THE AMERICAN SOCIETY OF MECHANICAL ENGINEERS

Three Perk Avenue, New York, N.Y.100164990 99-GT-418

The Society shall not be responsible for statements or opinions advanced hi papers or discussion at meetings of the Society or of its Divisions or
Sections, or printed in its publications. Dignission is printed only if the paper is published in an ASME Journal. Authorization to photocopy
for internal or personal use is granted to libraries and other users registered with the Copyright Clearance Center (CCC) provided
$3/article is paid to CCC, 222 Rosewood Dr., Danvers, MA 01923. Requests for special permission or bulk reproduction should be ad-
dressed to the ASME Technical Publishing Department.

Copyright 0 1999 by ASME All Rights Reserved printed in U.S.A.

FLOW FIELD MEASUREMENTS IN AN ANNULAR

GAS TURBINE EXHAUST DIFFUSER WITH STRUTS

11111 11 111111111111111111
Umberto Desideri and Stefano Ubertini
Dipartimento di Ingegneria Industriale, Universita di Perugia
Via G. Duranti 1N4 -06125 Perugia, Italy
Phone+Fax +39 0755852736
e-mail: desideriu@asme.org

ABSTRACT
This paper presents the velocity and turbulence characteristics of
the flow in an annular diffuser, which is a model of a gas turbine
exhaust diffuser with six struts. The diffuser where the measurements
were made is a scaled down model of a 10 MW gas turbine, built by
GE/Nuovo Pignone.
In a previous paper (Desideri and Manfrida, 1995) 2-D turbulence
and velocity measurements were presented with axial inlet velocity
conditions.
In this paper a more detailed 3-D analysis of the design and off
design behavior of the diffuser is presented. Turbulence characteristics
were determined by means of two hot split-film probes, which allowed
measuring axial, radial and tangential components of the mean velocity
and their fluctuating components. The measuring point is moved inside
the diffuser by means of two step-motors, which allow the rotation of
the hub and the radial displacement of the probe.
Off-design behavior of the annular diffuser was determined by
changing the inlet velocity angle of 10° from axial direction. The effect
of swirl on the performance of the diffuser will be presented.
Turbulence rnicroscales were also calculated in regions of interest
inside the diffuser, with particular attention to the strut wake.
NOMENCLATURE
lateral correlation coefficient
axial velocity component = Um+u
tangential velocity component = Vm+v
axial velocity component = Wm+w
Temperature coefficient
Dissipation rate
Kolmogorov length scale
Taylor's microscale
Cinematic viscosity
INTRODUCTION
Global parameters and equations, which correlate static and total
pressure at inlet and outlet, are generally used to determine diffusers'
performance. The design of diffusers is based on performance maps,
such as pressure recovery and pressure loss maps (Japikse, 1986). In
the cases, where the length of the diffuser is not a primary constraint,
diffuser angles are determined from performance maps with ample
margins to avoid separation phenomena. Design based on performance
maps is easy, but maps do not consider the presence of obstacles inside
the diffusers and the distortion of the flow at inlet.
The performance of diffusers located after a turbomachine (axial
compressors and turbines) is influenced by the highly turbulent flow at
inlet. Therefore the knowledge on the inner flow behavior is required,
in order to predict the flow field of the diffuser (Pfeil and Going,
1987). It is no longer appropriate to treat the diffuser, as an isolated
component in the design process, but the whole diffusing system
including the turbomachine has to be taken in account (Raab, et al.,
1996). In fact, the increased level of turbulence caused by the blades
of the turbine upstream of an exhaust diffuser, can sometimes improve
the pressure recovery (Zierer, 1995). All the above reasons push
towards a more detailed analysis of diffusers and mainly of their
interaction with the flow coming from the turbine or the compressor.
The exhaust diffuser of a gas turbine often features structural
elements (struts), which are necessary to support the shaft. The
presence of these objects influences the flow field and increases
pressure losses, even though they are generally designed using
aerodynamic profiles. Putting some attention to the design of the struts
can effectively modify the structure of the vortexes shedding from the
strut, and improve diffuser perfrornance (Fric, et al., 1996).
Experimental investigations made by Norris, et al., (1998) within
a diffusing S-shaped duct show that the presence of a row of struts has
two detrimental effects:

