Paper ID: ETC2017-128 Proceedings of 12th European Conference on Turbomachinery Fluid dynamics & Thermodynamics

ETC12, April 3-7, 2017; Stockholm, Sweden

PREDICTION AND AUGMENTATION OF NOZZLE GUIDE VANE

FILM COOLING HOLE PRESSURE MARGIN

N. E. Holgate1, I. Cresci1, P. T. Ireland1, A. Rawlinson2

1
Department of Engineering Science, Oxford University, Oxford, United Kingdom
2
Turbine Systems, Rolls-Royce plc, Derby, United Kingdom

(nicholas.holgate/peter.ireland/irene.cresci)@eng.ox.ac.uk, anthony.rawlinson@rolls-royce.com

ABSTRACT

To ensure adequate cooling and avoidance of hot gas ingestion, a high pressure turbine nozzle
guide vane must maintain a safe pressure margin by which the film coolant static feed pressure
exceeds the hot gas total pressure. This pressure margin is lowest for cooling holes near the
stagnation region, especially near the coolant inlet. This study investigates an insert device
which increases the pressure margin in these ingestion risk regions by altering the coolant
passage geometry, potentially allowing engine performance gains via reduced combustor
pressure loss requirements. Seven parametrically varied inserts are compared within a
pressure-tapped, cooled vane model, and design rules are suggested. For correctly designed
inserts, pressure margin results show significant improvement in the ingestion risk region,
without causing ingestion risks elsewhere. An analytical model of combined converging,
diverging and prismatic coolant channels is validated by experimental data to be capable of
accurately predicting coolant pressure margins.

KEYWORDS

FILM COOLING, PRESSURE MARGIN, HOT GAS INGESTION

NOMENCLATURE
Combustor Turbine Interactions / , / Coolant total/static pressure, temperature
Mass Flow Rate Hydraulic diameter
Nozzle Guide Vane , Coolant density, specific gas constant
Normalised hole Pressure Margin Mainstream NGV inlet, throat velocity
, Pressure Side and Suction Side , Coolant velocity
, Channel and passage total ,/ Internal/external pressur
Momentum flux ratio Discharge coefficient
Covered and total cooling holes NGV coolant percentage of throat
, Coolant passage and insert lengths Fractional spanwise distance
Channel and passage areas Loss coefficient, pipe friction factor
Normalised internal pressure Axial distance, axial chord

OPEN ACCESS 1 Copyright © by the Authors

Downloaded from www.euroturbo.eu
Figure 1: Experimental wind tunnel showing bellmouth inlet, combustor simulator, test
section, tailboards and central vane with cooling hole locations and an insert at coolant inlet

INTRODUCTION
Film cooling of high pressure turbine nozzle guide vanes (NGVs) is essential to maintaining their
structural integrity at the extreme combustor exit temperatures of any modern jet engine gas turbine.
The effusion of film coolant without potentially destructive ingestion of hot mainstream gases
requires that the feed coolant total pressure be adequately in excess of the NGV external static
pressure at all engine conditions. This excess is commonly known as the ‘pressure margin’, and has
significant influence on engine performance and efficiency, via its consequences for parasitic
combustor pressure loss and NGV coolant mass flow rate ( ). For a given NGV film coolant , a
pressure margin increase permits a
reduction in combustor pressure loss,
which has historically been limited by
NGV pressure margin considerations.
Using the wind tunnel shown in Fig. 1.,
the present study investigates the
performance of a device which alters
the aerodynamic geometry of the NGV
leading edge coolant feed passage so as
to increase the margin independently of
the coolant , potentially allowing a
reduction in this rate, in the combustor
pressure loss, or a combination of both.
The tested device is referred to as an
‘insert’, since it is intended to be
attached to the inside of an airfoil
coolant passage (rather than cast as part
of the airfoil), in order to improve the
pressure margin at the cooling holes
most at risk of ingestion. Such a device
could also potentially be employed to
allow reshaping of an airfoil external Figure 2: NGV leading edge, with and without a half-
profile, where this may otherwise have span insert, with internal pressure contours from CFD
been compromised by the need to ( ). A static pressure rise inside the insert is
contour the coolant passage while evident. Pressures relative to the stagnation line ( =0)

