See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/267644620

HP Vane Aerodynamics and Heat Transfer in the Presence of Aggressive Inlet

Swirl

Article in Journal of Turbomachinery · March 2013

DOI: 10.1115/GT2011-46037

CITATIONS READS

37 924

3 authors:

Imran Qureshi Smith A.

University of Birmingham Rolls-Royce
17 PUBLICATIONS 621 CITATIONS 6 PUBLICATIONS 176 CITATIONS

SEE PROFILE SEE PROFILE

Thomas Povey
University of Oxford
110 PUBLICATIONS 1,672 CITATIONS

SEE PROFILE

All content following this page was uploaded by Thomas Povey on 21 September 2015.

The user has requested enhancement of the downloaded file.

 HP Vane Aerodynamics and Heat
Transfer in the Presence
of Aggressive Inlet Swirl
Modern lean burn combustors now employ aggressive swirlers to enhance fuel-air mixing
and improve flame stability. The flow at combustor exit can therefore have high residual
swirl. A good deal of research concerning the flow within the combustor is available in
Imran Qureshi1 open literature. The impact of swirl on the aerodynamic and heat transfer characteristics
Department of Engineering Science, of an HP turbine stage is not well understood, however. A combustor swirl simulator has
University of Oxford, Parks Road, been designed and commissioned in the Oxford Turbine Research Facility (OTRF), previ-
Oxford, OX1 3PJ, UK ously located at QinetiQ, Farnborough UK. The swirl simulator is capable of generating
e-mail: [email protected] an engine-representative combustor exit swirl pattern. At the turbine inlet plane, yaw and
pitch angles of over 640 deg have been simulated. The turbine research facility used for
Andy D. Smith the study is an engine scale, short duration, rotating transonic turbine, in which the non-
Turbine Sub-systems, dimensional parameters for aerodynamics and heat transfer are matched to engine condi-
Rolls-Royce PLC, Moor Lane, tions. The research turbine was the unshrouded MT1 design. By design, the center of the
Derby, DE24 8BJ, UK vortex from the swirl simulator can be clocked to any circumferential position with
respect to HP vane, and the vortex-to-vane count ratio is 1:2. For the current investiga-
Thomas Povey tion, the clocking position was such that the vortex center was aligned with the vane lead-
Department of Engineering Science, ing edge (every second vane). Both the aligned vane and the adjacent vane were
University of Oxford, Parks Road, characterized. This paper presents measurements of HP vane surface and end wall heat
Oxford, OX1 3PJ, UK transfer for the two vane positions. The results are compared with measurements con-
ducted without swirl. The vane surface pressure distributions are also presented. The ex-
perimental measurements are compared with full-stage three-dimensional unsteady
numerical predictions obtained using the Rolls Royce in-house code Hydra. The aerody-
namic and heat transfer characterization presented in this paper is the first of its kind,
and it is hoped to give some insight into the significant changes in the vane flow and heat
transfer that occur in the current generation of low NOx combustors. The findings not
only have implications for the vane aerodynamic design, but also for the cooling system
design. [DOI: 10.1115/1.4006610]

Keywords: HP NGV, heat transfer, combustor swirl, aerothermodynamics, transonic tur-

bine, vane, stator, turbulence

1 Introduction bly, the flow incidence angle to the HP vane, the loading distribu-
tion on the HP vane, the inlet velocity profile, the inlet total
Highly swirling flows are now widely employed in gas turbine
pressure profile, the free-stream turbulence, the inlet boundary
combustors in order to improve the combustion aerodynamics.
layer profile, and the HP secondary flow characteristics would be
Swirl enhances the fuel-air mixing and thus reduces peak tempera-
affected. These changes in turn affect the aerodynamic and heat
tures, which reduces NOx emissions. In addition, the recirculation
transfer characteristics, and the performance of an HP turbine.
zone that develops as a result of strong swirl-induced axial and ra-
The influence of these parameters on the HP vane has been stud-
dial pressure gradients aids flame stability. Significant research is
ied in isolation (mostly in the linear cascade studies), but a com-
available in the open literature on these topics.
prehensive study incorporating the combined effect of these
Perhaps because of the rapid development in low-NOx high-
parameters in a flow representative of a low-NOx combustor has
swirl combustors, there has been little research that addresses the
not yet been reported.
influence of high residual swirl on the HP turbine. This paper
Studies relevant to the individual effects, which are combined
presents the first experimental investigation of the effects of high
in the research reported in this paper, are reviewed below.
swirl on HP vane aerodynamics and heat transfer.

2 Review of Past Literature 2.1 Off-Design Incidence and Load Distribution. A num-
A number of experimental and computational studies have been ber of HP turbine cascade studies have been reported which com-
reported which deal with the effects of swirl on combustion and pare the performance of turbine blades at design and off-design
emission characteristics, e.g., [1–3]. High residual swirl would be inlet conditions. Jouini et al. [4] investigated the effects of off-
expected to influence the flow in the first stage turbine. Most nota- design exit Mach number, Reynolds number, and inlet flow inci-
dence angle. Significant changes were observed in the loading
1
Corresponding author. Present address: Rolls-Royce PLC, PCF-2, P.O. Box 31, distribution in the vicinity of the leading edge for off-design inci-
Derby, DE24 8BJ. dence angle. For the design exit Mach number, an increase in the
Contributed by the International Gas Turbine Institute (IGTI) of ASME for publi-
cation in the JOURNAL OF TURBOMACHINERY. Manuscript received October 19, 2011;
losses was observed at off-design incidence angles. Corriveau and
final manuscript received October 26, 2011; published online November 19, 2012. Sjolander [5,6] investigated the influence of loading distribution
Editor: David Wisler. on the performance of HP turbine blades. They observed a

