Content uploaded by Nick Hills
Author content
All content in this area was uploaded by Nick Hills on Mar 23, 2015
Content may be subject to copyright.
Proceedings of the 8th International Symposium
on Experimental and Computational
Aerothermodynamics of Internal Flows
Lyon, July 2007
Virginie AUTEF
http://www.lmfa.ec-lyon.fr/ISAIF8/
Paper reference : ISAIF8-008
TURBINE STATOR-WELL FLOW MODELLING
Virginie N.D. Autef1, John W. Chew1, Nicholas J. Hills1 and Ivan L. Brunton2
1University of Surrey
School of Engineering (H5)
Guildford, GU2 7XH, UK
2Rolls-Royce plc
PO Box 31, Derby DE24 8BJ, UK
In axial gas turbines, hot air from the main annulus path
tends to be ingested into the turbine disc cavities. This leads
to overheating which will reduce the disc’s life time or lead
to serious damage. Often, to overcom e this problem, some air
is extracted from the compressor to cool the rotor discs. Th is
also helps seal the rim seals and to protect the disc from the
hot annulus gas. However, this will deteriorate the overall
efficiency. A detailed knowledge of the flow interaction be-
tween the main g as path and the disc cavities is necessary in
order to optimise thermal effectiveness against overall effi-
ciency due to losses of the cooling air from the main gas path.
The aim of this study is to provide better understanding of
the flow in a turbine stator-well, and evaluate the use of diffe-
rent CFD methods for this complex, 3-dimensional unsteady
flow. This study presents CFD results for a 2-stage turbine.
The stator-well cavity for the second row of stationary vanes
is included in the calculation and results for both turbine per-
formance and stator-well sealing efficiency are presented.
Keywords: Turbomachine, axial turbine, cavity, efficiency, cooling, thermal effectiveness
Introduction
The turbomachinery industry has to face continual
pressure to improve performance. A 1% improvement in
efficiency can save millions of pounds, as typically 1%
improvement in specific fuel consumption is approxi-
mately worth 560 tonnes of fuel per aircraft per annum
for a wide bodied airliner [1]. Some flow features that
used to be ignored in approximate design calculations
now have to be considered in detail. This is particularly
the case for secondary air systems and cavity flows. With
recent computer developments, the capability of compu-
tational fluid dynamics (CFD) is increasing, allowing
more geometry features to be modelled as well as fully
3-dimensional unsteady flows. The improved representa-
tion of the physics is expected to lead to improved engine
design.
Cavity flows between a rotating disc and a fixed stator
[2-5] and more particularly the interaction between the
main annulus flow and the cavity rim seal [6-11] have
been the subject of many studies. Hot air from the annu-
lus is ingested into the cavity, and will reduce the disc
life. A common practice to reduce the amount of ingress
is to inject coolant air extracted from the compressor into
the cavity. However, as this creates parasitic losses in
engine performance, its use has to be optimised. The in-
gestion has been shown to depend on the external flow
[12]. It is strongly influenced by the swirl velocities [13],
2 Proceedings of the 8th International Symposium on Experimental and Computational Aerothermodynamics of Internal Flows
Nomenclature
C
percentage of main annulus flow rate
Greek letters
m
&
mass flow rate (kg/m3)
!
Sealing efficiency
n
local unity vector normal to the surface
!
cooling effectiveness
P
static pressure (Pa)
!
rotational speed (rad/s)
Po
total pressure (Pa)
Subscripts
ro
outer radius of the cavity (m)
1
main annulus inlet
t
time (s)
c
coolant inflow
T
static temperature (K)
in
inward mass flow through the
rim seal
T
blade passing period (s)
L
labyrinth seal flow
To
total temperature (m)
out
outward mass flow through the
rim seal
Tcoolant
relative total coolant flow temperature at
the entrance of the cavity (K)
Abbreviations
Tdisc
rotor disc temperature (K)
TSW
turbine stator well
Tgas
relative total air temperature behind the
exit of rotor 1 (K)
MP
mixing plane
u
velocity vector (m/s)
SP
sliding plane
AO
annulus only
the rotor blades [5 and 7] and the rim seal geometry [6].
The flow in the cavity may be complex,
three-dimensional and unsteady. The flow pattern is also
strongly affected by the amount of coolant flow injected
into the cavity.