2
maintaining allowable metal thickness. One of the seven insert geometries investigated in this study
is shown in the lower image of Fig. 2 in its operational position within the NGV leading edge model.
The inserts could also find application in other cooled parts. As further discussed in the theory section,
their underlying principle is to slow the coolant flow in the region of the ingestion risk hole inlets,
increasing the local recovery of dynamic pressure and hence the hole feed static pressures.
In practice, slowing coolant flow with an insert will reduce the local internal heat transfer. This
could be partially offset by increased turbulation on the coolant passage wall to augment heat transfer
coefficient. Similarly, the coolant temperature at hole exits will be slightly increased, reducing film
cooling effectiveness. This will be partially offset by higher hole pressure margins than would have
been achieved without the insert. The use of one or more inserts may represent a calculated design
choice to improve engine cycle efficiency at the expense of a local increase in NGV temperature.

Literature Survey
There appears to be no previous academic literature proposing increases to airfoil pressure margins,
by geometric manipulations or otherwise. Film cooling performance studies usually concern the
effects of variations in airfoils, holes, and jet or hot gas flows. While some studies have investigated
the performance and cooling jet physics of various coolant supply conditions (e.g. Gritsch et al.
(1997), Thole et al. (1997), Walters and Leylek (1997)), little consideration is given explicitly to hot
gas ingestion concerns. The studies most closely related to the present one appear to be those of
internal cooling schemes which consider the associated coolant total pressure losses. In regions of
low ingestion risk, internal losses may be used to reduce film blowing rate. Total pressure loss is
inherent to the turbulent friction required to increase solid-to-fluid internal heat transfer, and as such
the pressure loss characteristics of conventional techniques (impingement jets, rib turbulators and
pin-fins) are well studied. Some studies are particular to leading edges, such as that by Domaschke et
al. (2012), who also reviewed the prior studies of rib-turbulated cooling in prismatic passages of
triangular cross-sections. There is a recent trend in airfoil internal cooling research to combine and
integrate the various existing technologies (Ligrani, 2013). A promising example is the impingement
of coolant into wall cavities bridged by pin-fin arrays. In these ‘double-wall’ configurations, there is
little additional pressure loss concomitant with the extra heat transfer surface area provided by the
pin-fins (Bamba et al., 2008). The complex internal flow field, combined with the use of low blowing
ratio, high hole density film cooling, mimics the operation and performance of a porous material
undergoing transpiration cooling (Murray et al., 2017). The insert concept of this study is investigated
in a purely film-cooled geometry, though it has potential to be integrated with double-wall cooling
designs, which could be permitted more aggressive impingement pressure losses as a result.

PRESSURE MARGIN INSERT THEORY

Engine Coolant Mass Flow Rate Matching

The NGV coolant in this study is expressed as a percentage of overall throat flow ,a
value which will vary between engines. The percentages calculated in this study are not engine-
representative because of the absence of downstream PS film cooling holes and slots. True engine-
matching in this respect would also require testing of geometry with a more engine-realistic coolant
flow path with varying coolant passage area and curvature, and potentially a variety of internal
cooling features. The primary purpose of varying in this study is to examine its influence on
the inserts’ cooling and flow attributes.

Inviscid Analysis
The following inviscid analysis shows that a sufficiently broad insert can increase its internal static
pressure, thus increasing the pressure margin of enclosed holes. The fundamental mechanism is to
slow the coolant, increasing the local recovery of dynamic pressure. Consider an inviscid and
incompressible analysis (constant coolant total pressure and density in which a prismatic
coolant passage of constant cross-sectional area and internal length feeds cooling holes from