Journal of Turbomachinery Copyright V

C 2013 by ASME MARCH 2013, Vol. 135 / 021040-1

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 reduction of about 20% in the total pressure losses for an aft-
loaded airfoil as compared to the baseline airfoil [5].
Weiss and Fottner [7] studied the effects of loading distribution
on secondary flows by considering two highly loaded turbine cas-
cades having similar overall loading but different loading distribu-
tions. They observed higher secondary loss for the front-loaded
cascade. A similar result was observed by Benner et al. [8] who
studied the influence of leading-edge geometry on the secondary
flow losses and concluded that leading-edge geometry was less
significant than loading distribution, which significantly affected
the passage vortex and thus the secondary losses.
Fig. 1 Schematic of the Oxford turbine research facility
2.2 Combustor Turbulence and Heat Transfer. Measure-
ments performed at the exit of gas turbine combustors show the
flow. Jacobi and Shah [19] reviewed studies of heat transfer
turbulence intensity ranges from 15% to 40% [9]. Such high levels
enhancement using longitudinal vortices. The overall (averaged)
of turbulence have a significant effect on the heat transfer to the
HP turbine stage. Hancock and Bradshaw [10] and Blair [11] heat transfer enhancement observed in the studies reviewed
ranged up to 100% with local heat transfer enhancement up to
developed correlations for the enhancement of heat transfer as a
200% in some cases.
result of increased freestream turbulence through their studies on
flat-plate fully-turbulent boundary layers. Krishnamoorthy and
Sukhatme [12] performed experimental investigations using vane/ 3 Experimental Facility
blade cascade geometry and correlated surface heat transfer coef- The Oxford Turbine Research Facility (OTRF), previously
ficient enhancement with freestream turbulence for the laminar called Turbine Test Facility (TTF) when sited at QinetiQ Farnbor-
and turbulent boundary layer regions. Ames et al. [13] measured ough, is a short duration transonic light-piston facility. It can be
vane surface heat transfer for different inlet turbulence conditions operated as a single or 1.5 stage turbine, and for the current study
in a large-scale low-speed linear cascade facility and observed was operated with the full-scale MT1 research turbine. The main
augmentation of heat transfer with high inlet turbulence (dry low components of the facility are (i) the high pressure reservoir; (ii)
NOx DLN combustor model with Tu  14%) compared to the low the pump-tube that contains a light-weight piston; (iii) a fast act-
inlet turbulence (combustor models with Tu  1%). They also ing plug valve; (iv) the turbine stage (working section); (v) the
observed early boundary layer transition on the SS with increased turbobrake. These are highlighted in the schematic of the research
turbulence. Similar observations were reported by Radomsky and facility shown Fig. 1.
Thole [14]. Nasir et al. [15] studied the effects of inlet turbulence The operating principles of this type of facility were first
on vane heat transfer in a transonic cascade facility and also described by Jones et al. [20]. Prior to an experimental run the
observed augmented heat transfer on both PS and SS of the vane plug valve is closed, the working section and exhaust tank are
with high freestream turbulence, and earlier transition on the SS. evacuated, and the turbine disc is accelerated to the design speed
However for exit Mach number of approximately unity, the aug- using an air motor. Air from the high pressure reservoir is then
mentation on the SS was comparatively small and no effect was injected into the piston tube behind the light piston. The piston
observed on the location of boundary layer transition. moves down the piston tube acting to compress and heat the air in
front of it (approximately) isentropically. When the desired test
2.3 Secondary Flows, Vorticity, and Heat Transfer. The gas conditions (of pressure and temperature) are achieved in this
fundamental features of secondary flow patterns in nozzle guide compression, the fast-acting plug valve is opened and the test gas
vane (NGV) passages were proposed by Langston [16] and Sie- (air) flows out of the piston tube into the large annulus. Here the
verding [17], among others. This early work was conducted in lin- gas settles in an annulus which is similar to combustor geometry,
ear cascades, and although highly significant in establishing the before passing through the working section. The test run ends
basic mechanisms that apply in all cascades (both linear and annu- when the piston reaches the end of the piston tube, at which point
lar) it is now acknowledged that the flow pattern in the annular the test gas has been processed by the turbine. Steady conditions
cascade is very different from the linear counterpart. For this rea- are achieved for approximately 500 ms, during which the experi-
son there is now more emphasis on the annular experiment, or the mental data are acquired. During the test period all relevant nondi-
full rotating turbine experiment, where there is a desire to accu- mensional parameters are matched to engine conditions. The
rately understand the extent of secondary flow vortices generated turbobrake [21] ensures approximately constant turbine speed dur-
in a modern gas turbine engine, and the impact of these secondary ing the run. The OTRF has been used to test a HP turbine stage
flows on loss. Annular cascades tend to have more restricted [22] and a 1.5 stage turbine [23] in the past.
regions of secondary flow, with less cross-passage flow than sug- The nominal operating conditions of the research facility for
gested by studies in linear cascades. In both linear and annular the MT1 turbine stage, used in the current study, are listed in
cascades, the generation of secondary flow vortices has been Table 1.
found to affect the local heat transfer characteristics on the end
walls, and on the vane SS near the end walls.
In the situation where the combustor exit flow has residual Table 1 MT1 turbine stage operating conditions
swirling motion, the flow at the inlet to the turbine stage will have
additional vorticity. This vorticity is propagated through the tur- Parameter Nominal Allowable run-to-run
bine stage which accordingly changes the aerodynamic and heat (unit) value variations around nominal value (%)
transfer characteristics. It is expected that the additional vorticity
will increase the mixing of boundary layer and freestream fluid p01 (bar) 4.6 61
T01 (K) 444 62
which will lead to an enhancement in heat transfer.
Tg/Tw 1.50 62
The generation of vortices is encouraged in heat exchangers for M2hub 1.054 61
this reason; the vortices having the effect of promoting mixing M2casing 0.912 61
and therefore thinning of the boundary layer. Yilmaz et al. [18] x (rpm) 9500 61
observed an augmentation of up to 98% in Nusselt number in their p02rel (bar) 2.697 61
swirling flows experiments compared to the baseline no-swirl

021040-2 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 4 Pitch angle profile measured 0.7 Cax upstream of NGV
inlet plane with inlet swirl

Fig. 2 The working section of the OTRF with the HP turbine

stage and turbobrake highlighted

The working section of the OTRF is shown in Fig. 2. The MT1

research turbine has 32 NGVs and 60 rotor blades. The NGV true
chord and the exit swirl angle are 70.1 mm and 74.4 deg at the
midspan location. The vane has an aspect ratio of 0.53 and an exit
Reynolds number of 2.6  106. For the current investigation, the
research facility was upgraded with the installation of an inlet
swirl generator. This is briefly described in the next section.

4 Inlet Swirl Simulator

In order to carry out detailed experimental investigations of the Fig. 5 Yaw angle profile measured 0.7 Cax upstream of NGV
effects of swirl on the HP turbine stage, an inlet swirl simulator inlet plane with inlet swirl
has been designed and commissioned in the Oxford Turbine
Research Facility. The target swirl profile at the turbine stage inlet
was based upon extreme exit swirl conditions for a low-NOx mod-
ern combustor. Peak yaw and pitch angles were over 640 degrees.
The development, installation, and commissioning of the swirl
simulator in the OTRF is discussed by Qureshi and Povey [24].
The swirl simulator module is designed to mount in the tunnel
flow path upstream of the OTRF inlet contraction. The module
allows rotation so that clocking of the vortex core with respect to
the NGV leading edge could be achieved. To aid in computational
fluid dynamics (CFD) comparison, and to allow repeatability
assessment around the annulus, an integer swirler-to-vane count
ratio of 1:2 was chosen (16 swirlers and 32 NGVs). The manufac-
tured swirl generation system (approximately 1 m diameter), is
shown in Fig. 3. Each swirler is composed of six flat-plate vanes
inclined at 40 deg to the axial direction (swirl number ¼ 0.7). The
flow conditions measured at the inlet to the HP turbine with inlet
swirl are presented in Sec. 5.
Fig. 6 Measured inlet secondary flow vectors (bold arrows)
and interpolated/extrapolated vectors (thin arrows)

5 Inlet Conditions With Swirl

Swirl profile measurements were performed using the rakes of
four-hole probes at about 0.7 axial chords upstream of the vane
inlet plane. Details of the necessarily unconventional four-hole
probes arrangements are discussed in Ref. [24]. Area survey was
conducted over one swirler pitch (or 2 NGV pitches). The results
for the measured pitch angle (b) and the yaw angle (a) are pre-
sented in Figs. 4 and 5, respectively.
The circular dots indicate the points at which pitch and yaw
measurements were possible. Because of the high compound swirl
angles, at some points it was not possible to reduce the data using
the calibration maps.
These regions have been filled with the results at the nearest
measured locations in the area plots. The profiles of pitch and yaw
angle correspond to a well-defined clockwise vortex (as viewed
from upstream) of approximately the desired target magnitude.
Fig. 3 The assembled inlet swirl simulator module The corresponding flow vectors are presented in Fig. 6. The thick

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-3

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 used to refer to the vanes similarly disposed with respect to the
vortex.
An inlet total temperature survey was also performed with inlet
swirl. No significant variation from the nominal uniform inlet
temperature condition was observed.

