Proceedings of ASME Turbo Expo 2014: Turbine Technical Conference and Exposition

GT2014
June 16 – 20, 2014, Düsseldorf, Germany

GT2014-25524

UNSTEADY EFFECTS OF A GENERIC NON-AXISYMMETRIC ROTOR ENDWALL

CONTOUR ON A 1½ STAGE TURBINE TEST RIG AT OFF DESIGN CONDITIONS

Dwain Dunn∗ Glen Snedden Theodor von Backström

University Of Stellenbosch, CSIR University Of Stellenbosch
Stellenbosch, South Africa Pretoria, South Africa Stellenbosch, South Africa
on staff at the CSIR
Pretoria, South Africa
Email: ddunn@csir.co.za

ABSTRACT that of the annular rotor. The rotor outlet flow was thus more
With the current drive to improve fuel efficiency and reduce uniform due to the more dispersed vortex system. As seen with
emissions, in gas turbine engines various methods have been in- the experimental results, the extent of the change due to the con-
vestigated. Previously it has been shown that a generic rotor toured rotor changes with loading. The differences present in the
endwall contour could improve the efficiency of a 1 ½ stage test decreased loading case being relatively insignificant.
turbine at design conditions. The current investigation looked It was concluded that the generic contour does not intro-
at the increased and decreased loading conditions to determine duce any unsteady effects at off design conditions that were not
if the contour introduces detrimental effects at off design condi- observed in the design case.
tions.
A previous unsteady analysis of the design condition found
that the contoured rotor does have an effect on the flow field, NOMENCLATURE
reducing the magnitude of the hub endwall secondary flow region α Absolute yaw angle (◦ )
as well as reducing fluctuations in the velocity. S Span (mm)
Experimental results showed that the increased load case C Absolute velocity (m/s)
presented with an increase in hub endwall secondary flow struc- W Relative velocity (m/s)
ture when compared to the design case. This increase was to be φ Flow Coefficient (-)
expected due to the increased turning of the flow due to the in-
creased loading operating condition. The contoured rotor had Subscripts:
a weaker hub endwall secondary flow system, with the high mo- t Tangential
mentum flow distributed more in the span wise direction. The x Axial
variation in the velocity was also found to be smaller for the 3 Rotor outlet
contoured rotor. stage From stator inlet to rotor outlet
The decreased loading case showed similar improvements,
but the extent of the change was less due to the lower turning of
the flow (due to a faster rotor).
INTRODUCTION
The numerical results show that the hub endwall secondary
Several researchers have been shown that secondary flows
flow vortex of the contoured rotor was not as tightly wrapped as
and the resulting losses [1, 2, 3, 4] exist in turbomachinery. Sec-
ondary losses are those losses that occur due to the endwall
∗ Address all correspondence to this author.

1 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 boundary layer flow being turned by an adjacent curved sur- ing did improve efficiency, even though the contouring had been
face [5] such as a turbine blade. Secondary flows were first iden- designed for a cascade.
tified in 1955 by Hawthorne [6] and have been extensively re- In a previous paper Dunn et al. [10] experimentally found
searched since then. However as recently as 2012 Denton [7] that at design conditions the hub endwall contour in question did
stated that the physical origins of secondary loss were not fully improve the flow field of the rotor exit flow. It was also found
understood. Denton [7] also indicated that more detailed mea- that the oscillations in the flow were reduced. The current inves-
surements in a rotating turbine (instead of in a cascade) with at tigation was aimed at looking at the influence of the contoured
least a full stage were required. rotor endwall at increased and decreased loading conditions (i.e.
positive and negative rotor inlet incidence angle) from both an
experimental and numerical view point.
TABLE 1: TEST RIG DESIGN SUMMARY

Inlet EXPERIMENTAL SETUP

Axial Velocity 21.38 m/s
1st Stator
No. of Blades 30
Rotor
No. of Blades 20
Design Rotor Speed 2300 RPM
Flow Coefficient Hub 0.625
Casing 0.439
Stage Reaction Hub 0.38
Casing 0.70
Midspan Stage loading 0.70
Midspan Outlet Relative Velocity 57.4 m/s
Modified Stage Power∗ 3.42 kW
∗ Rotor outlet angle adjusted according to [5]