3
a single end. Let represent the fractional spanwise distance from the inlet ( ). If the coolant
mass flow rate at the inlet is and the coolant is expelled evenly along the span, then the passage
at a position can be approximated by , so the coolant velocity
. Then the coolant static pressure within the passage simplifies to eqn. (1).
̇
, , (1)
Consider also the two adjacent channels of
length with , formed by an insert , ( )
(2)
of the same length , resembling those ( ) ( )
pictured in Figs. 1 and 2. The within the ( )
(3)
converging channel can be approximated by
where is the number of holes covered by the insert (i.e. whose inlets are
within the converging channel). If the channel's cross-sectional area is , then the static pressures
in this channel , can be expressed as eqn. (2). Employing eqns. (1) and (2), the condition
for , , simplifies to inequality (3). At the inlet ( ), the static pressure
within the insert converging channel increases above the prismatic passage static pressure when the
proportion of the total passage inlet area occupied by the insert's cross-section exceeds the proportion
of included cooling holes. In the results section, an extension of the preceding analysis including total
pressure loss estimations is described, which enables hole pressure margins to be predicted with good
agreement with the experimental data.

Insert Design
To determine a profile of minimum allowable cross-sectional area , the inviscid analysis
above can be repeated with inclusion of a desired minimum pressure margin
, , where , is the maximum (stagnation line) external airfoil static pressure. Since the
external pressure around the airfoil varies, the pressure margins between covered hole rows will be
different even if the static pressure inside the insert is uniform, but all will be greater than .

, ,

, (4)
In this study, the was three NGV mainstream inlet dynamic pressures as measured in the
test rig (estimated as representative of the equivalent pressure margin in a modern civil jet engine).
The other parameters in eqn. (4) were measurable from rig tests on the film-cooled NGV model with
no insert attached. These rig pressure margin results demonstrated a clear need to include within the
inserts all but the two rearmost suction surface (SS) rows. There is only a slight recovery of dynamic
pressure observed as the coolant velocity decreases along the NGV span, suggesting that the ingestion
risk has little spanwise dependency. This would also likely be the case for the same coolant passage
under engine conditions, since the recoverable coolant dynamic pressure is smaller than the rig case
relative to the respective estimated mainstream NGV dynamic pressures. Therefore, inserts should
cover all holes of rows at ingestion risk. Nevertheless, since many engines feature double-end fed
leading edge coolant passages, inserts of lengths , and were tested. Tab. 1 summarises the
geometric details of the tested designs. For each of the shorter two lengths, three inserts were designed
on the basis of profile predictions according to eqn. (4), for ,
employing the respective NGV and pressure data collected during rig operation with no insert.
Since testing on these designs indicated higher enclosed hole pressure margins for higher inlet areas,
only a single full-span insert of
Table 1: Inlet areas and lengths of tested inserts
generous inlet area was tested. All
Insert A B C D E F G
profiles of the tested inserts exceed all
0.22 0.33 0.41 0.34 0.53 0.60 0.89
three of their relevant profiles
0.3 0.3 0.3 0.5 0.5 0.5 1
at all span positions. All of the inserts
4
Figure 3: Measured SS1 hole row pressure margins for , all seven tested inserts
were tested at . The risk of diverging channel flow separation increases as
the inlet area of the insert increases, or the length decreases, as these changes increase the diffusion
rate. The cooling holes just downstream of the end of an insert are at particular risk of ingestion, as
they may be located within a flow separation from off this end. Fig. 3 shows a summary of the
experimental results more thoroughly conveyed in the results section, which illustrates the beneficial
effects of well-designed inserts. Non-dimensional pressure margins across a near-stagnation line
cooling hole row are shown for all seven inserts plus the datum case of no insert. While experimental
data are only available at , it is clear that half-span inserts E and F vastly improve
pressure margin near the inlet, while full-span insert G somewhat improves margin for the entire span.