6 Experimental Instrumentation and Data Reduction

Experimental pressure and heat transfer measurements were
conducted on the HP vane surface and the end wall. A brief
description of the instrumentation and data processing is given
below.

6.1 NGV Pressure Measurements. The HP vane static pres-

sure distributions were measured at 10%, 50%, and 90% span
with and without inlet swirl. Pneumatic tappings were distributed
over a set of vanes to achieve a good resolution of data at each
span location. Instrumented vanes were assembled in a removable
Fig. 7 Yaw angle profile at 20% and 80% span; comparison of
measurements in the OTRF with the target profile
cassette.

6.2 NGV Heat Transfer Measurements. Heat transfer

measurements were performed at 10%, 50%, and 90% spans on
the SS and PS of the HP NGV. Measurements were obtained with
and without inlet swirl. Thin-film platinum-resistance gauges
(0.2  2 mm in size) were used. A two-layered substrate was
used: an insulating layer of KaptonTM (polyamide) on a (ther-
mally) semi-infinite layer of metal. The theory for the use of mul-
tilayered gauges is given in Doorly and Oldfield [25].
For the uniform inlet condition, one vane was instrumented for
each of PS and SS at each span location. However for inlet swirl,
two vanes (labeled n1 and n2) were instrumented for each of PS
and SS at each spanwise location. The heat flux was evaluated
from the measured surface temperature time history using the
impulse response method, as described by Oldfield [26]. The heat
flux data was reduced to Nusselt number using the vane inlet total
Fig. 8 Normalized inlet total pressure profile with swirl
temperature and the vane true chord as given by Eq. (1). The ther-
mal conductivity was evaluated at the inlet total temperature,

_
qC
NuT01 ¼ (1)
ðT01  Tw ÞkT01

6.3 End Wall Heat Transfer Measurements. Heat transfer

measurements were obtained on the vane hub and casing end
walls using approximately 50 thin-film platinum-resistance gauges
on a MacorTM (machineable glass) single layer substrate. Heat
transfer electrical analog (HTA) circuits [27] were used to convert
measured resistance (surface temperature) into heat flux.
Using the measured heat flux data, the temperature history was
reconstructed for each gauge as described in Oldfield et al. [28].
Fig. 9 Schematic to show notation used to refer to HP vane During a run, the heat flux drops with time as the surface tempera-
geometric alignment with respect to the vortex center ture rises. By extrapolated regression of heat flux against recon-
structed temperature the local adiabatic wall temperature at each
gauge location was obtained [28]. Nusselt number, based upon the
arrows represent locations where a measurement was possible.
vane true chord, was obtained using the relation given in Eq. (2).
Points based on interpolated/extrapolated data are marked as thin
The thermal conductivity of air was evaluated at mean inlet total
arrows.
temperature,
The yaw angle profile measured in the OTRF is compared with
the target profile in Fig. 7. Results are presented at 20% and 80% _
qC
span. The measured peak yaw magnitude (approximately 50 deg) Nu ¼ (2)
is well over the target value indicating a well-formed clockwise ðTaw  Tw ÞkT01
vortex as desired.
The inlet total pressure profile was also determined using the The end wall heat transfer measurements with inlet swirl were not
four-hole probe data. The measured nondimensional pressure profile within the scope of the present program. The measurements with
is presented in Fig. 8. For the uniform inlet condition, the total pres- uniform inlet conditions are presented for CFD validation, fol-
sure variation from the nominal mass-mean value at the inlet plane lowed by a discussion of the CFD predicted changes in heat trans-
was approximately 60.3%. With inlet swirl a variation of 61.5% fer with inlet swirl.
from the nominal mass-mean value of total pressure was observed.
The notation C1 and C2 is used to refer to leading edge point 6.4 Measurement Uncertainty. A detailed treatment of
aligned with the vortex center (C1) and the leading edge between uncertainty in the heat transfer measurement process in the tran-
vortices (C2). This is illustrated in Fig. 9. The notation is also sient turbine experimental environment is provided by Povey

021040-4 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 2 Estimated precision and absolute uncertainty in meas-
ured and derived variables

Precision Absolute uncertainty

Variable Uniform Swirl Uniform Swirl

T01 61 K 63 K 65 K 67 K
Tw 61.5 K 61.5 K 61.5 K 61.5 K
q_ 1.5% 1.5% 3.4% 3.4%
Taw 61 K 63 K 65 K 67 K
Nu 1.9% 3.6% 5.6% 7.6%

[29]. Estimated precision and absolute uncertainties for uniform Fig. 10 Predicted difference of inlet incidence angle with inlet
inlet conditions and for inlet swirl are presented in Table 2 below. swirl; 0.25 axial chords upstream of the vane leading edge

