
vane cascades with the interprofile channels having
some degree of expansion, which are designed for
supersonic outflow velocities [9]. The principles used
to develop the profiles of such cascades are based on
the conditions of interaction between the suction side
and compression shocks from the neighboring profile,
and the inverse concavity on the suction side locates

334

MAYORSKIY, MAMAEV

close to the throat. Thus, these cascades differ essen


tially from modernized profile 3 both in the shape of
profiles and in the profile development principles.
The adopted approach to profiling opens the possi
bility of radically changing the suction side curvature in
the outlet area between the throat and trailing edge
without changing the suction side outline in the channel
and without affecting the profiles entire inlet part.
However, in order to retain the thickness of the profile
outlet part, it had to be duly increased on the pressure
side (see Fig. 6). The maximal change of thickness that
in fact had to be compensated is 0.25 mm. With the
adopted changes in the profile outlet part implemented,
the wedge angle 2 decreased and the angle 2b
increased by approximately 2 as compared with their
values in cascade 1.
The streamlining calculation results showed that
the adopted changes of the profile shape led, as should
be expected, first, to a growth of flow convergence
degree on the suction side downstream of the throat
and, second, to a more pronounced divergence effect
in the suction side outlet section and to a higher veloc
ity in the pressure side outlet section (see Fig. 5c).
However, the calculation carried out according to [7]
does not take into account the viscosity effects, which
manifest themselves primarily in the divergent sec
tions. Therefore, under real conditions it can be
expected that the streamlining pattern would become
more intricate and would even be qualitatively differ
ent from the calculated one.
Figure 5c shows the experimental dependences
( s ) . At 2 1 the divergent zone on the suction side
downstream of the throat ( s 0.2215) is suppressed,
and only one such zone at the trailing edge is noted
( s < 0.1). However, even at 2 > 1, positive pressure
gradients in this zone are smaller than they were in the
initial cascade (see Fig. 5a). Only small changes of the
dependence ( s ) as compared with the initial versions
are noted on the concave side, which locate mainly in
the trailing edge zone: some growth of is noted in the
range s 0.750.95, and on the contrary, in the inter
val s 0.951.0 the velocity gradients became some
what smaller. However, an important thing is that the
flow in the interprofile channels outlet part remains
convergent in nature.
The obtained experimental dependences ( 2 )
confirm the soundness of the approach used for
designing the cascade: at the design value of 2 = 1,
the value by which decreases as compared with the
initial version is 0.008, which is equal to approx
imately 24% of the loss coefficients initial value (see
Fig. 3). An assessment shows that with such modern
ization of a number of cross sections along the blade
height, the value of can be decreased by 0.50.6% on

the average. As a result, the efficiency of a particular


turbine stage can be increased by 0.20.3%.
At the same time, it should be pointed out that
optimization matters remained beyond the scope of
the performed investigation because only one funda
mentally new version of the profile outline was consid
ered. Therefore, a positive result from the viewpoint of
the fundamental approach to profiling has been
obtained. Possibly, with some correction of the
adopted shape of modernized profile 3, a more sub
stantial gain in efficiency could be obtained.
Thus, the performed experimental investigation of
turbine cascades intended to operate at transonic flow
outlet velocities and featuring high efficiency has
shown that their further aerodynamic improvement
can be achieved using a new approach to profiling,
according to which an inverse concavity of the suction
side outline is introduced near the trailing edge. This
approach can be recommended in designing heavily
loaded stages.
REFERENCES
1. M. E. Deich, Gas Dynamics of Turbine Machinery Blade
Cascades (Energoatomizdat, Moscow, 1996) [in Rus
sian].
2. V. Kh. Abiants, A Theory of Jet Engine Gas Turbines
(Mashinostroenie, Moscow, 1979) [in Russian].
3. B. M. Aronov, M. I. Zhukovskii, and V. A. Zhuravlev,
Designing the Profiles of Aircraft Turbine Blades (Mash
inostroenie, Moscow, 1978) [in Russian].
4. E. A. Gukasova, M. I. Zhukovskii, A. M. Zavadovskii,
L. M. ZysinaMolozhen, N. A. Sknar, and V. G. Tyry
shkin, Aerodynamic perfection of the steam and gas tur
bine blade systems (Gosenergoizdat, Moscow, Lem
ingrad, 1960) [in Russian].
5. B. I. Mamaev and E. K. Ryabov, Designing the turbine
blade cascade profiles using the dominating curvature
method, Therm. Eng. 26 (2) (1979).
6. B. I. Mamaev, On selecting the blade profile suction
side curvature in the turbine transonic cascade, Izv.
Vyssh. Uchebn. Zaved., Aviats. Tekhn., No. 2, 2932
(2011).
7. A. B. Bogod, A. V. Granovskii, and A. M. Karelin,
Achieving better accuracy and shorter computation
time in numerically studying the flows in turbine
machinery blade cascades, Therm. Eng. 33 (8) (1986).
8. M. E. Deich and B. M. Troyanovskii, Investigations and
Calculations of Axial Turbine Stages (Mashinostroenie,
Moscow, 1964) [in Russian].
9. M. E. Deich, A. V. Gubarev, L. Ya. Lazarev, and
A. Jachanmohan, New nozzle vane cascades for ultra
sonic velocities developed at the Moscow Power Engi
neering Institute, Teploenergetika, No. 10, 4152
(1962).

Translated by V. Filatov

THERMAL ENGINEERING

Vol. 62

No. 5

2015

You might also like