Presented at the International Gas Turbine & Aeroengine Congress & Exhibition
Indianapolis,
Downloaded From: https://proceedings.asmedigitalcollection.asme.org Indiana
on 01/10/2019 —ofJune
Terms Use: 7-June 10, 1999
http://www.asme.org/about-asme/terms-of-use
 They cause a blockage and consequently the flow accelerates,
reducing the diffusion;
They increase the size of the separation bubble.
This paper presents the 3D description of the flow field of a
model of an exhaust diffuser for gas turbines, including 24 blades at
inlet that simulate the effect of the turbulence induced by the last stage
of the upstream turbine.
DIFFUSER MODEL
The aim of this paper is the characterization of the flow in a
diffuser model built by Nuovo Pignone S.p.A. reducing in 0.35:1 scale
the exhaust diffuser of a POT 10 gas turbine.
The inlet of the model has been shaped in order to prevent flow
separation in the first part of the diffuser. The shell is made up of
transparent Plexiglas to check the position and the movements of the
probes. At the inlet section there are 24 guide vanes which generate a
wake similar to that produced by the blades of the last turbine rotor.
The guide vanes can be rotated about their axis to simulate non-Tidal
off-design flow at turbine outlet.
The interesting part in the characterization of the diffuser is the
distortion of the flow produced by the presence of six high-solidity
struts, which support one of the shaft bearings and have piping oil
supply inside for bearings lubrication of the gas turbine.
60 cm
1). To keep geometric similarity of the model with the OT diffuser, the
area ratio is maintained the same along the axis in both the model and
the OT diffuser.
The model was designed to operate in geometric and Reynolds
number similarity with the OT diffuser. The OT diffuser operates with
41.3 kg/s at 462 °C and has a Reynolds number higher than 10 6. The
Reynolds number in the model is high enough (Re > 6x10 5) to assume
similar flow conditions in the OT diffuser and in the 35% scale model.
The inner hub is a stainless steel cylinder rotatable over 360°.
Four automated carriages for probe radial movements are mounted
inside the inner hub, which has four holes to position the probes into
the flow field. The carriages motion is automated by small geared
stepping motors (Fig. 2). Another stepping motor has the function of
rotating the hub in order to realize the circumferential movement of the
probes. A 2-channels MSTEP-5 motor management board permits the
control of the stepping motors by personal computer
Step motor Probe support
Security
stop

Figure 2: Carriages for the movement of test probes inside

the inner hub.

60°

Oc-
18 ' cm 32 :m

45 cm

Ii 1.6 cm 1
75 cm
4 2 cm

25 cm

30 cm

Figure 1: Model of gas turbine exhaust diffuser.

The inner and outer diameters of the inlet section are 190 mm and
320 mm. Axial length of the diffuser is 340 mm with an outer diameter
Figure 3: Measurement sections arrangement.
at outlet of 420 mm. The resulting conicity of the shell is 1:0.28. The
area ratio of the diffuser is 1.53 and the length/wide ratio is 2.01 (Fig.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 MEASUREMENT SECTIONS The model is connected to a wind tunnel with a 22 kW
The measurement sections, shown in Fig. 3, are named as section centrifugal-flow industrial fan. The flow rate is controlled by a sluice
A, B, C and D. Section A is just behind the guide vanes; sections B gate, positioned 3.40 m before the exhaust.
and C allows the characterization of the flow in front and behind the
strut. Section D is the section at outlet.
Since the inner hub diameter is constant over the entire length, it
is also possible to move the measurement sections in the axial
direction. Figure 3 also shows the position of the measurement points
in each section. The first radial point of acquisition is 15 min far from
the inner hub and it corresponds to 0 mm radial position in all the
plotted graphs.
The radial movement of the probe is made up of 10 mm steps, to a
safety distance from the shell. 31 angular positions correspond to every
radial one, covering a sector of 60° containing one strut in the middle.
Results have been produced for a 60°sector because the flow field
repeats itself every 60 0 . The 60° sector was chosen in the upper part to
avoid any interaction with the step-motor and its support, which
disturb the flow behind in the lower part of the model. Figure 4: Reference measurement system.
60 60
50