7 Numerical Simulations
Three-dimensional computational simulations were conducted
with and without inlet swirl and the predictions were compared
with experimental results. The measured values of pitch and yaw
(Figs. 4 and 5) were used as inlet boundary conditions for the
swirl simulations. Uniform total pressure fields were used for both
the uniform simulation and that with inlet swirl, and this must
be considered when interpreting results (see measured profile,
Fig. 8).
The swirler-to-vane count ratio was 1:2, so a structured compu-
tational grid comprised of two NGVs and four rotor blades (actual
vane/blade count is 32/60), was developed using the Rolls-Royce
in-house tool PADRAM, having over 5  106 mesh cells. The
rotor airfoils were skewed by a small angle to reproduce the rotor Fig. 11 Predicted streamlines for uniform flow (blue) and flow
capacity of the original geometry. The computational solutions with inlet swirl (red), showing shift of the stagnation line
were obtained using the Rolls-Royce in-house CFD solver
HYDRA, which uses an efficient edge-based data structure, as
described by Moinier and Giles [30]. A MUSCL based flux- Fig. 10. The positions of the vane leading edges C1 and C2 are
differencing algorithm is used to integrate the flow equations marked.
around median-dual control volumes. Block Jacobi preconditioning Vane C1 is downstream of the vortex core, and is subject to the
of the discrete flow equations is applied [31] and a 5-stage Runge- largest change in incidence, approximately 50 deg at the casing
Kutta scheme is employed for convergence of the steady state solu- and þ 50 deg at the hub. Between vortex cores, in position C2, the
tion [32]. An element-collapsing multigrid algorithm is used to general positive circulation of flow near the casing, and negative
accelerate the convergence to steady state [33]. Nonlinear unsteady circulation of the flow at the hub is still evident, though less pro-
calculations were performed using implicit dual time-stepping. nounced. Here the incidence is approximately 30 deg at the cas-
Both steady and unsteady solutions were obtained. The results ing and þ20 deg at the hub. Near the midspan on both vane C1
from unsteady solutions were time-averaged for comparison with and C2 there is a small region of approximately nominal flow
steady solutions. The Spalart–Allmaras turbulence model was angle (0 deg), but the flow-field is otherwise dominated by high
implemented with wall functions. The boundary layers were mod- whirl angles.
eled as turbulent. For each case, two solutions were obtained at The incidence angle difference causes a significant change in
two isothermal wall temperatures, Tw1 and Tw2 , allowing the heat the position of the stagnation line on the vane leading edge. The
transfer coefficient to be determined from the heat fluxes using shift in the stagnation point near the hub and the casing is shown
Eq. (3). Adiabatic wall temperature was obtained by substituting using predicted streamlines in Fig. 11.
heat transfer coefficient in either Eq. (4) or Eq. (5). The Nusselt For vane C1, the stagnation point with inlet swirl moves
number was obtained using Eq. (2), towards the PS near the hub, and towards the SS near the casing.
The effect is pronounced, and sufficient to cause a change in the
ðq_ 1  q_ 2 Þ loading of the vane, the heat transfer and the loss. This is dis-
h¼ (3) cussed subsequently.
ðTw2  Tw1 Þ
On vane C2 a similar shift of the stagnation line is observed,
q_ 1 ¼ hðTaw  Tw1 Þ (4) though less pronounced at the casing (compared to C1) and close
q_ 2 ¼ hðTaw  Tw2 Þ (5) to negligible near the hub.
The change in flow incidence affects the loading distribution of
8 Results and Discussion the airfoil. The measured HP NGV isentropic Mach number distri-
butions at 10%, 50%, and 90% span with inlet swirl are presented
Experimental measurements and computational predictions of in Figs. 12, 13, and 14, respectively, for vane C1. Results obtained
pressure and heat transfer on the NGV and platform surfaces are with uniform inlet conditions are plotted for comparison. The
now presented and discussed. time-averaged unsteady CFD predictions are also presented for
comparison.
8.1 NGV Isentropic Mach Number. The presence of There is a significant increase in measured aerodynamic loading
intense swirl in the flow significantly alters the incidence angle at at 10% span, associated with the high positive incidence near the
the inlet of the HP vane. The predicted difference in incidence at a hub section of the vane. This is confined to the region of about
plane, 0.25 axial chords upstream of the vane leading edge, 0–40% axial chord on the suction surface, but is evident along the
between the case of inlet swirl and uniform conditions is shown in entire pressure surface.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-5

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 12 NGV isentropic Mach number at 10% span for uniform Fig. 15 NGV isentropic Mach number at 10% span for uniform
inlet conditions and inlet swirl for position C1; comparison of inlet conditions and inlet swirl for position C2; comparison of
measurements and CFD measurements and CFD

Fig. 13 NGV isentropic Mach number at 50% span for uniform

inlet conditions and inlet swirl for position C1; comparison of Fig. 16 NGV isentropic Mach number at 50% span for uniform
measurements and CFD inlet conditions and inlet swirl for position C2; comparison of
measurements and CFD

The computational predictions were generally in very good

agreement with the measurements on all the surfaces at each span
with the exception of 50%-PS (significant difference) and 10%-PS
(small difference). Here the change in isentropic Mach number
with inlet swirl was lower (50%-PS) or higher (10%-PS) in the
prediction than in the experimental case. This is likely due to the
approximation of uniform inlet total pressure in the computations.
In the experiment, vane C1 is subject to higher than average total
pressure in a region near the PS at 10% span, and lower than aver-
age total pressure in a region near the PS at 50% span. These non-
uniformities would act to lower the inferred isentropic Mach
number in the experiment at 10% span and increase it at 50%
span, which is consistent with the measurements. It is noted that
Fig. 14 NGV isentropic Mach number at 90% span for uniform the predictions are in excellent agreement for all span locations
inlet conditions and inlet swirl for position C1; comparison of for position C2 (presented below), where the nonuniformity in
measurements and CFD total pressure is lower.
The measured HP NGV isentropic Mach number distributions
with and without inlet swirl at 10%, 50%, and 90% span for vane
At 50% span, a reduction in the measured loading was C2 are presented in Figs. 15, 16, and 17, respectively. The general
observed, caused by an increase in the pressure surface Mach trends with inlet swirl are similar to those observed for vane C1,
number. At 50% span the SS isentropic Mach number was not sig- but the change with swirl is less significant. This is expected when
nificantly affected. The apparent change in PS isentropic Mach the whirl distribution is considered.
number may be an artifact of the low total pressure vortex core At 10% span, a small increase in loading was measured, which
being ingested predominantly on the PS of vane C1. At 30% axial is in-line with the small change in incidence expected, and
chord, a change from approximately M ¼ 0.15 to M ¼ 0.25 would observed in the CFD solutions. The decrease in loading at 50%
be consistent with a change in total pressure of 2.7%. Figure 8 was also relatively small.
reveals a deficit in total pressure of approximately 1.5%, which is At 90% span where the expected incidence difference was high-
the correct order of magnitude. This argument is also consistent est, measurements show a significant reduction in loading (though
with the disappearance of a stagnation region in the leading part less than for vane C1) in the region 0-50% axial chord. There is a
of the vane when presented in terms of isentropic Mach number. region of mild negative pressure between 5% and 30% axial chord
At 90% span, there is a decrease in loading with inlet swirl on the PS.
associated with high negative incidence in this region. This causes The agreement between CFD and experiment is good at all
a region of negative pressure gradient on the PS of the vane span heights on both the PS and SS of the vane. Figure 8 reveals
between approximately 5% and 30% span. relatively uniform total pressure at the inlet to the vane for

021040-6 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 17 NGV isentropic Mach number at 90% span for uniform
inlet conditions and inlet swirl for position C2; comparison of
measurements and CFD