Stage Pressure Ratio 1.0393

Exit Reynolds Number 127500
Zweiffel Coefficient [8] @ hub 0.94
2nd Stator
No. of Blades 30 FIGURE 1: SCHEMATIC OF THE TEST SECTION OF THE
1½ STAGE TURBINE TEST RIG

As manufacturing and analysis tools advance it has become

easier to implement hub endwall contouring as a means to reduce The experimental turbine test rig used was the same as that
endwall flows. Non-axisymmetric hub endwall contouring has used by Dunn et al. [10] for the design speed measurements. The
been investigated and when designed correctly have produced test rig can be seen in Fig. 1 and can be described as a vertical
a positive change in the turbine performance. Ingram [3] de- axis, low pressure test turbine. The test rig was originally de-
signed a set of non-axisymmetric hub endwall contours which signed for tip clearance measurements resulting in all the blades
were tested in the Durham linear cascade. Snedden et al. [4, 9] being attached at the hub, with a tip gap at the casing. The ro-
then took an adaptation of the most promising of Ingram’s [3] tor had a 1mm (1.6% span) tip gap, while the stators had 0.5mm
designs (the P2 endwall) and performed steady state measure- (0.8% span) tip gap. A summary of the design specifications can
ments in an annular rotating test rig and found that the contour- be seen in Table 1. It should be noted that the design speed was

2 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 at a sampling frequency of 100kHz (±130× blade passing fre-
quency). The TSI Model 1240 Standard Cross Flow “X” Probe
has an uncertainty of ±0.77% mean velocity resulting in a max-
imum uncertainty in the current investigation of 0.3 m/s (the
uncertainty of the design flow condition being 0.19 m/s). The
hot-film probes were calibrated before each test using TSI model
1129 auto-calibrator. The calibration curves were fitted to a
Kings Law curve due to its consistently lower mean squared er-
ror.

FIGURE 2: DIMENSIONS OF THE HOTFILM PROBE USED

FOR EXPERIMENTAL TESTING

not tested in this investigation, the rotor speeds tested can be seen
in Table 3. The instrumentation can be seen in Table 2
The Durham cascade blade profile used by Ingram [3] was
used at the hub to allow the use of a similar generic hub end-
wall contouring. The tip and the stators were designed using the
inverse design method [11] such that the rig had the operating
conditions as listed in Table 1. Fig. 3 shows a 3D representation
of the final hub endwall contour as used in the current investiga-
FIGURE 4: CONVENTIONS USED FOR ANGLES AND VE-
tion.
LOCITIES

The current investigation was performed at rotor speeds that

correspond to a 5◦ change in the rotor flow incidence angles. The
speeds tested at can be seen in Table 3.
The current experiment only performed measurements
downstream of the rotor (X3 in Fig. 1). The sample grid con-
sisted of 342 points, 18 in the tangential direction (at 1◦ incre-
ments from 31◦ to 48◦ ) and 19 in the radial direction, starting at
3.5mm from the hub, thereafter every 3.1mm. At each point 64k
samples (the IFA300 converts this to 65536 samples) were taken.
The resulting data was Phase Lock Averaged (PLA) [13, 14], to
facilitate noise reduction. A step wave was sampled in conjunc-
tion to the 2 components of the “X” probe to give an indication
of the rotor phase angle. The first rise of the step wave was used
to phase lock the velocity data. The step was generated after each
full rotor revolution dependant on the start position of the rotor.
FIGURE 3: ADAPTED P2 ENDWALL CONTOUR APPLIED Dunn et al. [10] found the variation in the flow angle during
TO THE ANNULAR TURBINE HUB testing exceeded the cone of acceptance of the probe. For this
reason the probe was set to three different yaw angles, as shown
in Table 4 in order to accurately capture the flow and it’s fluc-
tuations. The steady state measurements of Snedden [15] were
used to determine the appropriate probe yaw angle (shown in Ta-
EXPERIMENTAL TECHNIQUE ble 4). Results were still obtained when the flow was outside of
Table 2 shows the rig instrumentation and the associated un- the cone of acceptance, the values measured were however erro-
certainties. Measurements were performed using the TSI IFA300 neously reduced in magnitude. Thus to accurately measure the