30:=3" — 0AN7 1.°"-

g 20 2
10- 10
0 "---473,. 0
0 5 10 115 20 25 10 0 0 ao
Radial Position (rrm): Section A Radial Position (rim): Sedion B
60
50
40 .c.41Cri=
0
t:Fa-
A' 30
207r
-V 56.0=rl
d 20
1 0 1. 10
0 M(.0 , 0
o 10 70 zo ao so
Radial Position (rim): Section C Radial Position (rrm): Sedion D
Figure 5: Contour plots of axial velocity Um (m/s) with straight guide vanes.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 10
I 5
a-- Radial Position at Orrrn
>j5 Rada Position at 30mn _
R3F1I4 position at 50mn
0 10 20 30 40 50 60
Angle (deg)
Figure 6: Vm velocity component vs. angular position at sections B and C.
MEASUREMENT TECHNIQUES
The 3D characterization of the diffuser flow was made by using
two types of split-film probes:
DANTEC 55R56 split-film probe with R20 = 4.5 0 for one film
and 4.80 for the other and a20 = 0.0039 for one film and 0.0041
for the other
DANTEC 55R57 split film probe with R2o = 5.27 0 and 4.57 0,
and a20 = 0.0042 and 0.0041.
The 55R.57 probe allows the measurement of the x and y
components of the velocity vector, whereas the 55R56 measures the
components on the x-z plane (Fig. 4).
The measurements with the two probes are made with the same
flow velocity in the tunnel at two different times, so it is not possible
to correlate turbulence quantities obtained on the x-y plane to those on
the x-z one.
The probes were connected to a dual-channel anemometric system
by A.A. Labsystems, which includes amplification and filtering for
each channel. The anemometer output was then transmitted to a
sampling board and a PC elaborated the results. The sampling board
used is a Metrabyte DAS-I 600, which features 16 channel and a 12-
digit AJD converter. The maximum acquisition rate is 100 kHz.
The similarity of flow conditions among measurements made at
different times is guaranteed by keeping the wind tunnel flow rate
constant. This assumption also allows making measurements with only
one probe at a time thus avoiding flow interaction between probes
simultaneously located in the flow field.
EXPERIMENTAL RESULTS
The flow field of the diffuser model was studied in two different
conditions:
- with straight inlet blades;
- with 10° inclined inlet blades.
Figure 5 shows the contour plots of the axial component of
velocity at the four sections with straight inlet blades.
The presence of the guide vanes can be easily detected in the plot
of section A. A flow deviation and a reduction of the axial velocity,
Um, is evident at 15°, 30° and 45°, which are the angular positions of
the guide vanes.
Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
a— Radial Position at Ortm
Radial Positirn at 3Orrrn
Rada Position at 53mm
0
8 _5
0 10 20 30 40 50 60
Are (deg)
In section B the downstream strut at 30° causes a strong deviation
of the fluid with a significant reduction of the axial velocity. The
presence of the guide vanes at 15° and 45° is still traceable. The y-
component velocity graph, shown in Fig. 6, confirms the deviation
caused by the presence of the strut.
The plot of section C shows a wide area of low axial velocity,
caused by the wake of the strut. Approaching the inner hub, axial
velocity slows down and this can be mainly attributed to the growth of
the boundary layer. The presence of the guide vanes is still detectable.
It seems that the interaction between the guide vanes and the strut
wakes causes the onset of the flow separation at the shroud. The Y-
component of velocity graph (Fig. 6), shows how the flow reattaches
behind the strut.
In section D, the growth of the boundary layer is more evident,
since the axial velocity decreases from 55 m/s, at mid-span, to 20 m/s
near the hub. A clear zone of separation can be observed between the
strut and the guide vanes close to the shroud.
The Vm and Wm components of the velocity vector are shown in
Figs. 7 to 10. Vm and Wm components are very small in Section A
(Fig. 7). Span-wise deviations from the axial velocity are due to the
presence of guide vanes. Since the thickness of the guide vanes is 5
mm and they present a rounded trailing edge, the V m component is
always in the interval t4 m/s. The W m component is generally zero,
except along the guide vanes, where a slight upward direction from the
hub to the shroud can be observed.
Before the strut there is a significant deviation in the V m
direction, with a left-wise direction on the left of the strut and a right-
wise direction on the other side (Fig. 8). Near the hub the W m
component is negative (upward to the center of the section), whereas
near the shroud it is positive, i.e. directed downward. This
characteristics is the same along all the hub and the shroud, but the
highest Wm component can be found along the sides of the strut (at
30°) and along the wakes generated by the guide vanes at 15° and 45°.
In section C a secondary flow is quite evident on the sides of the
main wake generated by the strut. The V m component is positive on
both sides of the strut in the region near the hub from 0 to 35 mm, and
negative near the shroud (Fig. 9). If we combine this component with
the W, two vortexes with counterclockwise direction can be observed.
 In section D (Fig. 10), the flow is mainly redirected along the axis behind the strut and the W m component due to the development of the
and Vm and Wn, components have almost disappeared. There still boundary layer.
remains aslight Vm component due to the reattachment of the flow
60 • 60