Fig. 19 Predicted pressure loss coefficient difference at vane

exit between swirl and uniform conditions; position C2

the vane there is a region of lower loss (c). This is in the boundary
layer region, and is likely the result of lower surface velocities on
the PS at 10% span, resulting from high positive incidence. With
reference to Fig. 29, it is clear that the PS streamline pattern for
vane C1 has strong divergence of streamlines (compared with the
uniform inlet case) which leads to redistribution of loss from the
hub region (c) to the casing region where there is enhanced total
pressure loss in the whole of the boundary layer region (f). The
enhanced loss (f) is likely exacerbated by increased velocities and
possible separation of the PS boundary layer at the 90% span sec-
tion (see Fig. 14). Close to 90% span SS there is a region of signif-
icant loss reduction (e) which appears to be associated with lower
surface velocities (see Fig. 14) and streamline divergence (see
Fig. 29). The features (g) and (h) appear to be weakened and
strengthened corner vortices, respectively.
Fig. 18 Predicted pressure loss coefficient difference at vane Consider the change in total pressure loss with inlet swirl for
exit between swirl and uniform conditions; position C1 vane C2, presented in Fig. 19. The general picture is similar to
vane C1, but with reduced magnitude of most of the features. This
is in accord with general trends in Fig. 15 to Fig. 17, and the
position C2, making the approximation in the computational
streamline pattern in Fig. 29.
boundary conditions more sound at this location.
Loss core (a) is less enhanced than was the case for position
The measured changes in aerodynamic loading would cause a
C1, which is explained by the streamline pattern (Fig. 29) which
change in loss characteristics of the vane. This is examined with
shows less accumulation of SS boundary layer fluid in this region
reference to the computational results. The area plots of the differ-
for position C2. The weakening of the loss regions (d) and (e) on
ence in total pressure loss coefficient (where CPL is computed
the SS are less pronounced for C2 than for C1, which is in line
using Eq. (6)) evaluated at the vane exit plane, between the swirl
with the reduced change in vane loading. Likewise the PS bound-
inlet condition and uniform inlet condition are presented in Fig.
ary layer loss band (f) is less enhanced than for position C1, which
18 (C1) and Fig. 20 (C2). Results presented are viewed from
is the result of less lateral redistribution of loss on the PS (Fig. 29)
upstream to downstream,
and a reduction in the increase in surface velocity (Fig. 17) com-
ðP01  P02 Þ pared with the C1 position (Fig. 13). In the hub region on the SS,
CPL ¼ (6) the enhancement in loss core (b) is less than for position C1, and
ð0:5qu2 Þj02 the vortex is pushed closer to the end wall. This leads to an area
of reduced loss (g) which marks the position the vortex adopted in
Consider the change in total pressure loss with inlet swirl for vane the uniform inlet flow case.
C1, presented in Fig. 18. In the hub region there are two additional A comparison of the circumferentially-averaged pressure loss
loss cores (a) and (b) on the suction side of the vane. Loss core (a) coefficient at vane exit is presented in Fig. 20 for the uniform sit-
is close to the vane surface at approximately 15% span. Examina- uation, and for vanes with inlet swirl in position C1 and C2.
tion of the surface streamlines presented in Fig. 29 shows that For C1 an increase in loss is observed in the hub region which
there is significant downwash of flow on the SS with inlet swirl, extends to approximately 25% span. The peak at 10% span is
causing the boundary layer fluid to collect at approximately 15% associated with the displaced passage vortex (b), and the increase
span. This is matched by divergence of streamlines for flow arriv- between 10% and 25% span is associated with the redistribution
ing in the 20%–60% span region at the trailing edge (TE). This of the boundary layer loss from the casing to the hub region. This
spreading of boundary layer fluid accounts for the deficit in loss in is manifested as loss region (a) in Fig. 18. From 25% to 90% span
this region (d). The second loss core (b) suggests enhanced sec- loss is reduced. This is caused by the redistribution of boundary
ondary flow caused by the higher loading in the hub region. layer fluid to the hub region and the decreased loading of the vane
Higher loading leads to higher surface velocities over the majority in the casing region. The overall difference in loss between C2
of the vane SS, and also higher boundary layer loss. On the PS of and the uniform case is less striking than for C1.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-7

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 22 Predicted passage streamlines at 90%, span for inlet
swirl

Fig. 20 Predicted radial pressure loss coefficient distributions

downstream of NGV for uniform inlet conditions and for inlet
swirl in positions C1 and C2

Fig. 23 Predicted passage streamlines at 50%, span for inlet

swirl

Fig. 21 Predicted passage streamlines at 10%, 50%, and 90%

span for uniform inlet conditions

8.2 NGV Surface Heat Transfer Results. With inlet swirl,

the NGV surface heat transfer is affected by increased inlet turbu-
lence (compared to the uniform condition), changes in the second-
ary flow pattern, and changes in vane loading. Fig. 24 Predicted passage streamlines at 10%, span for inlet
The measured freestream turbulence intensity at the turbine swirl
inlet plane in the OTRF for uniform inlet flow was found to be
6.5% (Chana [34]). Beard [35] measured the turbulence intensity
with inlet swirl as 11.2%. The impact on heat transfer coefficient SS of vane C2, consistent with an additional vortex in the passage
of increased inlet turbulence was estimated using the correlation bounded (from L to R) by C1-C2. The same effect is observed in
given in Eq. (7) (see Ref. [12]). Using this, an increase in h of the passage bounded (from L to R) by C2-C1, but it is less
11% was estimated, pronounced.
At 10% span in passage C1-C2 there is significant upwash on
hx ¼ hðTu¼0Þ ð1 þ 0:0235Tux Þ (7) the PS of vane C1, and mild downwash on the SS of vane C2. The
flow in passage C2-C1 is comparatively less affected by inlet
Although this is significant, it might be expected that larger swirl.
changes would arise because of gross changes in the flow structure Computational results of secondary flow velocity components
within the passage. Figure 21 shows the predicted streamlines for can be helpful in visualizing the flow field. In Figs. 25–28 the dif-
the uniform case, obtained by tracking particles marked at the vane ference between the secondary flow velocity vectors for the uni-
inlet plane, at the 10%, 50%, and 90% span locations (where the heat form and swirl cases are presented.
transfer measurements have been performed). For the uniform inlet Figure 25 presents the difference in the secondary flow vectors
condition there is little radial movement of the streamlines, demon- at 0.25 axial chords upstream of the NGV inlet. Two clockwise
strating the relatively low secondary flow within the passage. vortices, generated by two neighboring upstream swirlers are
For inlet swirl, there is significant radial movement of stream- aligned with the leading edge of vanes in position C1. The vane in
lines as shown in Figs. 22, 23, and 24 for 90%, 50%, and 10% position C2 is approximately aligned with the middle of these vor-
span, respectively. tices as expected.
At 90% span the effect of the clockwise inlet swirl vortex is sig- At the vane inlet plane (Fig. 26), the vortex is divided by vane
nificant, causing significant upwash on the PS of vane C1. At this C1, resulting in a clockwise vortex within the passage (from L to
span, the SS of vane C1 and the PS and SS of vane C2 are less R) C1-C2. This causes upwash on the PS of vane C1 and down-
affected. wash on the SS of vane C2. In passage C2-C1 (from L to R) the
At 50% span the effect of swirl is most pronounced. There is effect is less pronounced, with the vortex being restricted to the
significant upwash on the PS of vane C1, and downwash on the right hand side (RHS) of the passage, near the SS of vane C1.