Until recently, the disc cavity and main gas path flows
have generally been treated separately. In the present
contribution CFD is evaluated and used for combined
disc cavity/main gas path modelling. It is considered that
this will allow better flow modelling, with the potential
of capturing cavity/main annulus interaction effects ear-
lier in the engine design and development programme.
The scope of the present paper is to provide a better
understanding of the flow behaviour in the main annulus
flow of a 2-stage turbine and a turbine stator well (TSW),
as presented in Fig. 1. The results section presents the
main aerodynamics of the flow and considers the impact
of flow modelling on the turbine efficiency. The effect of
the geometry (cavity, labyrinth seal flow clearance) will
be evaluated. Steady and unsteady results will be com-
pared. The influence of the coolant flow on the cavity
flow pattern, the ingestion and the turbine efficiency are
also presented.
Model description
Geometry and mesh
The geometry studied is a two-stage axial-flow turbine,
based on an existing rig [14]. Each turbine stage com-
prises 39 Nozzle Guide Vanes (NGV) and 78 rotor
blades. Cooling can be supplied through 39 equispaced
cooling holes as shown in Fig. 1. The hub line is inclined
at an angle of 6°. The simulations will consider the sim-
plified case of zero blade tip clearance. The TSW is
composed of a short inner foot, two rotor-stator cavities
linked together by a labyrinth seal and to the main annu-
lus flow by two rim seals. The geometry was designed so
that the labyrinth seal clearance could be altered easily.
The geometry being 1/39th periodic, only a 1/39th model
will be represented.
Fig. 1 Model geometry.
The model and mesh, presented in Fig. 2, are the same
as used by Dixon et al. [14], and tested by those authors
for mesh dependency. Its 2 million cells provided a mesh
with y+ values within the accepted range for use with the
!
"k
turbulent model with standard wall function, ex-
cept in the labyrinth seal region. In this region, further
refinement was necessary to correctly resolve the local
velocities, resulting in y+ values below the accepted
range for this model. The Spalart-Allmaras turbulence
model was selected for this work, but no further mesh
testing was done. An annulus only model was also built
for this study to evaluate the influence of the cavity. The
Rotor 1
TSW
NGV 2
Cooling air holes
Rotor 2
Labyrinth seal
NGV 1
Virginie AUTEF et al. Turbine stator-well flow modelling 3
geometry was identical to the main model, but without
the TSW.
CFD modelling
The CFD program chosen for this work was a modi-
fied version of the Rolls-Royce time-marching code Hy-
dra [15]. Mixing plane techniques were used for steady
calculations, and sliding planes for unsteady calculations.
100 times steps per NGV passing period were specified
for the latter. To ensure convergence, flow residuals were
monitored. Momentum, mass flow and enthalpy balances
were also checked. The unsteady calculations were run
long enough to stabilise flow variables at monitored
points. Using multigrid acceleration, typical convergence
times were three days on 8 parallel processors for steady
calculations and two weeks for unsteady calculations.
The design rotational speed is 10,630rpm. The boun-
dary conditions chosen are appropriate for subsonic inlet
and outlet flow conditions in the main annulus. The main
inlet total pressure was 255,000Pa and the total tempera-
ture 443K. The outlet static pressure was set to
109,785Pa. The boundary condition for the coolant inlet
hole also assumed subsonic inflow, with a total tempera-
ture of 300K. The total pressure necessary to get the de-
sired inlet mass flow was obtained by using a trial and
error method. The Spalart-Allmaras variable was set to
0.000176165m2/s at the inlets, which corresponds to a
turbulent to laminar viscosity ratio of 10. The walls were
considered adiabatic with no-slip conditions. The outer
casing walls of the rotating zones were considered statio-
nary.
Results
Uncooled TSW
Table 1 presents the main results of this subsection in
which the TSW coolant flow is zero. Five different mo-
dels were investigated: two full unsteady models, one
with normal and one with reduced labyrinth seal clea-
rance, a full steady model and two annulus only models,
steady and unsteady. Mass flows parameters
(
11 oo PTm
&
) were in good agreement with experimental
data [16]. Two ways of calculating adiabatic efficiencies
were considered. The actual work was either based on
computed enthalpy change or torque. Theoretically these
should give identical results, but some discrepancies
were observed, especially for the steady cases. This is
considered to be due to slight mass flow imbalances in
the CFD solutions.
The highest turbine efficiency is obtained with the un-
steady calculation with the reduced seal clearance. Ta-
king the seal away by simulating only the annulus blade
row reduces the efficiency according to the unsteady
model, while providing a larger seal clearance also re-
duces the efficiency. Steady and unsteady annulus-only
simulations predict similar efficiency but, when the ca-
vity is included, a higher efficiency is given by the un-
steady calculations than the steady calculations. This
suggests a significant unsteady interaction between the
main passage and the cavity.
Fig. 2 CFD geometry and mesh, close-up on the rotor blades.
Interface planes
4 Proceedings of the 8th International Symposium on Experimental and Computational Aerothermodynamics of Internal Flows
Table 1 Turbine performance.
Mass flow parameter
11 oo PTm
&
( kg.K-1.s-1. Pa-1)
Enthalpy based adia-
batic efficiency %
Torque based adiabatic
efficiency %
Unsteady
0.00033659
89.09583361
88.77957131
Unsteady, reduced lab. seal clearance
0.00033489
90.63401474
90.57233389
Steady
0.00033381
84.71223195
83.93174113
Unsteady AO
0.00032999
86.91523198
86.92508157
Steady AO
0.00033267
87.51445042
87.03225829
Experiment [16]
0.000339
Instantaneous circumferentially averaged radial pro-
files are presented in Fig. 3, on a cross-section behind the
first rotor, just upstream of the first seal, for the five cal-
culations presented in Table 1. For the full unsteady
models with regular labyrinth seal clearance, four diffe-
rent profiles at four different physical times spread in a
blade passing period are presented.
Looking at the radial flow profiles in the annulus from
the five calculations presented in Table 1, it appears that
as the flow goes through the first stator, the profiles ob-
tained are relatively consistent between the different mo-
dels. As the flow goes through the first rotor blade row,
some larger differences appear. As can be seen in Fig. 3,
the curves start to split in two different families: mixing
plane, and sliding plane calculations. The mixing plane
calculation tends to have a smaller core region and the
sliding planes calculations have much stronger near-wall
gradients. Time variations are small in comparison to
differences between the steady and unsteady models.
There is no obvious evidence of the seal affecting the
upstream flow, whereas there is a very clear impact on
the hub boundary layer just downstream of the first seal.
As shown in Fig. 4, two new families can be noticed in
this near-wall region: calculations with or without a ca-
vity. Similar effects occur around the downstream seal.
In the second stage, the mixing plane/sliding plane fa-
mily effect is accentuated, and the scatter between the
curves increases as the flow is going down the annulus.
The highest amplitude difference is to be found around
the annulus mid-radius. Interestingly, the results indicate
strong sensitivity to whether or not the cavity is included
in the model.
Fig. 3 Instantaneous radial profiles of absolute tangential
velocities at the exit of rotor 1 and just before the upstream rim
seal.
Fig. 4 Instantaneous radial profiles of absolute whirl angle
just after the upstream rim seal.
Virginie AUTEF et al. Turbine stator-well flow modelling 5
Coolant flow simulations
Total pressures varying for different flow rates and a
fixed total temperature of 300K were specified at the
inlet of the coolant holes. The selection of flow rates co-
vers configurations from strong ingestion to completely
cooled discs.
Main flow features
As shown in Fig. 5, the amount of coolant flow sup-
plied affects the labyrinth seal flow. The smallest seal
flow rate (0.78% of the main annulus mass flow) is ob-
tained with the reduced seal clearance model, as expected.
The seal flow increases with the coolant flow rate, ran-
ging from 1.32% to 1.80% for the mixing plane calcula-
tions, and 1.66% to 2.12% for the sliding plane calcula-
tions over the range of coolant flow rates simulated. At
identical coolant flow rates, the sliding plane calculations
predict around 0.35% more leakage than the mixing
plane calculations.
Fig. 5 Effect of the coolant flow on the labyrinth seal flow.
The flow pattern in the upstream cavity showed high
sensitivity to cooling, and as the cooling air is being sup-
plied by cooling holes, it will naturally create some
strong asymmetries in the cavity. Fig. 6 shows the cavity
flow pattern on a cut plane going through the cooling
hole. The configuration shown on the left corresponds to
a very strong ingestion case. The coolant flow being too
low to satisfy the labyrinth seal flow demand, the com-
plement is sucked into the upstream cavity from the main
gas path. There is a very small recirculation near the up-
stream seal, and a main recirculating core region cover-
ing the entire cavity. The disc entrainment creates a
pumping effect, with strong outward radial velocity on
the back of the disc. The coolant air having a relatively
low radial velocity is quickly deflected and directly in-
gested through the labyrinth seal. The rotor disc remains
totally uncooled, which is consistent with rotor disc adia-
batic temperature.