EXPERIMENTAL APPARATUS AND METHOD

Combustor Turbine Interactions Rig

The Oxford Combustor Turbine Interactions (CTI) rig features a two-passage cascade test section
with linear extrusions of a modern engine NGV midspan geometry, and an upstream combustor
simulator (Fig. 1). The rig-to-engine scale is 4.45. The mainstream flow is induced by a downstream
centrifugal fan, while the plena for combustor simulator dilution ports and endwall film cooling holes
are supplied by an additional blower, via a branching pipe system with butterfly valves permitting
individual control of the coolant mass flow rates, measured by orifice plate meters. The plenum
coolant temperatures are recorded by thermocouples, due to the coolant temperature rise and density
reduction through the blower. The core into the simulator is inferred using an upstream pitot tube,
which was referenced to the detailed NGV mainstream inlet velocity distribution measured by Cresci
et al. (2015a). The NGV throat Reynolds number based upon the chord varies between approximately
for no NGV coolant flow and 7.2 for the maximum investigated NGV coolant .
Increased NGV coolant very slightly reduces the induced mainstream . Rather than
compensating by increasing the mainstream fan speed, the vane mainstream inlet dynamic pressure
is determined separately for each run, and used to normalise the pressure results from that run. The
mainstream change is insufficient to significantly affect or coolant momentum flux ratios
( ). The latter are determined at the dilution ports ( ) and outer and inner endwall film
cooling holes ( and , respectively), and are representative of those in a modern Rich-Burn,
Quick-Mix, Lean-Burn (RQL) annular combustor. This parameter is well established as the most
crucial in determining jet-in-crossflow penetration and mixing, and hence the fluid property profiles
at the NGVs (Hatch et al., 1992, Holdeman et al., 1996, Barringer et al., 2002).

5
NGV Leading Edge Test Pieces
The central cascade vane features a removable leading edge section, allowing interchange of test
pieces suited to different measurements. These are fed coolant from a single end by attachment to the
same plenum as feeds the inner endwall film cooling. The coolant supply passage is prismatic along
the span and feeds 9 rows of cooling holes, labelled in Fig. 1. All NGV film cooling holes eject
parallel with the mainstream flow. The hole inlets and exits are uniformly positioned along the whole
span at midspan engine-representative positions. The cooling holes per row, hole diameter and hole
pitch are also engine-representative. As the main body of the central NGV is uncooled, it lacks several
downstream pressure side (PS) cooling hole rows and a trailing edge slot, which would be present in
an engine. Cresci et al. (2015a, 2015b, 2015c) provide more details about the rig and the flow
properties approaching the NGV cascade. The two pressure-tapped leading edges used were 3D
printed. One features coolant passage internal pressure tappings at fractional span positions
, while the other features internal and external surface tappings at ,
respectively. A prior study on the same cooled geometry found negligible variation between external
pressure measurements at and , so the external pressure distribution is considered
uniform along the span. The pressure margin inserts were also 3D printed, and their edges were sealed
to the inside of the coolant passage with flowable silicone, as shown in Fig. 2. Occasionally, the
coolant passage tappings were obstructed by the inserts or by excess silicone, and these were removed
from the data as necessary. Additional tappings in the downstream central vane section
permit external pressure measurements around the rest of the vane. tappings in the two half-
vanes allow the tailboards downstream of the cascade to be adjusted so that the external pressure
distributions mimic those of a periodic, annular cascade.