021040-8 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 25 Difference of secondary flow velocity vectors between
swirl and uniform; 0.25Cax upstream of inlet

Fig. 29 Predicted NGV surface flow streamlines, with and with-

out inlet swirl

Fig. 26 Difference of secondary flow velocity vectors between

swirl and uniform at NGV inlet plane

Fig. 30 NGV Nu at 10% span; uniform and swirl-C1

Figure 27 shows the difference in secondary flow vectors at

0.25 axial chords downstream of the vane inlet plane. In passage
C1-C2 the vortex has become more regular and filled the passage,
causing, in particular, significant upwash on the PS of vane C1. At
the same axial station, the vortex on the RHS of passage C2-C1
remains confined to this region of the passage.
At the vane exit plane the vortex structure is weakened, but
there is still evidence of upwash on the PS of vane C1.
The impact of these vortices on the NGV surface streamline
pattern is shown in Fig. 29. The streamline pattern is significantly
altered by inlet swirl. Of particular note is the extreme upwash on
the PS of vane C1, which causes boundary layer fluid to accumu-
Fig. 27 Difference of secondary flow velocity vectors between late at the 90% span location. This causes a region of increased
swirl and uniform; 0.25Cax downstream of inlet loss (f) illustrated in Fig. 18. The strong downwash on the SS of
vane C1 causes boundary layer fluid to accumulate at approxi-
mately 15% span, which is manifested in loss core (a) in Fig. 18.
Where there is streamline divergence there is a thinning of the
boundary layer, and lower loss. It is expected that the heat transfer
will follow an inverse relationship, being low where there is accu-
mulation of boundary layer fluid and high where there is thinning
of boundary layer fluid.
HP NGV surface heat flux was measured with uniform inlet
conditions and with inlet swirl, and reduced to Nusselt number as
described in Sec. 6.2. At each vane location (C1 and C2) a set of
two different vanes was instrumented so that repeatability com-
parisons could be performed. These vanes are termed as n1 and n2
for each surface-span in the following plots.
The results for inlet swirl for the NGV in position C1 are com-
pared to the uniform inlet data in Figs. 30, 31, and 32 for 10%,
50%, and 90% span, respectively.
At 10% span an increase in Nu is observed over most of the
Fig. 28 Difference of secondary flow velocity vectors between chord. On the PS of the vane the enhancement is significant, being
swirl and uniform, at NGV exit plane greatest on the leading part of the vane. This is a region of

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-9

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 31 NGV Nu at 50% span; uniform and swirl-C1 Fig. 33 NGV Nu at 10% span; uniform and swirl-C2

Fig. 32 NGV Nu at 90% span; uniform and swirl-C1 Fig. 34 NGV Nu at 50% span; uniform and swirl-C2

strongly diverging boundary layer fluid (see Fig. 29) in which

thinning is expected. Higher loading will also tend to increase sur-
face heat transfer coefficient, as observed—less significantly—on
the SS of the vane. It is noted that the SS Nu profile with inlet
swirl suggests earlier boundary layer transition than with uniform
inlet conditions. It should be recalled that increased turbulence in-
tensity may account for an increase of approximately 11% in the
heat transfer coefficient (an order of magnitude lower than the
observed changes). Vanes n1 and n2, which are exposed to differ-
ent vortices on account of the different positions in the annulus,
are in reasonably good agreement.
At 50% span, there is significant augmentation on both the PS
and SS with inlet swirl. On the PS Nu is increased by approxi-
mately between 50% and 200%, with the greatest changes near
the leading edge region. The profiles suggest earlier transition to a Fig. 35 NGV Nu at 90% span; uniform and swirl-C2
fully turbulent boundary layer. High augmentation is also caused
by boundary layer divergence. On the SS there is significant aug- described in Sec. 6.3. Competitive measurements with inlet swirl
mentation particularly in the leading region of the vane. This is were not part of the current program. In this section, experimental
likely caused by early transition to a turbulent boundary layer. and computational results are compared for the uniform inlet con-
Vanes n1 and n2 are in good agreement on the SS and in fair dition (by way of benchmarking the CFD), and the effects due to
agreement on the PS, suggesting sensitivity to the location of the inlet swirl are subsequently examined with reference to the com-
upstream vortex. putational predictions.
At 90% span the change in Nu with inlet swirl was minimal, Figure 36(a) shows the measured Nu at the hub end wall for
although it is noted that there were limited measurements in the uniform inlet conditions. The equivalent CFD predictions are pre-
leading region. Examination of the surface flow (Fig. 29) reveals sented in Fig. 36(b). The corresponding measurements and predic-
relatively little swirl induced secondary behavior in this region, tion on the casing are presented in Fig. 37. Measured data points
although it is perhaps surprising that there is not a reduction in Nu are shown as dots on the interpolated surface plots. Although the
on the PS at 90% span associated with the accumulation of bound- data are relatively sparse, the interpolated results can highlight
ary layer fluid (f) observed in Fig. 18. It is possible that this effect general trends.
is mitigated by higher inlet turbulence intensity. The overall trends in Nu distribution are similar between the
The results for the NGV in position C2 are presented in experiments and CFD predictions. There is an early region of low
Figs. 33, 34, and 35 for the 10%, 50%, and 90% span, respec- Nu, where there is low inlet flow velocity and a relatively thick
tively. The trends with inlet swirl are similar to those for vane inlet boundary layer. Nu increases in the region of higher flow ve-
position C1, though less pronounced. locity in the downstream region of the passage and near the SS of
the vane. Although the vane has minimal secondary flow, evi-
8.3 End Wall Heat Transfer Results. Heat flux measure- dence of a horseshoe vortex can be seen in the CFD and experi-
ments were performed on the hub and casing end wall of the NGV mental solutions. This is marked by a thin line of low Nu, with a
passage for uniform inlet conditions and were reduced to Nu as region of high Nu in the PS corner of the vane where the flow

021040-10 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 36 Hub end wall Nusselt number with uniform inlet condi-
tions (a) measured (b) predicted Fig. 38 Predicted surface flow streamlines, with and without
inlet swirl: hub end wall

Fig. 37 Casing end wall Nusselt number with uniform inlet

conditions (a) measured (b) predicted

Fig. 39 Predicted surface flow streamlines, with and without

reattaches with a thin boundary layer, having separated in the
inlet swirl: casing end wall
region of the horseshoe vortex. This effect is also described by
Harvey and Jones [36].
Near the hub and casing end wall, as the flow passes beyond the enhancement. Nu is higher by up to 25% in the region of the
point of maximum velocity (reached at approximately 80% axial crown due to increased velocity, caused by higher loading. There
chord—see Figs. 12 and 14), it is moderately diffused, and a is also enhancement in the PS TE region downstream of the horse-
reduction in Nu is seen. shoe vortex lift off line. In passage C1-C2, the relatively low Nu
The predicted surface streamline patterns on the hub and case line that marks the PS horseshoe vortex line is apparent, and
walls are shown in Figs. 38 and 39 for the case of uniform inlet appears to be enhanced due to the combination of high positive
conditions and inlet swirl. incidence on the vane and a clockwise vortex due to swirl. In pas-
Inlet swirl acts to increase incidence at the hub and decrease sage C2-C1 there is less evidence of a PS horseshoe vortex leg,
incidence at the casing. On the hub end wall this causes the stag- and the enhancement on the SS crown is less than for passage C1-
nation point to move towards the PS of the vane, increasing the C2. This is consistent with the loading distribution on the vane
loading in the section. Over-turning of the end wall boundary shown in Figs. 12 and 15.
layer fluid is mitigated by the clockwise vortex which propagates The predicted casing end wall Nu is compared in Fig. 41 for
through the passage, which offsets the effect of enhanced loading uniform inlet conditions and for inlet swirl. There is a small
of the hub section. enhancement on the SS crown of vane C1. This is not explained
On the casing end wall the effect of inlet swirl is a decrease in by an increase in surface velocities, and may be the result of
incidence, which lowers the loading on the vane. The strong reduced concentration of boundary layer fluid in the crown region
clockwise vortex in passage C1-C2 causes greater overturning of (see Fig. 39) causing higher momentum in the near wall region.
the end wall boundary layer fluid, an effect that is less apparent in The passage C1-C2 shows a large band of low Nu flow which sug-
passage C2-C1. gests accumulation of boundary layer fluid in this region (see
The predicted hub end wall Nu is compared in Fig. 40 for uni- Fig. 39) ahead of the PS leg of the horseshoe vortex. Behind the
form inlet conditions and for inlet swirl. The averaged Nu with separation line there is a region of high Nu. Passage C2-C1 is less
swirl is higher than for uniform conditions due to localized affected by inlet swirl.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-11