As the coolant mass flow increases, this supplies the
labyrinth seal, resulting in less annulus flow being in-
gested through the upstream rim seal into the cavity. This
corresponds to the configuration shown on the right in
Fig. 6. It is to be noticed that as the coolant mass flow
increases, the labyrinth seal flow increases slightly. As
the radial velocities of the coolant flow increases, the
coolant flow penetrates further outward in the cavity and
impinges upon the stator foot, reducing the size of the
main re-circulation zone and suppressing the small
near-seal recirculation as the flow in the cavity becomes
stronger. The lower part of the cavity starts benefiting
from the cooling flow, the disc temperature dropping
slightly at low radius compared to the uncooled model.
Fig. 6 Flow pattern in the upstream cavity, steady calcula-
tion, Cc = 0.43% (left) and Cc = 1.13% (right).
By further increasing the coolant flow rate, a distinct
recirculation zone appears. The flow recirculates around
the lockplate shoulder of the disc before being ingested
through the labyrinth seal. The size of this zone increases
with the coolant flow rate. Once the coolant flow rate is
larger than the labyrinth flow seal requirement, and the
radial velocity is strong enough, the flow penetrates up to
the centre of the cavity and part of the coolant flow is
sucked into the rotor disc boundary layer before being
ejected out of the cavity to the main annulus flow.
The flow in the downstream cavity is presented in Fig.
7. Whatever the coolant flow rate supplied, the vortex
pattern stays similar, only the strength of the vortex
changes slightly. The flow exiting of the labyrinth seal
arrives with strong axial and tangential velocities; the
swirl coming from the upstream cavity and the moment
exerted by the rotating fins in the labyrinth seal. The
fluid flows onto the upstream face of the downstream
6 Proceedings of the 8th International Symposium on Experimental and Computational Aerothermodynamics of Internal Flows
disc, and is entrained into its boundary layer, the axial
velocity being transformed into outward radial velocity.
Arriving at the outer radius of the cavity, some of the
flow turns back into the main annulus flow. This strong
flow stream leaves a large but weak vortex filling most of
the cavity. A small vortex can also be found behind the
stator foot.
Fig. 7 Flow in the downstream cavity, steady calculation,
Cc = 0.425%.
Unsteady effects
The main flow patterns in the unsteady calculations
are identical to the steady flow pattern exposed above.
However, at the same coolant flow rate, the degree of
coolant jet penetration and overall level of swirl in the
upstream cavity could vary significantly between steady
and unsteady solutions. The most noticeable unsteadiness
affecting the stator well flow was found in the rim seal
where asymmetric pressure and velocity profiles (Fig. 8)
linked to the NGV position were identified (fixed posi-
tion in the NGV frame). The high pressure region in Fig.
8 matches the trailing edge of the NGV. No such asym-
metry could be transferred through the interface planes in
the steady calculations, as data are circumferentially av-
eraged through the mixing planes.
Some very strong unsteadiness can also be seen be-
hind the rotor blades. Successive wakes travelling down-
stream as the rotor blades pass the NGVs can clearly be
seen behind the second set of rotor blades (secondary
flows). Combined unsteady effects of the blades and
NGVs affect the seal flow. This is confirmed by Fourier
transforms of static pressures at three different monito-
ring points. The first point, situated near the upstream
rim seal in the rotating zone, shows strongest unsteadi-
ness linked to the NGV frequency (half of the blade fre-
quency). The second point is situated in the downstream
rim seal. Unsteadiness is again strongest at the NGV fre-
quency. The third point is in the cavity, at mid-radius,
close to the rotor wall. There is very little unsteadiness in
the cavity, and the strongest amplitude frequency (in the
stationary frame) is at the blade passing frequency for the
lower flow rates. For the highest simulated coolant flow
rate of Cc = 2.12%, there seems to be some natural un-
steadiness of frequency around 0.2 to 0.3 of blade pas-
sing frequency. This may be linked to unsteadiness in the
rim seal. As identified in earlier studies [7, 8], rim seal
flows may well be inherently unsteady for low net rim
seal throughflow rates, as occurs at this coolant flow rate.
Fig. 8 Instantaneous static pressure (top) and axial velocity
(bottom) in the upstream rim seal, Cc=2.15%.