40- ec

30 -400

.g
m 20 4s)
t
Errro

10->0
0 T
5 10 15 20 25 30 5 10 15 20 25 30
Radial Position (rim): Section A, V Radal Position (mm): Section A W
Figure 7: Contour plots of the Vm and Wm velocity components (mls) with straight guide vanes in section A.
60 60

50 50

40

30
a)
.1.3 20

5 10 15 20 25 10 20 30 40 50
Radial Position (mm): Section B, V Radial Position (mm): Section B, W
Figure 8: Contour plots of the Vm and Wm velocity components (m/s) with straight guide vanes in section B.

60 60 LCO
•CdP
—

50 50- tW
a °3

- 00
40-E–_-_-_3. 40- 0 04

C-rato
o 4, ac4 -03
13 -4 1c
7
a) Thco
20-r o
r3.03
n %to>. 7-07
00------
k■- CIO
1 0- -.0- 4 I

? 4C0 -C3r
a- 240
CI ,--- , ) •7 ,
...--,--,--.-- --4
0
0 10 20 30 40 50 50 70 10 23 33 ao 5O 70

Radial Position (rrm): Seclion C, V Radial Positiai (mm) : Secticn C, W -

Figure 9: Contour plots of the V m and Wm velocity components (m/s) with straight guide vanes in section C.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 60
-203
•=1:C•--
50 00

< N(1.3 <

40 40 .6.0)
11) <::E7 ttoorc
ID

3.1 30- --aw 21 30—

2 If
,g) "c-4- ,g)
1 0-
-0 4.60-

20 40 80 80 eo 80
Radial Position (m). Section D, V Radial Posffion (mm): Section D, W

Figure 10: Contour plots of the V m and Wm velocity components (m/s) with straight guide vanes in section D.
60

50-
gE4—aomat=soovia
40 10.0016D
amm° 5atra cabi
&Ed
g) 20
s 002.4110160

10

o
o 5 10 15 20 25 30 10 20 30 40 50
Radial Position (rim): Sedion A Radial Position (rim): Section B
60 60

50
0.016_
3-27:
40-ta

30 I --=-2 1 r-

e g 20 6Aiw —0.0160 <

10

0
0 10 20 30 40 50 60 70 80
Radial Position (rrrn): Sedion C Radial Position (n-m): Section D

Figure 11: Contour plots of the axial fluctuating velocity component, u, (straight guide vanes).