3 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TABLE 2: EXPERIMENTAL INSTRUMENTATION AS USED BY [12, 10]

Primary Instrumentation

Parameter Instrument Uncertainty

Torque Himmelstein MCRT 28002T(5-2)CNA-G ±0.03Nm
Speed + Model 721 Mechanical Power Instrument 2RPM
Barometric Pressure Siemens Sitrans 2 Pa
P 7MF4233-1FA10-1AB6-Z A02+B11
Temperature PT1000 RTD’s ±0.05◦ C

Secondary Instrumentation

Unsteady Flow Mapping TSI Model 1240 Standard Cross Flow “X” Probe ±0.77% mean velocity
Tangential Traverse Custom cable system Better than 0.01◦
Radial and Custom two component backlash free 0.01mm
Yaw traverse traverse driven by Cool Muscle® drives 0.1◦

unsteady velocity field, the absolute rotor exit flow angle vari-
ation is required. The probe was rotated using a custom yaw
traverse as shown in Table 2. To reduce the effects of hysteresis
TABLE 3: SPANWISE DISTRIBUTION OF YAW ANGLE
the probe was always rotated the same direction, in cases where
USED DURING UNSTEADY TESTING
the opposite direction of rotation was required the angle was ex-
ceeded then rotated back.
Test Case Rotor Speed Incidence
(RPM) Angle (◦ ) The FFT’s presented were taken of the raw data (all 65536
samples), and not the phase lock averaged data to prevent
Increased loading 1907 +5◦ elimination of potentially interesting frequencies (ambient noise
would be consistent). The raw data was parsed such that the start
Decreased loading 2820 −5◦
and end of the data set were at the same phase with respect to the
rotor location (i.e. rectangular window).
TABLE 4: SPANWISE DISTRIBUTION OF YAW ANGLE
USED DURING UNSTEADY TESTING

Span Increased Loading Case

(%) Yaw angle (◦ ) COMPUTATIONAL FLUID DYNAMICS SETUP

5.74 - 26.1 50 The numerical work presented is a continuation of previ-

ous work (more detail in [12, 16, 17]) and for brevity it is only
31.2 - 76.9 30
summarised below. The simulations were carried out using Nu-
82 - 97.5 0 meca’s FINET M /Turbo v8 [18]. Domain scaling with a dual time
stepping technique was used for the unsteady simulations requir-
Span Reduced loading case ing matching periodicities of consecutive blade rows. Thus 3
(%) Yaw angle(◦ ) inlet stator blades, 2 rotor blades and 3 outlet stator blades were
meshed (shown in Fig. 5). The SST k − ω turbulence model, as
5.74 - 21.0 10 used by Snedden et al. [15] and Dunn et al. [16], was used for the
26.1 - 76.9 0 simulations. The mesh had approximately 5 million cells with an
average y+ < 1. Snedden [9] performed a grid dependence study
82.0 - 97.5 -20
and found the solution to be independent on mesh.

4 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 INCREASED LOADING CASE: Looking at the veloc-
ity magnitudes in Fig. 6 it can be seen that the contoured ro-
tor has a lower peak magnitude in the hub endwall secondary
flow region. The peak seen in the annular case at 31% span
(Vmag = 37.3 m/s) was reduced by 2.9 m/s and the peak region
was moved down to 26% span for the contoured rotor. The mag-
nitude of the fluctuations present in the hub endwall secondary
flow region have also been reduced, except at 5.7% span where it
was increased. As seen with the design speeds in Dunn et al. [10]
the magnitude of the tip leakage flow was strengthened, with the
oscillations in the flow increasing as well.
Looking at the FFT magnitudes at the blade passing fre-
quency in Fig. 8 it can be seen that there was an increase in the
magnitude below midspan. The overlaid contour plots in Fig. 6
show that the contoured hub endwall secondary flow vortex core
was not as azimuthally distributed as the annular case, making
the wake more pronounced, causing a larger FFT magnitude at
FIGURE 5: MESH USED FOR CFD SIMULATIONS the blade passing frequency.