Cooling effectiveness
The cooling effectiveness
!
is presented in Fig. 9
against the cooling flow presented as percentage of the
main inlet annulus flow Cc. These parameters are defined
as:
coolantgas
discgas
TT
TT
!
!
="
(1)
and
1
.100
m
m
Cc
c&
&
=
(2)
with Tgas the relative total air temperature behind the exit
of rotor 1, Tco olant the relative total coolant flow tempera-
ture at the entrance of the cavity and Tdi sc the rotor adia-
Flow direction
Virginie AUTEF et al. Turbine stator-well flow modelling 7
batic disc temperature. ro in Fig. 9 is the outer radius of
the cavity. The cooling effectiveness does not seem to be
strongly affected by unsteadiness, though steady results
tend to predict slightly higher effectiveness (better disc
cooling) than unsteady ones in the upstream cavity. Since
there is always some coolant flow ingested through the
labyrinth seal, the downstream disc benefits even from
low coolant flow, (high gradient at low Cc), which ex-
plains the different trends between the upstream and
downstream cavities.
Fig. 9 Cooling effectiveness against cooling mass flow on
the upstream and downstream rotor discs.
Mainstream gas ingestion
Flow can be ingested from the main annulus into both
the upstream and downstream cavities, but the most criti-
cal rim seal in this configuration is the upstream one. As
work is done in the main annulus, the temperature drops
considerably, and by the time the flow reaches the second
seal, its temperature is already considerably lower. More-
over, as mentioned previously, the downstream cavity
will benefit even from small coolant flow rates, leading
to overall lower temperature on the upstream disc of the
downstream cavity. This ingestion study will concentrate
on the upstream cavity.
Following other workers [17] and as explained belo w,
an estimate of the sealing efficiency can be obtained
based on the mass flow through the rim seal. Considering
the upstream cavity, as shown in Fig. 10, this method has
however two main restrictions:
(i) The splitting of outflow between labyrinth seal
flow and upstream seal is not considered. This is
quite limiting in this particular case, especially
since the steady cases predict a lower labyrinth
flow than the unsteady cases. Moreover, the laby-
rinth seal flow tends to slightly increase as the
coolant mass flow increases.
(ii) The flow ingested in the cavity, through the rim
seal, is assumed to come from the main annulus
but in fact may contain coolant flow being
re-ingested.
Fig. 10 Sketch of the upstream cavity.
The designations
c
m
&
and
L
m
&
correspond to coo-
lant and labyrinth seal mass flows, and the inward and
outward mass flows through the rim seal are defined as in
[16] by:
( )
!"=S
in dSuum ....
2
1nn
#
&
(3)
and
( )
!+= S
out dSuum ....
2
1nn
"
&
(4)
where S is a cutting surface through the rim seal, and
n
the local unity vector normal to the surface (outward di-
rection). Assuming that
c
m
&
is only composed of flow
coming from the coolant inlet, and
in
m
&
only of flow
coming from the main annulus inlet, the average concen-
tration of coolant flow in the upper section of the cavity
can be estimated by:
( ) in
c
c
T
inc
T
c
mm
m
dtmm
dtm
&&
&
&&
&
+
=
+
=
!
!
2
0
2
0
.
.
"
(5)
Fig. 11 Evaluation of ingestion in the upstream cavity with
the mass flow ratio through the rim seal
!
.
8 Proceedings of the 8th International Symposium on Experimental and Computational Aerothermodynamics of Internal Flows
Fig. 11 presents a comparison of the mass flow ratio
!
(sealing effectiveness) between the MP and SP cases.
MP calculations predict lower ingestion than the SP cal-
culations for a given coolant flow rate. This is consistent
with what has been seen in Fig. 5. Since the labyrinth
seal takes higher mass flow in the SP calculations than in
the MP ones, it has to ingest more flow from the annulus
(through the upstream seal) to complement the flow al-
ready coming from the cooling hole.
Turbine performance effects
Annulus mass flows are plotted against coolant mass
flows in Fig. 12. Looking at the calculations for the annu-
lus inlet flow only, it can be observed than the main flow
rate does not stay constant but decreases slightly as the
coolant mass flow rates increases, in both steady and
unsteady calculations. Unsteady calculations predict
overall higher annulus flow rates than steady calculations,
but follow similar trends. Modelling the cavity seems to
increase the main annulus flow rate, both the steady and
unsteady calculations with the cavity predicting higher
flow rates than the annulus only models.
Fig. 12 Mass flow parameter
oo PTm
&
.
Turbine efficiencies are presented in Fig. 13. These
are isentropic efficiencies taking account of both the
main annulus and coolant inlet flows. The enthalpy based
method seems to always predict around 0.75% higher
efficiencies than the torque based method for steady cal-
culations with coolant flo w. Penalties in efficiency when
introducing coolant flow rates are relatively low at low to
medium flow rates, but then get worse with higher cool-
ant rates. A coolant flow rate of around 1.5% would pos-
sibly appear as a good compromise between efficiently
cooling the rotor discs, and more especially the upstream
disc as it requires higher coolant flow rates to cool, and
reducing turbine efficiency.
The highest efficiency is obtained from the model
with the reduced seal clearance, which also corresponds
to the lowest labyrinth seal flow. Unsteady calculations
with regular seal clearance always predict higher effi-
ciency than steady calculations by approximately 5%.
The efficiency of unsteady calculations also decreases as
the coolant flow increases. However, there is very little
effect at small flow rates. Agreement between the two
efficiency methods is very good in all the unsteady cal-
culations, and as a general rule the level of convergence
obtained on momentum and mass balances on the cavity
and global model are also better than the steady calcula-
tions. The large difference in efficiency between steady
and unsteady calculations is not fully understood. It may
be associated with some flow separation in the steady
calculations.
Fig. 13 Influence of coolant mass flow on efficiency.
Conclusion
Using a model of the whole turbine including the sta-
tor well and a model of the annulus only to evaluate the
influence of the cavity, flow predictions were success-
fully obtained from the CFD code Hydra using the
Spalart-Allmaras turbulence model. Mass flow and adia-
batic turbine efficiencies were obtained for both steady
and unsteady simulations. Annulus only simulations pre-
dicted lower efficiencies than the full model, and showed
little difference in efficiency between steady and un-
steady models. Reducing the labyrinth seal clearance
increased the efficiency. The unsteady model of the full
geometry predicted higher efficiency and showed un-
steady interaction between the main annulus flow and the
cavity. Such unsteadiness was particularly noticeable in
the upstream seal.
A coolant flow was then introduced via coolant holes
to cool the rotor discs. The influence of this coolant flow
on the cavity flow and main annulus flow efficiency was
studied. The flow pattern in the upstream cavity was af-
fected by the amount of coolant flow supplied through
the cooling hole. At low flow rates, all the cooling flow
Virginie AUTEF et al. Turbine stator-well flow modelling 9
was directly sucked into the downstream cavity, through
the labyrinth seal. As coolant flow rate increased, the
flow requirements of the labyrinth seal flow were satis-
fied, allowing some coolant flow to penetrate into the
upstream cavity and cool the upstream rotor disc. Dis-
tinct recirculation zones could be seen in both cavities.
The downstream cavity flow pattern was not influenced
much by the variation in cooling flow, as the variation of
flow rate through the labyrinth seal was quite small. In-
creasing the coolant flow rate increased the cooling of
both upstream and downstream discs; the downstream
disc benefiting most at the lower cooling flow rate. How-
ever, increasing cooling efficiency adversely affects the
overall turbine efficiency and the balance between those
two has to be studied carefully, in connection with the
resulting disc temperatures. It may be concluded that the
calculated turbine performance is sensitive to both the
modelling assumptions and the cooling flow. The trends
predicted by steady and unsteady models were generally
consistent.
Two different methods of calculating the turbine effi-
ciency were used: an enthalpy based method and a torque
based method. The enthalpy based method always
seemed to predict higher efficiencies than the torque
based method. Consistency between those two methods
was very good for the unsteady calculations, but dis-
crepancies were noticed for steady calculations which
might be explained by small mass flow imbalances.
Results presented in this project were found to be in
relatively good agreement with the few experimental data
matching the running conditions chosen here which were
available at the time of execution of this project. This
study contributes to the understanding of cavity flows,
and their interaction with the main annulus. It also gives
insight into the use of cooling flow and its effect on effi-
ciency.
Acknowledgements
This work was funded by Rolls-Royce plc and EPSRC.
The authors are grateful to colleagues at Rolls-Royce for
their technical support and permission to publish this
paper, and to Susanne Svensdotter for her useful com-
ments on a draft of this paper. The authors would also
like to thanks Peter Childs, Vassilis Stefanis and Sussex
University for their experimental data.
References
[1] Young, C.: Rolls-Royce, Private communication, (2006).
[2] Staub, F. W.: Rotor cavity flow and heat transfer with in-
let swirl and radial outflow of cooling air, 92-GT-378,
Proceedings of ASME International Gas turbine and Ae-
roengine Congress and Exposition, Cologne, Germany,
(1992).
[3] Djaoui, M., Dyment, A., Debuchy, R.: Heat transfer in a
rotor-stator system with a radial inflow, Eu r. J. B-Fluids,
vol. 20, pp. 371-398, (2001).
[4] Beretta, G.P., Malfa, E.: Flow and heat transfer in cavi-
ties between rotor and stator discs, Int. J. of Heat and
Mass transfer, vol. 46, Issue 15, pp. 2715-2726, (2003).
[5] Bohn, D.E., Decker, A., Ma, H. and Wolff, M.: Influence
of sealing air mass flow on the velocity distribution in and
inside the rim seal of the upstream cavity of a 1.5-stage
turbine, GT2003-38459, Proceedings of ASME Turbo
Expo 2003, Atlanta, Georgia, USA, (2003).
[6] Bohn, E.D., Decker, A., Ohlendorf, N. and Jakoby, R.:
Influence of an axial and radial rim seal geometry on hot
gas ingestion into the upstream cavity of a 1.5-stage tur-
bine, GT2006-90453, Proceedings of ASME Turbo Expo
2006, Barcelona, Spain, (2006).
[7] Boudet, J., Autef, V.N.D., Chew, J.W., Hills, N.J. and
Gentilhomme, O.: Numerical simulation of rim seal flows
in axial turbines, The Aeronautical Journal, pp. 373-383,
(2005).
[8] Boudet, J., Hills, N.J. and Chew, J.W., Numerical simula-
tion of the flow interaction between turbine main annulus
and disc cavities, GT2006-90307, Proceedings of the
ASME turbo expo 2006, Barcelona, Spain, (2006).
[9] Cao, C. and Chew, J. W., Millington, P. R., Hogg, S. I.:
Interaction of rim seal and annulus flows in an axial flow
turbine, Transaction of the ASME, J Engineering for Gas
Turbine and Powers, Vol. 126, pp 786-793, (2004).
[10] Scanlon, T., Wilkes, J.: A simple method for estimating
ingestion of annulus gas into a turbine rotor stator cavity
in the presence of external pressure variations,
GT2004-53097, Proceedings of ASME Turbo Expo 2004,
Vienna, Austria, (2004).
[11] Jakoby, R., De Vito, L ., Larson, J., Lindblad, K., Zierer,
T., Bohn, D., Decker, A., Funcke, J.: Numerical simula-
tion of the unsteady flow field in an axial gas turbine rim
seal configuration, GT2004-53829, proceeding of the
ASME turbo expo 2004, Vienna, Austria, (2004).
[12] Phadke, U.P., and Owen, J.M.: Aerodynamic aspect of
the rim sealing of gas rotor-stator systems parts 1-3. Int
J.Heat and Fluid Flow, vol.9, pp.98-117, (1988).
[13] Hills, N.J., Chew, J.W., Green, T., and Turner, A.B.:
Aerodynamics of turbine rim seal ingestion, 97-GT-268,
Proceedings of ASME Turbo Expo 1997, Orlando, USA,
(1997).
10 Proceedings of the 8th International Symposium on Experimental and Computational Aerothermodynamics of Internal Flows
[14] Dixon, J. A., Brunton, I. L., Scanlon, T. J., Wo-
jciechowski, G., Stefanis, V. and Childs, P. R. N.: Turbine
stator well heat transfer and cooling flow optimization,
GT2006-90306, proceeding of the ASME Turbo Expo
2006, Barcelona, Spain, (2006).
[15] Hills, N. J.: Achieving high parallel performance for an
unstructured turbomachinery code, The Aeronautical Jour-
nal, to be published, (2007).
[16] Georgakis, C., Whitney, C., and Woollatt, G., Turbine
stator well CFD studies: Effect of upstream egress inges-
tion, GT2007-27406, Draft for the ASME Turbo expo
2007, Montréal, Canada, (2007).
[17] Boudet, J.: Numerical simulation of rim seal flows in ax-
ial turbines, PUMA DARP project, deliverable D2-24C,
University of Surrey internal report, TFSUTC/2005/05,
(2005).