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 60 60
UV
50- 50-
•100303
400101
40 c Th.o2 4o -

I I 0/11
42CI
" .1 11—n>
30-n --crasfr-015- ?ui ao -4:ono-
-0.01)3

g 20 =21-C
10-
03
0
0 10 20 30 40 50 60 70 0 10 30 40 50 do 70
Radial Position (rrm) Radial Position (ran)
Figure 12: Contour plots of the product of tangential components of velocity fluctuations (straight guide vanes)
60 60

50

40

g 30 146.ar2=-0048--0 .02,0

10- 10

o 0
o 1'0 40 50 60
Radial Position (mini)
Figure 13: Contour plots of the v and w components of velocity fluctuations (straight guide vanes)
Figure II shows the contour plot of the axial component of
velocity fluctuations. Measured turbulence quantities are the axial and
tangential components of velocity fluctuations and their product in
both the xy and the xz-planes. The rms of turbulent components
(proportional to Reynolds stresses) is normalized by mean velocity.
Turbulence at inlet increases behind the vanes but it is lower than
1% of average velocity over the largest part of section A. In section B
axial velocity fluctuations reach values close to 2-3% only near the
inner hub. The tangential components in section A and B are also
lower than 2%; this means that turbulence is quite isotropic. The
region interested by turbulence higher than 1% is quite narrow behind
the guide vanes and the shape of the wakes can be easily traced.
The axial fluctuations are considerably increased in section C,
where it reaches 20% behind the strut. From section B to the outlet, it
is also evident the growth of the boundary layer, which is more
remarkable near the hub, where the width of the paccage is smaller.
The tangential components of the Reynolds stresses at section C are
shown in Fig. 12. Behind the strut, tangential fluctuating components
reach 10%. Fluctuating components along y and z, shown in Fig. 13,
confirm the high level of turbulence produced by the wake of the strut.
70 0 10 20 30 40 50 eo 70
Radial Position (rrm)
In section D the axial velocity fluctuations are remarkable only
near the hub, where they exceed 20%. Instead, the turbulent
fluctuations generated by the strut are considerably damped. However,
the interaction of the wakes of the strut and the guide vanes near the
shroud seems to be the main reason for the development of the
boundary layer near the shroud. In the region between the guide vanes,
for angles between 0° and 15 and between 45° and 60°, the turbulence
is much lower and the boundary layer is not developed as in the region
between 15° and 45°.
A second set of measurements was made with the guide vanes
rotated of 10°. The differences between this case and the case with
straight guide vanes are clear from Fig. 14, where axial velocity
contour plots are shown.
In section A, the size of the wakes of the blades at 15 °, 30° and
45° are almost double, as each one covers a sector of about 8°. The
flow field at section B is characterized by larger slowdown zone, since
the axial velocity reaches 70 m/s only in the first and in the last 15°
sectors. There is no distinction between the wakes of the guide vanes at
15° and 45° and the stagnation zone produced by the strut.

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 60 74._0=g3 r_

50 ;73 tM--
Zt The? C-S- 74
-c62P

eiS4D
-c!tt cap

20
.,SEL75.011

10
611-711-(=
0 o
0 5 10 15 20 25 30 o 140 20 30 4b 50
Radial Position (inn): Section A Radial Position (rim): Sedion 8
60

50
50 /60 ]

40-air €5.0 40 7c=r3"..

56 560

---55
-13 30 ----45a g 30?

ga) t 20 --360 48P-7

10a) 605 10- 4

600 6513-
j
10 20 30 40 50 80 70
Radial Position (rim): Sedion C
Figure 14: Contour plots of the axial velocity, U rn , (m/s) with inclined guide vanes
In section C the wake of the strut is more evident. The influence
of the guide vanes has disappeared and this confirms the strong
directional power of the strut.
Moreover, no separation is detected before section D, where a
large zone close to the upper hub is characterized by a very low and
irregular axial velocity.
The distribution of the axial fluctuating component, shown in Fig.
15, confirms a low level of turbulence in section A and B. In section C
the axial Reynolds stress increases behind the strut. The y and z-
components of velocity fluctuations follow the case with straight inlet
vanes, since they increase behind the strut (8+15 %).
The tangential components of velocity fluctuation, shown in Fig.
16, point out the anisotropy presence behind the strut, since they reach
values close to 10%.
The average level of turbulence is the same of that in the case
with straight guide vanes, but the peak values are lower. Moreover, the
disappearance of the wakes of the guide vanes damps the separation at
the shroud due to the interaction between vanes and struts, especially
on the pressure side of the strut.
40 60 80
Radial Position (rim): Sedion D
On the suction side, with the angle ranging from 00 to 15 °, a
higher turbulence level can be observed near the shroud (Fig. 15,
Section D). Therefore, the presence of the strut is not negative for the
performance of the strut, when inlet velocity is not axial. The leading
edge has such a large curvature and the thickness/chord ratio of the
strut is so high that flow with some degrees of incidence is easily
straightened.
TURBULENCE MICROSCALE
Turbulence has always all three components of velocity even if
the mean flow has only one direction. Therefore, within a turbulent
flow, eddies of different sizes govern the energy transfer between the
mean flow and the turbulence quantities. The boundary conditions of
the flow determine the maximum scale of eddies and vortex stretching
causes an energy transfer from large to small eddies. Viscous forces,
which dissipate the kinetic energy of the smallest eddies, stop this
«cascade» of energy. Therefore, viscous dissipation of turbulent
energy determines the scale of the smallest eddies (Bradshaw 1971).