Experimental Uncertainty Analysis

The uncertainties in the derived quantities mentioned in Tab. 2 are calculated with those propagated
by orifice meter, pressure tapping, pitot tube and thermocouple measurements. values are
highly sensitive to adjustment of the relevant butterfly valve, with a maximum variation from the
stated target quantities across all runs of 0.09 (which is added to the uncertainty based on orifice plate
calculations). Since goes with the square of the jet mass flow rates, its adjustment was particularly
sensitive, with a maximum variation from the target values of approximately 0.5. These variations
are accounted for since all vane pressure data are normalised by simultaneously recorded NGV
mainstream dynamic pressures. Calculation of the non-dimensional hole pressure margins
requires interpolation of the tapping data to the centres of the hole inlets and exits. The hole exit
pressure uncertainty for SS5 and SS6 is considerable, as they reside in the region of greatest pressure
gradient on the NGV exterior. This additional uncertainty in is estimated as 0.1, based on the
maximum residual of a linear fit through the pressure data points in this region. The normalized
internal pressures are not subject to this additional uncertainty as they represent the pressures
measured at tappings. The main contributor to uncertainty in both and is the cumulative mass
flow uncertainty in the estimation of the normalising mainstream dynamic pressures. Their largest
percentage uncertainties apply only to low NGV coolant mass flow cases for which some internal
pressures approach external pressures.
Table 2: Estimated maximum uncertainties for the derived flow quantities of interest
Quantity Symbol Range Maximum uncertainty
NGV coolant % 4.7 - 9.9 0.26 (2.6 - 5.5%)
Momentum flux ratio 7 - 11 0.9 (8.2 - 12.9%)
Vane internal pressure -0.56 - 50.8 0.18 (0.3 - 32.1%)
Non-dimensional hole pressure margin 2.5 - 61.7 0.28 (0.5 - 11.2%)

6
EXPERIMENTAL AND COMPUTATIONAL RESULTS

External Pressure Distributions

Fig. 4 shows experimentally measured NGV external pressures for the case of no insert and for all
values. The external pressures with inserts exhibited negligible departure from these results.
Unlike the margin results, these pressures are normalised by the more usual throat dynamic pressure
. This is calculated in two different ways: by excluding the NGV coolant (empty circles)
and by including it (filled circles). The small gap where the SS of the removable leading edge NGV
piece meets that of the downstream piece causes a small separation region, around which the flow
accelerates, causing a pressure drop. For all experiments, tape was sealed over the gap, though little
improvement was yielded. For higher NGV coolant , the film thickness increases, generating a
greater effective NGV blockage, which increases the velocity of the mainstream flow, reducing the
NGV external surface absolute pressures, as indicated by the empty circles (the effect of NGV coolant
on the mainstream is very small). The effect is most profound on the SS where there are more
cooling holes, but it is also present to a minor degree on the PS, except for , where
there is no clear trend. The trend appears to reverse when normalising by including NGV
coolant, which increases with the NGV . A series of uncooled vane pressure distribution
measurements with values matching those of each cooled case could be compared to the
respective cooled distributions in order to examine the effects of film-mainstream mixing total
pressure loss on the pressure distribution. However, this is beyond the scope of the present study.

Figure 4: Experimental pressures on NGV exterior, with no insert present. Filled/empty circles
are normalised by including/excluding the NGV coolant

Internal Pressure Distributions

The measured internal pressures relative to the maximum measured external static pressure and
normalised by the mainstream inlet dynamic pressure are shown in Fig. 5. describes
positions around the cross-section of the coolant passage inner surface (PS negative, SS positive, and
=0 is the vane’s geometric stagnation line). The internal pressure tends to increase with coolant
because higher velocity coolant recovers more dynamic pressure as it slows. Considering the
experimental data with no insert, a spanwise increase in static pressure due to progressive effusion of
coolant is only just apparent for .

7
Figure 5: Experimentally measured static pressures inside the NGV coolant passage for seven
pressure margin inserts. and are the coolant passage wall regions internal
to the NGV pressure and suction sides, respectively. Note the vertical axis scale variations