RESULTS DECREASED LOADING CASE: Figure 7 shows the

As in Dunn et al. [10] the experimental results obtained con- decreased loading case which presents with a similar trend, with
sist of axial and tangential velocities only, due to space con- the peak velocity magnitude of the hub endwall secondary flow
straints making it difficult to obtain all 3 components simulta- region being reduced from 29.6 m/s to 28.3 m/s. The peak value
neously. For this reason the standard measures of rotor perfor- was again moved closer to the hub endwall, but the difference in
mance cannot be used as they either require all 3 components of the peak values can be considered negligible. Flow oscillations
the flow or temperatures and pressures. were reduced, most notably at 26% span. The tip leakage flow
was again affected, but the peak velocity magnitude remained
EXPERIMENTAL RESULTS approximately the same.
For brevity only the velocity magnitude experimental results Even though the FFT magnitudes were affected, as shown in
will be presented. Since the velocity magnitude was the result of Fig. 8, the maximum change was only 0.29 m/s which was only
the measured quantities (done at each discrete time step) it was 1.7× the uncertainty of the probe. The magnitude of the har-
representative of the axial and tangential velocities and fluctu- monics of the blade passing frequency exhibit some differences
ations thereof. It should be noted that the velocity magnitude as well, but again the magnitude was such that it oscillated about
presented does not contain the radial velocity component. measurement uncertainty.
The error bands on the unsteady velocity profiles indicate the
range of velocities measured (minimum and maximum values). NUMERICAL RESULTS
The FFT data was plotted on the same set of axes as the veloc- The numerical results are presented in Fig. 9 and Fig. 12 as
ity components. The FFT’s use the secondary X-axis at the top time averaged meridional plots as well as time averaged relative
of the graphs, and the vertical scale can be found on the bottom reference frame contour plots in Fig. 10, 11, 13 and 14 (at the
right. The scale of which is multiples of blade passing frequency axial location that the rotor exit experimental measurements were
(increased loading case: 635.67 Hz, and decreased loading case taken, X3 in Fig. 1). The oscillations that were observed were
case 940 Hz). The design value was listed for reference, but it about the time averaged values presented.
should be noted that a disagreement is to be expected as the tur- In general the rotor exit experimental results compare well
bine was operated in off design modes. In all cases “improve- with the numerical results. In a previous numerical investigation
ment”, or “improved” should be taken to mean that the flow is Dunn et al. [17] noted that the spanwise location of the flow fea-
more aligned, in magnitude and/or trend, to the design flow con- tures were over-predicted, but the trends were the same and the
ditions. “Spanwise location” was measured from the hub (0% magnitudes were sufficiently captured.
span) to the casing (100% span), but experimental data was not
captured at these points to preserve the probe. The overlaid con-
tours represent the ensemble averaged data that was used to gen- INCREASED LOADING CASE: The most notable dif-
erate the velocity profiles in the underlying plots. ference in Fig 9a and 9b is the deviation in the tip leakage com-

5 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Multiples of blade passing frequency (-) Multiples of blade passing frequency (-)
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
100 100
90 90
80 80
70 70
Spanwise Location (%)

Spanwise Location (%)

60 60
50 50
40 40
30 30
20 20
10 10

1 m/s
1 m/s

1 m/s
0 0
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44
Velocity magnitude (m/s) Velocity magnitude (m/s)

(a) Annular endwall (b) Contoured endwall

FIGURE 6: INCREASED LOADING CASE: COMPARISON OF VELOCITY MAGNITUDE DOWNSTREAM OF THE ROTOR

Multiples of blade passing frequency (-) Multiples of blade passing frequency (-)
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
100 100

90 90

80 80

70 70
Spanwise Location (%)

Spanwise Location (%)

60 60

50 50

40 40

30 30
20 20
10 10
1 m/s

1 m/s
0 0
10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42
Velocity magnitude (m/s) Velocity magnitude (m/s)

(a) Annular endwall (b) Contoured endwall

FIGURE 7: REDUCED LOADING CASE: COMPARISON OF VELOCITY MAGNITUDE DOWNSTREAM OF THE ROTOR

parison between the numerical and experimental results. Sned- The contour plots in Fig. 10 and Fig. 11 show that at approx-
den et al. [15] found a similar difference, however there was a imately mid span the hub endwall secondary flow vortex system
difference between the annular and contoured results evident in has reduced in strength for the contoured rotor, as can be seen
both numerical and experimental results. by the reduced underturning (blue contour at “A”) and overturn-
ing (red contour at “B”). In other words there was an increase
As seen by Dunn et al. [19] for the design case, the compari- in relative yaw angle in the under-turned region, while having
son of the flow in the tip leakage region was not as well predicted a reduction to relative yaw angle in the overturned region. The
as would be expected. It was speculated that this difference was contoured rotor relative outlet flow angle having one less contour
due to the experimental setup that was independent of measure- level. The relative velocity magnitude contour in Fig. 11 shows a
ment technique. Both steady and unsteady experimental results hub endwall secondary flow vortex structure which was slightly
(using different measurement principles, 5 hole probe as opposed reduced in size in both azimuthal and spanwise directions.
to hotfilm) show similar differences in the tip leakage region and
therefore cannot be attributed to sampling technique. Looking at the tip leakage vortex in Fig. 10 and 11 it can be

6 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 100 100

90 90

80 80

70 70
Spanwise Location (%)

Spanwise Location (%)

60 60

50 50

40 40

30 30

20 20

10 X-probe Annular 10 X-probe Annular

X-probe Contoured X-probe Contoured
0 0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6
Velocity Magnitude (m/s) Velocity Magnitude (m/s)

(a) Increased loading case (b) Decreased loading casel

FIGURE 8: COMPARISON OF THE ANNULAR AND CONTOURED FFT MAGNITUDE AT THE BLADE PASSING FREQUENCY

1000
Design Annular Contoured Annular CFD Contoured CFD
Relative Yaw Angle (deg)
(%)
an
Sp

0 10 20 30 40 50 60 70 80

100 100
90 90
80 80
70 70
60 60
Span (%)

Span (%)

50 50
40 40
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70
Relative Yaw Angle (deg) Relative Velocity (m/s)

(a) Relative outlet flow angle (b) Relative velocity magnitude

FIGURE 9: INCREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD AND EXPERIMENTAL
RESULTS

seen that the tip region (specifically in the region of the vortex) in blade loading. The numerical and experimental results show
for both the annular and contoured have negligible differences. reasonable correlation, except at the tip and hub endwall sec-
ondary flow vortex as discussed previously.

DECREASED LOADING CASE: As is to be expected,

the decreased loading case presents with a much smaller hub end- DISCUSSION
wall secondary flow vortex system. Little difference between the The unsteady experimental results show that the contoured
annular and contoured results was anticipated, since the endwall rotor had an improved rotor exit flow field at off design condi-
contour was designed to control the hub endwall secondary flow tions. The velocity profile was improved with the fluctuations
vortex. Fig. 12 to Fig. 14 shows that the annular and contoured in the flow being reduced as well, producing a more temporally
results show negligible differences. Therefore it can be seen that constant velocity profile. The CFD results show that this was due
the effect of the contoured rotor has been reduced by a decrease to a reduced hub endwall secondary flow vortex structure, as was

7 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 (a) Annular endwall (b) Contoured endwall

FIGURE 10: INCREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD RELATIVE OUTLET
FLOW ANGLE CONTOURS

(a) Annular endwall (b) Contoured endwall

FIGURE 11: INCREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD RELATIVE VELOCITY
MAGNITUDE

found for the design case, [19]. and modified to prevent the blade from rubbing against the cas-
The steady CFD analysis of Snedden et al. [15] showed that ing. Transfer of the engineering blue to the casing indicated the
the tip leakage flow of the annular and contoured rotor were sig- blades that rubbed, but it also indicated the region of rub on the
nificantly different when compared to the difference seen in the casing. Using this it was found the axial chord of the rotor blade
unsteady results presented above. Since Snedden et al. [15] used at the tip was smaller than the region of rub (i.e. the axial chord
the same mesh as used in the current investigation, using the during operation). Which implies that the blade (which was man-
same turbulence model and solver settings, thus it was felt that ufactured from polyamide) flexes in the tip region during testing.
the differences in annular and contoured steady state numerical The rotor exit relative yaw angle in Fig. 9a and 12a agrees well
results due to the tip leakage flows was due to unsteadiness in the with this as it shows the flow under-turns in both increased load-
flow that could not be dealt with using a steady state solver. ing and decreased loading cases. With the rotor blade acting as
In order to investigate the cause of the experimental differ- a cantilever with the thinnest point being the blade tip it was ex-
ence of the contoured and annular rotor tip flow in this and previ- pected that the influence of this flexure would reduce with span,
ous investigations, the test rig was scrutinised. At the start of the yet be consistent between annular and contoured rotor. This is
test campaign it was found that some of the blades rubbed against born out of the correlation of the numerical and experimental re-
the casing. Using a high contrast dye these blades were identified sults below 70% span. The extent of the flexure will form part of

8 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 1000
Design Annular Contoured Annular CFD Contoured CFD
Relative Yaw Angle (deg)

(%)
an
Sp
0 10 20 30 40 50 60 70 80

100 100
90 90
80 80
70 70
60 60
Span (%)

Span (%)
50 50
40 40
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80 0 10 20 30 40 50 60 70
Relative Yaw Angle (deg) Relative Velocity (m/s)

(a) Relative outlet flow angle (b) Relative velocity magnitude

FIGURE 12: DECREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD AND EXPERIMENTAL
RESULTS

(a) Annular endwall (b) Contoured endwall

FIGURE 13: DECREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD RELATIVE OUTLET
FLOW ANGLE CONTOURS

a future fluid-structure interaction investigation. TABLE 5: DIFFERENCE OF THE EXPERIMENTAL ANNU-

LAR AND CONTOURED ROTOR VELOCITY AT 31% SPAN
COMPARED TO THE ANNULAR RESULTS
Table 5 provides an indication of how the hub endwall sec-
Loading Ct velocity Cx velocity Velocity magnitude
ondary flow vortex systems changes with loading for both the
annular and contoured rotor at the location of peak over-turned m/s (%) m/s (%) m/s (%)
flow for design loading of the annular endwall. It can be seen
Increased -1.29 (-0.37) -2.57 (-7.98) -2.90 (-7.77)
that for all cases the hub endwall contour has reduced the veloc-
ity in question by some amount. The exception was the decreased Design -1.10 (-10.46) -1.90 (-6.12) -2.20 (-6.71)
loading case tangential velocity which increased by 0.23 m/s Decreased 0.23 (19.77) -1.25 (-4.24) -1.23 (-4.16)
(from the annular value of 1.14 m/s). Therefore it can be seen
that as the loading increases so too does the reduction in the over-
turned flow.

9 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 (a) Annular endwall (b) Contoured endwall

FIGURE 14: DECREASED LOADING CASE: COMPARISON OF ANNULAR AND CONTOURED CFD RELATIVE VELOCITY
MAGNITUDE

Conclusions turbomachines, Vol. AGARD-CP-214 Paper 11. VKI PP

It has been shown that a generic hub endwall contour op- 1977-11.
timised for a linear cascade has no detrimental effects on the [2] Hodson, H., and Dominy, R., 1987. “Three-dimensional
unsteady flow field of an annular rotating turbine test rig at off- flow in a low pressure turbine cascade at its design condi-
design conditions. In general all the flow features that were found tion”. Transaction of the ASME Journal of Turbomachin-
to have changed in the design case were similarly changed at off- ery, 109, pp. 177–185.
design, but the extent of the change was different depending on [3] Ingram, G., 2003. “Endwall profiling for the reduction
the loading. of secondary flow in turbines”. PhD thesis, University of
Durham, July.
1. The hub endwall secondary flow vortex system strength has [4] Snedden, G., Dunn, D. I., Ingram, G. L., and Gregory-
been reduced by a contoured rotor endwall at off design con- Smith, D. G., 2009. “The application of non-axisymmetric
ditions. The effect of the contour was found to reduce with endwall contouring in a single stage, rotating turbine”. In
decrease in loading (i.e. less effect on weaker hub endwall ASME Turbo Expo, GT2009-59169.
secondary flow vortex). [5] Saravanamuttoo, H., Cohen, H., and Rogers, G., 2001. Gas
2. The fluctuations in the flow in the region of the hub endwall Turbine theory, 5th ed. Pearson Education Limited.
secondary flow vortex were reduced. [6] Hawthorne, W., 1955. “Rotational flow through cascades”.
3. Changes to the experimental tip leakage flows seen previ- Journal of Mechanical and Applied Mathematics, 3.
ously were due to blades untwisting, while numerical insta- [7] Denton, J., and Pullan, G., 2012. “A numerical inves-
bility in the tip region created a similar change to the steady tigation into the sources of endwall loss in axial flow
results (but which was not present in the unsteady numerical turbines”. In Proceedings of ASME Turbo Expo 2012,
results). GT2012-69173.
[8] Zweifel, O., 1945. The spacing of turbo-machine blading,
especially with large deflection. The Brown Boveri Review,
ACKNOWLEDGMENT December.
This document is the result of a research effort funded [9] Snedden, G., 2011. “The application of non-axisymmetric
by the Defence Research Development Board under contract endwall contourning in a 1.5 stage, rotating turbine”. PhD
KT521910. thesis, School of Engineering, Durham University.
[10] Dunn, D. I., Snedden, G., and von Backström, T. W., 2013.
“Unsteady effects of a generic non-axisymmetric endwall
contour on the rotor of a 1 stage low speed turbine test rig.”.
REFERENCES
In ASME Turbo Expo 2013: Power for Land Sea and Air,
[1] Marchal, P., and Sieverding, C., 1977. “Secondary flows
GT2013-94961.
within turbomachine bladings”. In Secondary flows in

10 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 [11] NREC, 1972. The design and performance analysis of
axial-flow turbines: Volumes I and II. Northern Research
and Engineering Corporation.
[12] Snedden, G., Dunn, D., Ingram, G., and von Backström,
T., 2010. “Observations on the selection of objective func-
tion for the optimisation of turbine endwalls using compu-
tational fluid dynamics”. In SACAM10, South African As-
sociation for Theoretical and Applied Mechanics.
[13] Perlman, B. S., and Auerbach, V. H., 1977. “A phase-
locking technique for estimating the ensemble average of
time-series data”. In IEEE Transactions on Acoustics,
Speech and Signal Processing, Vol. ASSP-25, pp. 295–299.
[14] Joslyn, H., Dring, R., and Sharma, O., 1983. “Unsteady
three-dimensional turbine aerodynamics”. Transactions of
ASME Journal of Engineering for Power, 105, pp. 322–
331.
[15] Snedden, G., Dunn, D., Ingram, G., and Gregory-Smith, D.,
2010. “The performance of a generic non-axisymmetric
end wall in a single stage, rotating turbine at on and off-
design conditions”. In ASME Turbo Expo 2010: Power for
Land Sea and Air, GT2010-22006.
[16] Dunn, D. I., Snedden, G., and von Backström, T. W.,
2011. “Numerical investigation into the unsteady effects
of non-axisymmetric turbine endwall contouring on sec-
ondary flows”. In ISABE-2011, ISABE-2011-1715.
[17] Dunn, D. I., Snedden, G., and von Backström, T. W., 2009.
“Turbulence model comparisons for a low pressure 1.5
stage test turbine”. In ISABE-2009-1258.
[18] Numeca, 2007. User Manual FINE/Turbo v8 (including
Euranus) Documentation v8a, v8a ed. Numeca Interna-
tional, 5, Avenue Franklin Roosevelt, 1050 Brussels, Bel-
gium, October.
[19] Dunn, D. I., Snedden, G., and von Backström, T. W.,
2013. “An unsteady numerical analysis of a generic non-
axisymmetric turbine endwall in a 1.5 stage turbine test
rig”. International Aerospace Symposium of South Africa,
Aeronautical Society of SA.

11 Copyright © 2014 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 02/01/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use