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 60 60

50. 50
00:060 Ca m eo
.4100160
0

40- . 4.004wici
00160— toes co.
001E0 "§) 30 J0032
1:11 coe:
20-Aur ----u13
nr. tu 20
011160— SF-d
10 M002410:00321 10
00160.
` 5E-4
0 0
e lb 15 20 25 30 o 40 50
Radial Position (ran). Section A Radial Position (rm): Sedion B
60 .rnAk
0.01
a.0123

50
0.01 CO
40- 40
r gq)' 011220

'1 30- - c. --(maw

tv 0.01
-1"2°
1 20 ' "Mar .
t
10h,
4.0160
4
10 -9 ° °243trel
11,71
7"?
0.016C 1l 2 m
. s:0 1/4

o 0160

0 10 20 30 40 50 60 70 Lio 80
Radial Position (rim): Section C Radial Position (mm): Section D

Figure 15: Contour plots of axial fluctuating velocity component (inclined guide vanes)
60 60
4100200
tiv 400:t03
0.1:05t
50- 50-
0.009r"

40 40 009 1002-40020
0,,,
30 j ma atsomEmmMilnA
CI0
g 30 J------att
C) low
-cao
- -a .0090C
20- 1 20-
o.00sr--
10- 10 ----;002,
-e-aomaP-

lb Lio 0 o 70
rwr
0 10 zu :30 tio 0 o 70
Radial Position (rin) Radial Position (rrrn)

Figure 16: Contour plots of the tangential components of velocity fluctuations (inclined guide vanes).

9
Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use
 60 60
0120
°rum-
50

c rc 40-

g 30-

,g) 20-

al
Q120
0 ,
0 10 20 30 40 80 do 70 2o 4o 60
Radial Position (am). Section C Radial Positicn (rim): Section D
Figure 17: Contour plots of Kolmogorov length scale (mm) with straight guide vanes (finite difference calculation)
60 60
I10:1turn>
al
C11
50- QI2D <
ca120 (lc
a mt=-_
40-
111 abiX>

g 30- 060-
o -ur C Q060

(Q076 <
60 10-r---ct
2e111

0 U°813 - 1°,-CLV
0
,
10 20 30 4b - 800
cum

6 70
0 4°"-
0
< 20 40 80 80
Radial Position (rim): Section C Radial Positicn (nrm): Section D
Figure 18: Contour plots of Kolmogorov length scale (mm) with straight guide vanes (by transverse correlation coefficient).
60
6(8( —mac
a
an 60
.10) „
50-? 11

40 40-1 )4 a Ca°9-L5r0976
%

■ •••■••• 30: taiz Os°

2 .---,
---T
\------a07 ca 07(C
aft.
g 20- afaCcrntrioo
7 a ao9---_ct___, 0
e>
.9cc
10 Caasa
1) ai1I Q081:C__Q070
I (Si
o
o l'o o o tic do do 70
Radial Position (am). Sedion C
Figure 19:
19: Contour plots of Kolmogorov length scale (mm) with inclined guide vanes.