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 43 Casing end wall Nu, with and without inlet swirl

less good, the former being higher by as much as 20% at approxi-

mately 50% axial chord. The predicted effect of inlet swirl is to
reduce the circumferentially averaged Nu in the trailing part of
the passage by up to approximately 8%.
Fig. 40 Predicted hub end wall Nu (a) uniform (b) swirl

9 Conclusions
A detailed experimental and computational investigation of the
effect of high inlet swirl on the aerodynamic and heat transfer
characteristics of a HP vane has been conducted. Experiments
were performed in a rotating transonic turbine facility, which was
modified to incorporate an inlet swirl generator which replicated
conditions based on a low-NOx combustor design. It is believed
that this is the first study of this type.
The results of the study indicate significantly altered vane load-
ing distributions at 10% and 90% span. The stagnation point
migrates towards the PS at the hub and towards the SS at the cas-
ing. The result of high positive and negative incidence at the hub
and casing sections, when combined with a vortex which propa-
gates though the vane passage, is to significantly alter the second-
ary flow structure of the vane. The surface flow pattern on the
vane and end wall surfaces is altered such that the boundary layer
fluid is redistributed, causing it to collect in certain regions and
dissipate in others. In the loss characteristics of the vane the effect
of boundary layer fluid redistribution and altered loading charac-
teristic is clearly evident. In particular there are loss cores associ-
Fig. 41 Predicted casing end wall Nu (a) uniform (b) swirl
ated with regions where there is streamline convergence from the
vane surface flow, and regions of higher and lower loss across
large regions of the span associated with higher and lower loading
of the vane.
Heat transfer measurements on the vane surface demonstrate
significant changes in Nu with inlet swirl. The changes are
explained by streamline divergence and convergence on the sur-
face (causing accumulation and dissipation of boundary layer
fluid) and by changes in vane loading, and therefore surface
velocity.
Vane end wall heat transfer predictions compared well with
measurements conducted for uniform inlet conditions. Predictions
for the inlet swirl condition showed significant changes in end
wall Nu distribution, which can be explained with reference to the
end wall secondary flow field and vane loading distribution.
Higher SS crown velocities resulted in higher Nu in this region.
Fig. 42 Hub end wall Nu, with and without inlet swirl The highest Nu values were in a region downstream of the PS leg
of the horseshoe vortex, where reattachment occurs.
Perhaps the most striking observation is the extent of streamline
Circumferentially averaged Nu predictions on the hub and cas- redistribution on the PS of the vane geometrically aligned with
ing end wall are presented in Figs. 42 and 43, respectively, and the vortex center. Here, extreme streamline divergence would
compared to experimental data for uniform inlet conditions. On make cooling system design challenging. The findings indicate a
the hub end wall there is good agreement between the experimen- requirement to rethink the cooling philosophy for vanes operating
tal data for uniform inlet conditions and the prediction for the in this environment.
same. For inlet swirl there is little effect in the leading region of
the vane, but between 50% and 100% axial chord there is an
enhancement in heat transfer which rises to approximately 17%. Acknowledgment
On the casing end wall the agreement between the experimental The authors would like to acknowledge the financial support
data for uniform inlet conditions and the prediction for the same is provided by the EC for the TATEF II Project, and the technical