Effects of Overly Broad or Narrow Inserts

The tested inserts illustrate the need to design inserts for low diffusion rate in the
diverging channel. The tappings lie just downstream of the inner ends of these inserts (many
were prone to blockage by adhesive, causing the irregular pressures just downstream of insert A).
Those just downstream of the portion of inner surface covered by insert B experience the same
uniform pressure as those farther around the passage surface. By contrast, the pressures measured just
behind insert C are substantially lower than those farther around the passage surface, and indeed much
8
lower than with no insert. This is because insert C has the widest inlet area of the three
candidates and hence the greatest diffusion rate in the diverging channel, causing a flow separation
and low pressure region near its end. The pressure inside insert A is slightly less than in the absence
of an insert, indicating that the converging passage is too narrow and accelerates the flow. The
effective cross-sectional area will always be less than the ideal inviscid one, because of boundary
layers within the converging passage. The restriction of mass flow into the converging channel
explains why insert A exhibits the highest static pressures at ; the associated increase in
to the diverging channel increases the recoverable dynamic pressure. Compared to no insert, inserts
B and C succeed in significantly increasing the pressures within the inserts, while reducing the
pressures outside at . Coolant pressures at are only very slightly reduced by the
three inserts. The inserts make this reduction significant, especially for
larger inlet areas, because slowing the coolant flow at more holes in the converging channel must
increase per hole in the diverging channel. This trend is continued by the insert G, for
holes within the diverging passage. At , insert D increases coolant pressure outside of the
insert, and affects little change inside it. At , it yields an equal but small static pressure gain
both inside and out. These changes are not useful; it is preferable to realise a significant gain within
the inserts, at the expense of the static pressures outside them (at the inlets of low ingestion risk holes).

Effects of Successfully Designed Inserts

Both inserts E and F produce much increased static pressures inside, with only slight reductions
outside at and , compared to no insert pressures. The trade-off between them is
that the even higher pressures within F cause lower static pressures outside at . The choice
between them in a double-end feed arrangement would therefore depend upon the relative static
pressure requirements at the ingestion risk film cooling holes and any (internal or film) cooling
features fed from the channels outside of the inserts. As mentioned previously, a full-span insert such
as G may be required for application of the concept in a single-end feed NGV leading edge. This
device also succeeds in increasing its internal static pressures at and . The static
pressure results for are increased inside the insert, but appear to decrease towards the
stagnation region . This indicates a leak where the insert is not fully sealed by the silicone
adhesive, which proved difficult to apply at locations deep within the passage. If the two channels
were properly isolated in this region, the pressure within the insert would further increase, while the
pressure outside it would further decrease. Moreover, only a single profile has been tested.
Therefore, the results show that the insert concept is viable for both single- and double-end fed NGVs.

Cooling Hole Pressure Margins

The internal and external pressure data at each of the three measurement spanwise positions can be
interpolated to estimate the static pressures at the centres of each hole inlet ( , ) and exit ( , ),
respectively. Fig. 6 shows the normalised differences in static pressure (margins) implied by this
process, across all hole rows. The external pressure distribution causes SS5 and SS6 margins to be
much higher than others. To facilitate comparison of insert performance for all rows, the hole pressure
margin results for each row are normalised once more by the highest margin achieved in that row by
any insert. For insert A, the margins of covered holes decrease at , in accordance with the
decreased internal pressures noted previously. Inserts B and C increase these margins at the expense
of the SS5 and SS6 margins. The effect is more significant at higher , for which insert C
almost equalises the margins of all rows. The inserts slightly reduce the downstream margins at
and , except for the separation region behind insert C, which suffers significantly
reduced margins (note the change of vertical axis scale for ). Insert D is too narrow to
produce a significant effect. Inserts E and F significantly improve the hole margins within them for
all , and at high F essentially equalises the margins at all rows. Both tend to reduce
the pressure margins at , though this is only significant for high . Insert G increases
the at-risk row margins for all and experimental . At-risk margins appear unaffected at
, but this is a relic of the leakage from this region of the diverging channel due to poor sealing.
9
 Figure 6: Pressure margins across film cooling hole rows for seven pressure margin inserts.
Note the change of vertical axis scale for s=0.30 results. Covered hole rows are shaded

Viscous Modelling for Prediction of Coolant Pressure Margin

The coolant passage internal pressures (Fig. 2) generated by ANSYS Fluent RANS CFD were
significantly overestimated, independently of mesh density and turbulence model. This is attributed
( )
to the overly large vena contra i.e. underestimated discharge coefficient = ( ) predicted within
each cooling hole. The reduced effective flow area due to larger vena contra means that an
unrealistically high driving pressure is needed in order to produce the mass flow rates equivalent to
experiments. For long cooling holes (length-to-diameter ratios greater than two) such as those of this
study, underestimation is ubiquitous with RANS. The effect has been observed in comparisons to
10
experimental data, even when individual holes are simulated with a variety of airfoil film cooling
inlet and outlet conditions and various turbulence models and mesh fidelities (Gupta et al., 2008). In
the absence of easily obtainable, reliable CFD results, the following enhanced analytical method may
complement or replace limited resolution experimental data. Total pressure losses in coolant channels
of length are estimated by pipe flow Darcy friction factors in converging sections and diffuser
loss coefficients in diverging sections. In this analysis, is derived from the definition of total
pressure and , and is the fraction of inlet dynamic pressure lost for a circular pipe
of length , according to the Darcy-Weisbach equation. is the specific gas constant for air.
is estimated according to Serghides (1984), making it in fact a function of Reynolds number,
hydraulic diameter (ultimately functions of ) and surface roughness (assumed constant).
( ) ̇ ( ) ( )
( ) ( ) ( ) ( )

Channel static pressure can be expressed with and the cumulative total pressure loss , :
,
A converging channel is treated as a series of pipes of infinitesimal length , yielding the following
simplified, viscous form of eqn. 2, which applies also to prismatic sections. is a dummy variable.
1 2
, 0 0 2
(5)
The static pressure in a diverging channel , is determined more simply, by assuming a linear
, amounting at the channel’s exit to times the dynamic pressure at its inlet. In practice,
total pressure loss can be calculated at specific spanwise intervals which divide the channels into
short, successive sections of diffuser or pipe. This avoids the need to solve exactly the integral in eqn.
5. Connected channels can be treated consecutively. For comparisons to the experiments of this study,
inlet total pressures were estimated from static pressure measurements within inserts at , the
channel velocities, and pipe flow losses. Area distributions of the channels and channel mass
flow reductions were approximated as linear. The results are quite insensitive to estimated
surface roughness ( was used). The values were estimated as 1.0 for insert F and 0.5 for
the lower diffusion rate insert G. Little basis currently exists for more accurate prediction of
performance of irregularly shaped diffusers with flow extraction. The pressure margin for a cooling
hole row with spanwise constant exit static pressure , is , . Despite
these simplifications, accurate and useful hole pressure margin predictions were made, as shown in
Fig. 7. SS1 and SS5 are typically the lowest margin hole rows inside and outside of the inserts,
respectively. The near-inlet pressure margin in the diverging channel is revealed to be significantly
reduced by the overly broad insert G. This suggests that for a full-span insert, special care must be
taken near the coolant inlet to balance the coolant pressures inside and outside of the insert.

Figure 7: Viscous model hole pressure margin predictions for inserts F and G,

11
CONCLUSIONS
A large-scale, film-cooled nozzle guide vane in engine-representative combustor flow conditions
has been used to experimentally measure the internal coolant and external surface pressures at various
coolant mass flow rates. Stagnation region holes are at considerably greater risk of hot gas ingestion
than rows far around the suction side. Seven coolant passage inserts were tested with the goal of
reducing the ingestion risk by increasing the margin by which the static pressure at the entrances of
at-risk holes exceeds that at their exits. This is achieved by slowing the flow to these holes by
enclosing them within a converging channel with cross-sectional areas which are large relative to the
number of holes supplied by the channel. The pressure margins for the unaltered prismatic leading
edge coolant passage are compared to those in the presence of the inserts, in order to evaluate the
efficacy of the concept and to identify design rules. The main conclusions drawn are:
1. In the diverging channel formed outside of an insert, the rate of diffusion must not be so large as
to produce separations (which create localised low pressure zones where ingestion may occur)
2. Broader inserts increase/decrease the pressure margins for cooling holes inside/outside the inserts
3. Higher coolant flow rates exacerbate both of the effects in (2), resulting in more complete
equalisation of the margins of holes within the inserts and outside
4. The insert concept can improve pressure margins, while potentially also controlling blowing
ratios and/or internal cooling feature feed conditions, for single- and double-end fed airfoils
5. The data have validated an analytical method for predicting the pressure margin performance of
combined converging, diverging and prismatic cooling passages

ACKNOWLEDGEMENTS
The research was supported by Rolls-Royce plc and the UK Engineering and Physical Sciences
Research Council under the Centre for Doctoral Training in Gas Turbine Aerodynamics.
The authors are grateful to the Osney Laboratory workshop for manufacturing
and assembling the CTI rig, and to Dr. James Turner for assisting with additive manufacturing.

REFERENCES
Bamba, T., Kumagai, T., Mimura, F., Yamane, T., Fukuyama, Y., Usui, T., Yoshida, T., (2008).
Leading edge cooling performance of an integrated cooling configuration. ASME Turbo Expo 2008,
Berlin, Germany
Barringer, M. D., Richard, O. T., Walter, J. P., Stitzel, S. M., Thole, K. A. (2002). Flow field
simulations of a gas turbine combustor. J. Turbomach., vol. 124
Cresci, I., Ireland, P. T., Bacic, M., Tibbott, I., Rawlinson, A., (2015a). Realistic velocity and
turbulence intensity profiles at the combustor-turbine interaction (CTI) plane in a nozzle guide vane
test facility. Proceedings of the 11th European Turbomachinery Conference, Madrid, Spain
Cresci, I., Ireland, P. T., Bacic, M., Tibbott, I., Rawlinson, A., (2015b). Velocity and turbulence
intensity profiles downstream of a long reach endwall double row of film cooling holes in a gas turbine
combustor representative environment. ASME Turbo Expo 2015, Montreal, Canada
Cresci, I., Ireland, P. T., Bacic, M., Tibbott, I., Rawlinson, A., (2015c). Realistic velocity,
turbulence and temperature profiles at the combustor-turbine interaction plane in a rig. International
Gas Turbine Congress 2015, Tokyo, Japan
Domaschke, N., von Wolfersdorf, J., Semmler, K., (2012). Heat transfer and pressure drop
measurements in a rib roughened leading edge cooling channel. J. Turbomach., vol. 134 (6)
Gritsch, M., Schulz, A., Wittig, S., (1998). Adiabatic wall effectiveness measurements of film-
cooling holes with expanded exits. J. Turbomach, vol. 120 (3), pp. 549-556

12
 Gupta, A. K., Ramerth, D., Ramachandran, D., (2008). Validation of CFD predicted discharge
coefficients for thick plate orifices with approach flow perpendicular and inclined to the orifice axis.
ASME Turbo Expo 2008, Berlin, Germany

Hatch, M., Sowa, W., Samuelsen, G., Holdeman, J., (1992). Jet mixing into a heated cross flow in
a cylindrical duct: influence of geometry and flow variations. NASA Technical Memorandum 105390
Holdeman, J. D., Liscinsky, D. S., Samuelsen, G. S., Oechsle, V. L., Smith, C. E., (1996). Mixing
of multiple jets with a confined subsonic crossflow in a cylindrical duct. ASME International Gas
Turbine and Aeroengine Congress and Exhibition 1996, Birmingham, UK
Ligrani, P., (2013). Heat transfer augmentation technologies for internal cooling of turbine
components of gas turbine engines. Int. J. Rotat. Mach., vol. 2013
Murray, A. V., Ireland, P. T., Rawlinson, A. J., (2017, in press). An integrated conjugate
computational approach for evaluating the aerothermal and thermomechanical performance of
double-wall effusion cooled systems. ASME Turbo Expo 2017, Charlotte, USA
Serghides, T. K., (1984). Estimate friction factor accurately. Chem. Eng., vol. 91 (5), pp. 63-64
Thole, K., Gritsch, M., Schulz, A., Wittig, S., (1997). Effect of a crossflow at the entrance to a
film-cooling hole. J. of Fluids Engineering, vol. 119 (3), pp. 533-540
Walters, D. K., Leylek, J. H., (1997). A detailed analysis of film cooling physics part 1:
Streamwise injection with cylindrical holes. J. Turbomach., vol. 122 (1), pp. 102-112

13