10

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 The Kolmogorov length scale, it is usually considered as a The use of Taylor microscale to detect turbulent flow has
measure of the "dissipating" eddies: produced interesting results, by showing with a good resolution the
most important turbulent zones inside the diffuser.
3 1/4
V
(1) REFERENCES
6 Bradshaw, P., 1971, "An Introduction to turbulence and its
measurements", Pergamon Press Ltd., Oxford, UK, pp. 1-17.
The Taylor's microscale X can be used to evaluate q: Bruun, H.H., 1995, "Hot Wire Anemometty principles and signal
analysis", Oxford University Press Inc., New York, pp. 62-70.
Desideri, U., Manfrida, G., 1995,"Flow and Turbulence survey
(2) for a model of gas turbine exhaust diffuser", ASME paper 95-GT-I39.
_(au/at)2 Kline, S.J., Abbott, D.E., Fox, R.W., 1959, "Optimum Design of
Straight-Walled Diffusers", Transactions of the ASME, Journal of
Taylor's microscale is not a direct measure of either large eddies Basic Engineering, pp. 321-330.
or of small eddies, but it can be related to both (Bruun 1995). Fric, T.F., Villarel, R., Auer, R.O., James, M.L., Ozgur, D.,
To determine Kolmogorov length scale, the dissipation rate has to Staley, T.K., 1996, "Vortex shedding from struts in an annular exhaust
be calculated by using the following expression: diffuser", ASME paper 96-GT-475.
2
Norris, G., Dominy, R.G., Smith, A.D., 1998, "Strut influences
U within a diffusing annular s-shaped duct", ASME paper 98-C T-425.
c=15v • i—
v (3) Pfeil, H., Going, M., 1987, "Measurements of the Turbulent
Boundary Layer in the Diffuser Behind an Axial Compressor",
Figures 18 and 19 show the contour plots of at section C and Transactions of the ASME. Journal of Turbotnachinery, -Vol. 109, pp.
section D with straight guide vanes. Even if the flow is not isotropic 405-412.
everywhere, they give a good reference of the distribution of the vortex Raab, I., Lippert, D., Hennecke, D.K., 1996, "A new concept for
microscale. Taylor's microscale was evaluated in two ways: the design of turbine diffusers", AWE paper 96-C T-450.
The contour plots of Fig. 17 were determined by a finite Runstadler, P.W. Jr., Dolan, F.X., Dean, R.C. Jr., 1975, "Diffuser
difference calculation based on equation (2); data book", Creare, TN-186.
In the contour plots of Fig. 18, X was determined by the Zierer, T., 1995, "Experimental Investigation of the Flow in
transverse correlation coefficient as illustrated by Bruun (1995). Diffusers Behind an Axial Flow Compressor", Transactions of the
The results obtained are practically coincident. ASME, Journal of Turbomachinery, Vol. 117, pp. 231-239.
In section C the microscale near the strut tends to reduce, mainly
close to the inner hub. As expected where the turbulence is stronger the
dissipating eddies are smaller. Close to the shroud there are two areas,
characterized by small values of length scale, between the strut and the
guide vanes at 15° and 45°. This confirms the hypothesis of separation,
due to the interaction of the strut and the guide vanes.
In section D at 30° angular position the microscale is smaller,
probably because the wake of the strut is still present even if it is not
detectable by measuring the axial velocity only. The separation of the
flow at the shroud is pointed out by the small values of eddies
dimensions.
Figure 19 shows the Kolmogorov length scale contour plots, with
guide vanes rotated of 10°. As seen before, the wake of the strut in
section C is narrower. In section D smaller eddies are located close to
30°. The average value of Ms smaller than in section C, probably
because the turbulence is spread all over the area.

CONCLUSIONS
A 3D flow field analysis is presented in a model of a gas turbine
exhaust diffuser. This has allowed the calculation of the characteristics
of the wakes before and after the struts, which support the shaft's
bearing in the real gas turbine.
The struts proved to have a strong directional power, even when
inlet flow is swirled of 10° from axial direction. However, the struts
produce a highly turbulent flow that favors separation between the
strut and the vanes and near the hub.

11

Downloaded From: https://proceedings.asmedigitalcollection.asme.org on 01/10/2019 Terms of Use: http://www.asme.org/about-asme/terms-of-use