021040-12 / Vol. 135, MARCH 2013 Transactions of the ASME

Downloaded 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 staff who supported the instrumentation and experiments at [9] Goldstein, R. J., Lau, K. Y., and Leung, C. C., 1983, “Velocity and Turbulence
Oxford and QinetiQ Farnborough. Measurements in Combustion Systems,” Exp. Fluids, 1, pp. 93–99.
[10] Hancock, P. E., and Bradshaw, P., 1983, “The Effect of Freestream Turbulence
on Turbulent Boundary Layers,” ASME J. Fluids Eng., 105, pp. 284–289.
Nomenclature [11] Blair, M. F., 1983, “Influence of Free-Stream Turbulence on Turbulent Bound-
Romans and Greeks ary Layer Heat Transfer and Men Profile Development, Part II—Analysis of
Results,” ASME J. Heat Transfer, 105, pp. 41–47.
a¼ yaw angle, degrees [12] Krishnamoorthy, V., and Sukhatme, S. P., 1989, “The Effect of Freestream Tur-
b¼ pitch angle, degrees bulence on Gas Turbine Blade Heat Transfer,” ASME J. Turbomach., 111, pp.
C¼ NGV midspan true chord, mm 497–501.
q¼ density, kg/m3 [13] Ames, F. E., Wang, C., and Barbot, P. A., 2003, “Measurement and
Prediction of the Influence of Catalytic and Dry Low NOx Combustor Turbu-
h¼ heat transfer coefficient lence on Vane Surface Heat Transfer,” ASME J. Turbomach., 125, pp.
k¼ thermal conductivity 221–231.
M¼ Mach number [14] Radomsky, R. W., and Thole, K. A., 2002, “Detailed Boundary Layer Measure-
Nu ¼ Nusselt number ments on a Turbine Stator Vane at Elevated Freestream Turbulence Levels,”
ASME J. Turbomach., 124, pp. 107–118.
p¼ pressure, bar [15] Nasir, S., Carullo, J. S., Ng, W., Thole, K. A., Wu, H., Zhang, L. J., and Moon,
q_ ¼ heat flux, kW/m2 H. K., 2009, “Effects of Large Scale High Freestream Turbulence and Exit
T¼ temperature, K Reynolds Number on Turbine Vane Heat Transfer in a Transonic Cascade,”
Taw ¼ adiabatic wall temperature, K ASME J. Turbomach., 131, pp. 021021–1-11.
[16] Langston, L. S., 2001, “Secondary Flows in Turbines—A Review,” Ann. N.Y.
Tg ¼ gas temperature, K Acad. Sci., 934, pp. 11–26.
Tw ¼ wall temperature, K [17] Sieverding, C. H., 1985, “Secondary Flows in Straight and Annular Turbine
u¼ velocity, m/s Cascades,” Thermodynamics and Fluid Mechanics of Turbo, Vol. 2 (A86-
x¼ turbine speed (rpm) 29376 12-02), Martinus Nijhoff Publishers, Dordrecht, Germany, pp. 621–664.
[18] Yilmaz, M., Comakli, O., and Yapici, S., 1999, “Enhancement of Heat Transfer
by Turbulent Decaying Swirl Flow,” Energy Convers. Manage., 40, pp.
1365–1376.
Subscripts, Superscripts, Abbreviations [19] Jacobi, A. M., and Shah, R. K., 1995, “Heat Transfer Surface Enhancement
0¼ total (absolute) Through the Use of Longitudinal Vortices: A Review of Recent Progress,” Exp.
1¼ NGV inlet plane Therm. Fluid Sci., 11, pp. 295–309.
2¼ NGV exit plane [20] Jones, T. V., Schultz, D. L., and Hendley, A. D., 1973 “On the Flow in an Isen-
tropic Light Piston Tunnel,” MoD (Proc Exec), Aeronautical Research Council
a; ax ¼ axial distance R&M No. 3731.
rel ¼ rotor relative [21] Goodisman, M. I., Oldfield, M. L. G., Kingcombe, R. C., Jones, T. V., Ains-
PL ¼ pressure loss worth, R. W., and Brooks, A. J., 1992, “An Axial Turbobrake,” ASME J. Tur-
Exp ¼ experimental bomach., 114(2), pp. 419–425.
[22] Hilditch, M. A., Fowler, A., Jones, T. V., Chana, K. S., Oldfield, M. L. G., Ains-
OTRF ¼ Oxford Turbine Research Facility worth, R. W., Andrew, S. J., and Smith, G. C., 1994, “Installation of a Turbine Stage
TTF ¼ Turbine Test Facility in the Pyestock Isentropic Light Piston Facility,” ASME Paper No. 94-GT-277.
HP ¼ high pressure [23] Povey, T., Chana, K. S., Jones, T. V., and Oldfield, M. L. G., 2003, “The
HTA ¼ heat transfer analog Design and Performance of a Transonic Flow Deswirling System: An Applica-
tion of Current CFD Design Techniques Tested Against Model and Full-Scale
NGV ¼ nozzle guide vane Experiments,” Advances of CFD in Fluid Machinery Design, R. Elder, A. Tour-
Tu ¼ turbulence lidakis, and M. Yates, eds., John Wiley & Sons, Hoboken, NJ, pp. 65–94.
PS, SS ¼ pressure side, suction side [24] Qureshi, I., and Povey, T., 2011, “A Combustor-Representative Swirl Simulator
LE, TE ¼ leading edge, trailing edge for a Transonic Turbine Research Facility,” Proceedings of the IMechE, Part G:
J. Aerosp. Eng., 225(7), pp. 737–748.
MUSCL ¼ monoton upstream-centered scheme for conservation [25] Doorly, J. E., and Oldfield, M. L. G., 1987, “The Theory of Advanced Multi-
laws Layer Thin Film Heat Transfer Gauges,” Int. J. Heat Mass Transfer, 30(6), pp.
n1 ¼ instrumented vane 1 1159–1168.
n2 ¼ instrumented vane 2 [26] Oldfield, M. L. G., 2008, “Impulse Response Processing of Transient Heat
Transfer Gauge Signals,” ASME J. Turbomach., 130, pp. 021023-1–9.
C1 ¼ position (or vane LE) aligned with vortex center [27] Oldfield, M. L. G., Burd, H. J., and Doe, N. G., 1984, Design of Wide-
C2 ¼ position (or vane LE) aligned with center of 2 vortices Bandwidth Analogue Circuits for Heat Transfer Instrumentation in Transient
Tunnels. Heat and Mass Transfer in Rotating Machinery, (A86-24451 09-34),
Hemisphere Publishing Corp., Washington, DC, pp. 233–258.
[28] Oldfield, M. L. G., Jones, T. V., and Schultz, D. L., 1978, “On-Line Computer
References for Transient Turbine Cascade Instrumentation,” IEEE Trans. Aerosp. Electron.
[1] Gupta, A. K., Lewis, M. J., and Daurer, M., 2001, “Swirl Effects on Combus- Syst., AES-14(5), pp. 738–749.
tion Characteristics of Premixed Flames,” ASME J. Eng. Gas Turbines Power, [29] Povey, T., 2003, “On Advances in Annular Cascade Techniques,” Ph.D. thesis,
123, pp. 619–626. Department of Engineering Science, University of Oxford, Oxford, UK.
[2] Li, G., and Gutmark, E. J., 2004, “Effects of Swirler Configurations on Flow [30] Moinier, P., and Giles, M. B., 1998, “Preconditioned Euler and Navier-Stokes
Structures and Combustion Characteristics,” ASME Paper No. GT2004-53674. Calculations on Unstructured Grids,” Proceedings of the 6th ICFD Conference
[3] Huang, Y., and Yang, V., 2005, “Effects of Swirl on Combustion Dynamics in on Numerical Methods for Fluid Dynamics, Oxford, UK.
a Lean-Premixed Swirl-Stabilized Combustor,” Proc. Combust. Inst., 30, pp. [31] Moinier, P., Mueller, J.-D., and Giles, M. B., 2002, “Edge Based Multigrid and
1775–1782. Preconditioning for Hybrid Grids,” AIAA J., 40(10), pp. 1954–1960.
[4] Jouini, D. B. M., Sjolander, S. A., and Moustapha, S. H., 2001, “Aerodynamic [32] Martinelli, L., 1987, “Calculations of Viscous Flows With a Multigrid Method,”
Performance of a Transonic Turbine Cascade at Off-Design Conditions,” Ph.D. thesis, Department of Mechanical and Aerospace Engineering, Princeton
ASME J. Turbomach., 123, pp. 510–518. University, NJ.
[5] Corriveau, D., and Sjolander, S. A., 2004, “Influence of Loading Distribution [33] Mueller, J.-D., and Giles, M. B., 1998, “Edge-Based Multigrid Schemes for
on the Performance of Transonic High Pressure Turbine Blades,” ASME J. Tur- Hybrid Grids,” Proceedings of the 6th ICFD Conference on Numerical Methods
bomach., 126, pp. 288–296. for Fluid Dynamics. Oxford, UK, March 31–April 3.
[6] Corriveau, D., and Sjolander, S. A., 2007, “Influence of Loading Distribution [34] Chana, K. S., 1992, “Heat Transfer and Aerodynamics of a High Rim Speed Tur-
on the Off-Design Performance of High Pressure Turbine Blades,” ASME J. bine Nozzle Guide Vane With Profile End Walls,” ASME Paper No. 92-GT-243.
Turbomach., 129, pp. 563–571. [35] Beard, P. F., 2010, “On Transient Turbine Efficiency Measurements With
[7] Weiss, A. P., and Fottner, L., 1995, “The Influence of Load Distribution on Second- Engine Representative Inlet Flows,” Ph.D. thesis, Department of Engineering
ary Flow in Straight Turbine Cascades,” ASME J. Turbomach., 117, pp. 133–141. Science, University of Oxford, Oxford, UK.
[8] Benner, M. W., Sjolander, S. A., and Moustapha, S. H., 2004, “The Influence of [36] Harvey, N. W., and Jones, T. V., 1990, “Measurement and Calculation of End
Leading-Edge Geometry on Secondary Losses in a Turbine Cascade at the Wall Heat Transfer and Aerodynamics on a Nozzle Guide Vane in Annular
Design Incidence,” ASME J. Turbomachinery, 126, pp. 277–287. Cascade,” ASME Paper No. 90-GT-301.

Journal of Turbomachinery MARCH 2013, Vol. 135 / 021040-13

Downloaded
View publication stats 18 Jan 2013 to 163.1.165.124. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm