https://ntrs.nasa.gov/search.jsp?

R=20150001428 2019-05-11T03:04:17+00:00Z

NASA/CR—2015-218390

Design Concepts for Cooled Ceramic

Composite Turbine Vane
Robert J. Boyle, Ankur H. Parikh, and Vinod K. Nagpal
N&R Engineering and Management Services, Parma Heights, Ohio

February 2015
 1$6$67,3URJUDPLQ3UR¿OH

6LQFHLWVIRXQGLQJ1$6$KDVEHHQGHGLFDWHGWRWKH ‡ &21)(5(1&(38%/,&$7,21&ROOHFWHG
DGYDQFHPHQWRIDHURQDXWLFVDQGVSDFHVFLHQFH7KH SDSHUVIURPVFLHQWL¿FDQGWHFKQLFDO
1$6$6FLHQWL¿FDQG7HFKQLFDO,QIRUPDWLRQ67, FRQIHUHQFHVV\PSRVLDVHPLQDUVRURWKHU
SURJUDPSOD\VDNH\SDUWLQKHOSLQJ1$6$PDLQWDLQ PHHWLQJVVSRQVRUHGRUFRVSRQVRUHGE\1$6$
WKLVLPSRUWDQWUROH
‡ 63(&,$/38%/,&$7,216FLHQWL¿F
7KH1$6$67,3URJUDPRSHUDWHVXQGHUWKHDXVSLFHV WHFKQLFDORUKLVWRULFDOLQIRUPDWLRQIURP
RIWKH$JHQF\&KLHI,QIRUPDWLRQ2I¿FHU,WFROOHFWV 1$6$SURJUDPVSURMHFWVDQGPLVVLRQVRIWHQ
RUJDQL]HVSURYLGHVIRUDUFKLYLQJDQGGLVVHPLQDWHV FRQFHUQHGZLWKVXEMHFWVKDYLQJVXEVWDQWLDO
1$6$¶V67,7KH1$6$67,SURJUDPSURYLGHVDFFHVV SXEOLFLQWHUHVW
WRWKH1$6$$HURQDXWLFVDQG6SDFH'DWDEDVHDQG
LWVSXEOLFLQWHUIDFHWKH1$6$7HFKQLFDO5HSRUWV ‡ 7(&+1,&$/75$16/$7,21(QJOLVK
6HUYHUWKXVSURYLGLQJRQHRIWKHODUJHVWFROOHFWLRQV ODQJXDJHWUDQVODWLRQVRIIRUHLJQVFLHQWL¿FDQG
RIDHURQDXWLFDODQGVSDFHVFLHQFH67,LQWKHZRUOG WHFKQLFDOPDWHULDOSHUWLQHQWWR1$6$¶VPLVVLRQ
5HVXOWVDUHSXEOLVKHGLQERWKQRQ1$6$FKDQQHOV
DQGE\1$6$LQWKH1$6$67,5HSRUW6HULHVZKLFK 6SHFLDOL]HGVHUYLFHVDOVRLQFOXGHRUJDQL]LQJ
LQFOXGHVWKHIROORZLQJUHSRUWW\SHV DQGSXEOLVKLQJUHVHDUFKUHVXOWVGLVWULEXWLQJ
VSHFLDOL]HGUHVHDUFKDQQRXQFHPHQWVDQGIHHGV
‡ 7(&+1,&$/38%/,&$7,215HSRUWVRI SURYLGLQJLQIRUPDWLRQGHVNDQGSHUVRQDOVHDUFK
FRPSOHWHGUHVHDUFKRUDPDMRUVLJQL¿FDQWSKDVH VXSSRUWDQGHQDEOLQJGDWDH[FKDQJHVHUYLFHV
RIUHVHDUFKWKDWSUHVHQWWKHUHVXOWVRI1$6$
SURJUDPVDQGLQFOXGHH[WHQVLYHGDWDRUWKHRUHWLFDO )RUPRUHLQIRUPDWLRQDERXWWKH1$6$67,
DQDO\VLV,QFOXGHVFRPSLODWLRQVRIVLJQL¿FDQW SURJUDPVHHWKHIROORZLQJ
VFLHQWL¿FDQGWHFKQLFDOGDWDDQGLQIRUPDWLRQ
GHHPHGWREHRIFRQWLQXLQJUHIHUHQFHYDOXH ‡ $FFHVVWKH1$6$67,SURJUDPKRPHSDJHDW
1$6$FRXQWHUSDUWRISHHUUHYLHZHGIRUPDO http://www.sti.nasa.gov
SURIHVVLRQDOSDSHUVEXWKDVOHVVVWULQJHQW
OLPLWDWLRQVRQPDQXVFULSWOHQJWKDQGH[WHQWRI ‡ (PDLO\RXUTXHVWLRQWR[email protected]
JUDSKLFSUHVHQWDWLRQV
‡ 3KRQHWKH1$6$67,,QIRUPDWLRQ'HVNDW
‡ 7(&+1,&$/0(025$1'806FLHQWL¿F  
DQGWHFKQLFDO¿QGLQJVWKDWDUHSUHOLPLQDU\RU
RIVSHFLDOL]HGLQWHUHVWHJTXLFNUHOHDVH ‡ :ULWHWR
UHSRUWVZRUNLQJSDSHUVDQGELEOLRJUDSKLHVWKDW 1$6$67,,QIRUPDWLRQ'HVN
FRQWDLQPLQLPDODQQRWDWLRQ'RHVQRWFRQWDLQ 0DLO6WRS
H[WHQVLYHDQDO\VLV 1$6$/DQJOH\5HVHDUFK&HQWHU
+DPSWRQ9$
‡ &2175$&7255(32576FLHQWL¿FDQG
WHFKQLFDO¿QGLQJVE\1$6$VSRQVRUHG
FRQWUDFWRUVDQGJUDQWHHV
NASA/CR—2015-218390

Design Concepts for Cooled Ceramic

Composite Turbine Vane
Robert J. Boyle, Ankur H. Parikh, and Vinod K. Nagpal
N&R Engineering and Management Services, Parma Heights, Ohio

3UHSDUHGXQGHU&RQWUDFW11;&$&

1DWLRQDO$HURQDXWLFVDQG
6SDFH$GPLQLVWUDWLRQ

*OHQQ5HVHDUFK&HQWHU
&OHYHODQG2KLR

February 2015
 Acknowledgments

:HZRXOGOLNHWRDFNQRZOHGJHWKHKHOSIXOGLVFXVVLRQVZHKDGZLWK0LFKDHO9LQXS.LQ3RRQDQG(GZDUG=XUPHKO\RI
+RQH\ZHOO,QWHUQDWLRQDO,QFDQGZLWK0LFKDHO+DOELJ5DPDNULVKQD%KDWW-DPHV'L&DUORDQG-HUU\/DQJRI
WKH1$6$*OHQQ5HVHDUFK&HQWHU0LFKDHO+DOELJZDVWKHFRQWUDFWPRQLWRUIRUWKLVZRUN

Level of Review7KLVPDWHULDOKDVEHHQWHFKQLFDOO\UHYLHZHGE\WHFKQLFDOPDQDJHPHQW

$YDLODEOHIURP
1$6$67,,QIRUPDWLRQ'HVN 1DWLRQDO7HFKQLFDO,QIRUPDWLRQ6HUYLFH
0DLO6WRS 6KDZQHH5RDG
1$6$/DQJOH\5HVHDUFK&HQWHU $OH[DQGULD9$
+DPSWRQ9$

$YDLODEOHHOHFWURQLFDOO\DWKWWSZZZVWLQDVDJRY
 Table of Contents
Page
1.0 Summary 1
2.0 Acknowledgements 4
3.0 Introduction 5

4.0 Benefits from CMC vanes and blades 8

4.1 SFC improvement from CMC vanes 8
4.2 NOx reduction from CMC vanes 13
4.3 SFC improvement from CMC vanes and blades 16
4.4 Conclusions 18

5.0 Stress analysis of CMC vane 19

5.1 No Trailing Edge Ejection Comparisons 27
5.1.1 Effect of Rib 28
5.1.2 Effect of Non-Uniform Temperature Gradient 31
5.1.3 Effect of Differential Wall Thickness 33
5.1.4 Effect of Combined Pressure and Temperature Loads 35
5.1.5 Effect of Insulating Connecting Rib 36
5.1.6 Effect of Modifying Rib Shape 37
5.1.7 Effect of No Internal Pressure 38
5.1.8 Effect of Revised Temperature Distribution 40
5.1.9 Effect of Trailing Edge Shape 42

5.2 Trailing Edge Ejection 43

5.2.1 Effect of Centerline Trailing Edge Ejection 44
5.2.2 Effect of Pressure Side Trailing Edge Ejection 47
5.2.3 Effect of Modified Rib Shape 49
5.2.4 Effect of Single Impingement Cavity 50
5.2.5 Effect of Rectangular & Square Ejection Tube Geometry 52
5.2.6 Effect of Square & Circular Tube Geometry 54
5.2.7 Effect of Revised CMC Temperature Gradients 57
5.2.8 Effect of Reduced Tube Heat Transfer Coefficents 59
5.2.9 Effect of Insulating Fork Region 60

5.3 Property and Boundary Condition Effects 61

5.3.1 Effect of Through Thickness Stiffness Modulus 62
5.3.2 Effect of Doubling Thermal Conductivity 64
5.3.3 Effect of Reducing End Plate Stiffness 66
5.3.4 Effect of Boundary Condition Variation 67

5.4 Mesh Refinement Studies 68

5.4.1 Effect of Vane & End Plate Mesh Density 69
5.4.2 Effect of Mesh Shape for Square Ejection Tube 70
5.4.3 Effect of Mesh for Circular Ejection Tube 71

5.5 Maximum Component Stresses 72

5.6 Conclusions 77

6.0 Radial Cooling 78

6.1 Half Scale EEE Vane 81
6.1.1 Radial Cooling Tubes - No Film Cooling 82
6.1.2 Radial Cooling Tubes & Film Cooling 86
6.1.3 Single Cavity & Trailing Edge Ejection 87

6.2 NASA Vane 91

6.2.1 Radial Cooling Tubes - No Film Cooling 92
6.2.2 Radial Cooling Tubes & Film Cooling 94
6.2.3 Single Cavity & Trailing Edge Ejection 95

6.3 Conclusions 96

7.0 References 97

NASA/CR—201-218390 iii
 Design Concepts for Cooled Ceramic Composite Turbine Vane
Robert J. Boyle, Ankur H. Parikh, and Vinod K. Nagpal
N&R Engineering and Management Services
Parma Heights, Ohio 44130

1.0 - Summary
The objective of this work was to develop design concepts for a cooled ceramic vane to
be used in the first stage of the High Pressure Turbine(HPT). To insure that the design
concepts were relevant to the gas turbine industry needs, Honeywell International Inc. was
subcontracted to provide technical guidance for this work. The work performed under this
contract can be divided into three broad categories. The first was an analysis of the cycle
benefits arising from the higher temperature capability of Ceramic Matrix Composite(CMC)
compared with conventional metallic vane materials. The second category was a series of
structural analyses for variations in the internal configuration of first stage vane for the High
Pressure Turbine(HPT) of a CF6 class commercial airline engine. The third category was
analysis for a radial cooled turbine vanes for use in turboshaft engine applications. The size,
shape and internal configuration of the turboshaft engine vanes were selected to investigate
a cooling concept appropriate to small CMC vanes.
Cycle benefits. The benefits to the engine cycle from using CMC vanes in the HPT are
improved efficiency and reduced NOx. The higher temperature capability of CMC materials
compared to conventional metallic vane materials results in less required vane coolant when
the rotor inlet temperature is unchanged. The higher temperature capability of CMC mate-
rials provide the greatest cycle benefits when these materials are used in the first HPT stage.
In a cycle analysis most vane cooling air is non-chargeable air, in that this air is available to
do work in the first stage rotor. Even if all vane cooling air is considered as non-chargeable
air, reduced vane cooling improves vane efficiency. With the same cooling approach for
CMC and metallic vanes the calculated reduction in required cooling was nearly 5.5% of the
compressor discharge flow rate over a wide range of conditions. The 5.5% reduction in com-
pressor discharge air represents approximately a one-third to one-half reduction in required
vane coolant. The calculated improvement in cycle efficiency due to reduced first stage vane
coolant was 0.4%. Cooling air introduced into the main gas flow path downstream of the
vane throat is sometimes considered chargeable air. Cooling air ejected from the vane trailing
edge falls into this category, since it is introduced into the main gas flow path downstream of
the vane throat. There is a reduction in trailing edge ejection air when CMC vanes replace
metallic vanes. If this air is considered to be chargeable air, the additional improvement
in cycle efficiency is 0.6% It is expected that CMC vanes can eliminate second stage HPT
vane coolant. When this is done the calculated additional improvement in cycle efficiency
was between 0.3 and 0.5%, depending on the compressor extraction location for second stage
cooling air. In summary, the calculated cycle efficiency, and thus specific fuel consumption,
improvement was as much as 1.5% due to replacing conventional metallic vanes with CMC
vanes in both stages of the HPT. The largest improvement in cycle efficiency occurs when
both vanes and rotors have higher temperature capability. The calculated improvement in

1$6$&5² 
cycle efficiency was between 5 and 6% when the first stage rotor sees an increased inlet tem-
perature. Raising the rotor inlet temperature while still using metallic rotor blades yields
little or no efficiency benefits. Rotor cooling air rapidly increases as rotor inlet temperature
is raised, and all rotor cooling air is chargeable air.
The production of NOx in gas turbine combustion is a major concern, and NOx produc-
tion is a non-linear function of the fuel-to-air ratio. As vane cooling decreases, and rotor
inlet temperature remains constant, the combustor outlet temperature, and thus the fuel-
to-air ratio decreases. If there was no vane coolant, the combustor outlet and rotor inlet
temperatures would be equal. If vane coolant is 10% of mainstream flow, the combustor
outlet temperature, and thus the fuel-to-air ratio, would increase by approximately 5%.
The 5% reduction in vane coolant was calculated to reduce combustor outlet temperature by
50◦ C(90◦ F ). Experimental data from different sources show that this temperature reduction
yields a NOx reduction between 15 to 40%.
Replacing first stage metallic vanes and rotor blades with CMC components in order to
raise the rotor inlet temperature was calculated to improve cycle efficiency by at least 5%.
A 5% increase in efficiency would reduce CO2 emissions by 5%. However, NOx emissions
would not be reduced, since T40 increases as T41 increases.
Full size vane Steady state aerothermal and structural analyses were done for a first
stage HPT vane, and this vane was sized to be consistent with vanes in CF6 class engines.
While alternate external vane shapes were identified as likely candidates to reduce maximum
stresses, only internal modifications to the reference vane shape were structurally analyzed.
Analyses were performed for both pressure and temperature loads. Stresses were determined
when only pressure loads were applied, and when only thermal loads were applied. Stresses
were also determined when both pressure and thermal loads were simultaneously applied.
Pressure loads were determined assuming gas and coolant total pressures of 50 atm. Thermal
loads arose from temperature gradients, which varied with the cooling scheme. Even though
the total pressure was greater than in current engines, stresses from thermal loads generally
exceeded those due to pressure loads. Stresses were calculated using the ANSYS computer
code with the linear static modeling assumption, so that creep effects were not included in
the analysis. Three stress components: (1) through thickness; (2) hoop; and (3) spanwise,
are discussed for each case. While shear stresses were calculated, they are not discussed.
The material properties used in the analysis were those of the N24A SiC/SiC CMC system
developed under the NASA’s Enabling Propulsion Materials (EPM) program. The N24A
material is formed from balanced two-dimensional woven plies. The CMC fiber orientation
was assumed such that the strength of the CMC was the same in the hoop and spanwise
directions. The material strength in the through thickness direction was less than one-
tenth of the strength in the other two directions. The through thickness direction was
perpendicular to the plane of the weave.
Perhaps the most significant conclusion from this work is that stresses which exceeded
design goals were seen for cooling configurations which resembled those used to cool cur-
rent metallic vanes. However, peak stresses were often highly localized. Significantly lower
maximum stresses were identified for configurations without trailing edge ejection, which is
a feature of current metallic vanes. However, these configurations required as much cooling
as metallic vanes, even though maximum CMC temperatures were much higher than the
maximum metal temperatures. A path was identified to reduce maximum CMC stresses,

1$6$&5² 
while maintaining the 5% reduction in required vane coolant. Stress analysis comparisons
are given for twenty five pairs of cases. The initial comparisons were done for configurations
without trailing edge ejection. For these configurations a rib connecting the suction and
pressure surfaces, as well as having a minimum pressure surface wall thickness, was bene-
ficial in terms of reduced maximum stresses. With trailing edge ejection pressure side, as
opposed to centerline, trailing edge coolant ejection resulted in lower maximum stresses. By
moving the location of the rear of the vane impingement cooling cavity forward, maximum
stresses were reduced when trailing. edge ejection was used.
The results from the structural analysis identify where future work is needed. The highly
localized regions of maximum stresses indicate that a creep analysis is likely to show reduced
maximum stresses. This is especially true for configurations with trailing edge ejection
tubes. Maximum tensile and minimum compressive stresses were often seen within the tube,
and relatively close to each other. Internal configurations that increased radii in either the
rib or fork region showed reduced maximum stress. Alternate external vane shapes which
allow for larger radii in high stress regions, and provide acceptable aerodynamics, should be
investigated.
Radial cooled vane Current cooling schemes rely on impingement and film cooling to
achieve acceptable temperatures. Radial cooling has been proposed for small CMC vanes,
such as would be found in turboshaft engine applications. With radial cooling the impinge-
ment scheme is replaced by using several cooling tubes, oriented in the spanwise or radial
direction. Without film cooling, air flowing through the radial tubes absorbs the entire heat
flux due to the gas-to-wall temperature difference. With film cooling, air is extracted from
the tubes and is used to reduce the external heat flux by insulating the wall with a film of
cooler air. For the same coolant amount, when film cooling air is extracted from the tubes
the average tube heat transfer coefficient is lower than when no film cooling is used. Two
configurations were analyzed. One was multi-tube radial cooling. The other was a single cav-
ity configuration. This single cavity configuration could be viewed as a single radial tube, or,
if there was enough space, as an impingement cooled vane. The single cavity configuration
always was coupled with trailing edge ejection tubes.
The thermal and aerodynamic performance was determined for two vane external con-
figurations. One was a half scale version of the vane used for the full size vane structural
analysis. The other was a vane shape specified by Dr. Jerry Lang at the NASA Glenn
Research Center. Both had acceptable aerodynamics in that calculated vane efficiency was
consistent with the full size vane. Neither vane could be effectively cooled using only radial
cooling. Film cooling was required to be used in conjunction with multi-tube radial cooling.
With film and multi-tube radial cooling, or single cavity and trailing edge ejection cooling,
the aggressive future engine goals of NASA’s Large Civil Tilt Rotor(LCTR) program could
be achieved. These goals are a gas temperature of 1927◦ C(3500◦F ), and a coolant tem-
perature of 594◦ C(1100◦ F ). The gas temperature includes a pattern factor to account for
temperature variations at the combustor outlet. The coolant temperature is significantly
warmer than ambient due to a takeoff pressure of 30 atm.

1$6$&5² 
 2.0 - Acknowledgments
We would like to acknowledge the helpful discussions we had with Michael Vinup, Kin
Poon, and Edward Zurmehly of Honeywell International Inc., and with Michael Halbig,
Ramakrishna Bhatt, James DiCarlo, and Jerry Lang of the NASA Glenn Research Center.
Michael Halbig was the contract monitor for this work.

1$6$&5² 
 3.0 - Introduction

Ceramic Matrix Composite(CMC) vane and rotor blades can significantly improve gas
turbine efficiency, due to their higher temperature capability compared to conventional
metallic blades. The CMC consists of Silicon Carbide(SiC) fibers in a SiC matrix. Us-
ing CMC vanes in the first stage of the High Pressure Turbine(HPT) provides the greatest
benefit in terms of reduced vane coolant. Reduced vane coolant results in improved vane
and turbine stage efficiency, which results in lower CO2 emissions. If rotor inlet temperature
is unchanged when CMC vanes are used, NOx emissions are reduced because the combus-
tor outlet temperature is lowered. However, vane pressure and thermal loads are also at a
maximum in the first stage of the HPT. CMC materials have directionally dependent prop-
erties, as well as directionally dependent strengths or load bearing capability. CMC have
significantly lower strength in the interlaminar direction, compared to their strength in the
plane of the fibers. This report gives stress analysis results which account for directionally
dependent material properties.
To illustrate the benefits and challenges associated with CMC vanes the report is divided
into three sections. The first section addresses the benefits in terms of improved efficiency
and reduced NOx emissions from using CMC vanes and blades. The second section presents
the results of stress analyses for a vane with varying internal configurations. In this section
component stresses in the through thickness, hoop, and spanwise directions are shown. This
vane analyzed is suitable for a large engine application. The third section presents the results
of a thermal analysis of cooled small size vanes suitable for a turboshaft engine application.
This section also gives details of the thermal analysis used for both the small and large engine
size vanes.
The benefits of replacing conventional metallic vanes with CMC vanes arise from the
higher temperature capability of CMC vanes. For the same rotor inlet temperature, T41 ,
CMC vanes require less coolant than conventional metallic vanes. Vane efficiency increases
as the amount of cooling air decreases. Hartsel[1] gave a correlation to estimate the im-
provement in vane efficiency due to reduced coolant. Increasing vane efficiency increases
cycle efficiency, and thus reduces CO2 emissions and fuel consumption. From a cycle stand-
point reducing cooling air ejected from the vane trailing edge may be more important than
reducing the overall amount of vane cooling air. Trailing edge ejection air is sometimes con-
sidered chargeable air. Chargeable air is unavailable to do work in the first HPT stage. NOx
emissions are reduced when the combustor outlet temperature is reduced. When T41 is held
constant, and vane cooling air is reduced, the temperature difference between T41 and the
combustor outlet temperature, T40 is reduced, since the coolant temperature, T3 is much less
than T41 A reduced temperature difference between T40 and T41 results in a lower T40 , since
T41 is unchanged with reduced coolant. Reducing T40 reduces the combustor fuel-to-air ratio,
and a small change in the fuel-to-air ratio has a disproportionately large change in NOx pro-
duction. The benefits analysis also includes the consequences of replacing both conventional
metallic vanes and rotor blades with CMC vanes and rotor blades. It will be shown that
replacing both vanes and rotor blades yields a much greater improvement in cycle efficiency,
and thus CO2 emissions, than just replacing metallic vanes with CMC vanes. The efficiency
improvement is achieved by raising the rotor inlet temperature, T41 . Since raising T41 also
raises T40 , the benefits of reduced NOx emissions are not available when CMC rotor blades

1$6$&5² 
are used.
Stress analyses were done for a vane sized for a CF6 engine size application. Component
stresses were evaluated. The component directions were the through thickness or interlam-
inar, the hoop or circumferential, and the spanwise or radial directions. The vane internal
configuration was varied, while the external geometry was held constant for the stress anal-
yses. Stress analyses were done for vanes with and without trailing edge ejection tubes.
Thermal analyses showed that trailing edge ejection was needed to achieve the reductions in
vane coolant. Without trailing edge ejection, excessive amounts of film cooling were needed
if film cooling rows were avoided for the aft portion of the suction surface.
The feasibility of manufacturing CMC vanes have been discussed in several references.
CMC vanes have been designed, fabricated, and tested, as discussed by Verrilli et al.[2],
Vedula et al.[3], Brewer et al.[4], Watanabe et al.[5], and Nakamura et al.[6]. Marshall
and Cox[7] discussed the use of ceramic textiles in gas turbine components. These references
discuss the desirability of using film cooling, since current design maximum gas temperatures
exceed the expected temperature capability of the Environmental Barrier Coating(EBC) used
to shield the CMC. The presence of film cooling rows was not included in the stress analysis.
This is not a short coming, since results showed that most of the vane can accommodate the
stress augmentation from film cooling rows without causing excessive stresses.
In the stress analysis section comparisons illustrating the effects of vane internal geometric
variations, property variations, and modeling assumptions are given. The vane external
geometry is that given by Halila et al.[8]. The effects of a rib connecting the pressure
and suction surfaces will be examined. The effects of using a thinner pressure surface wall
thickness will be examined, The effects of centerline and pressure side trailing edge ejection
will be examined. The effects of material property and boundary conditions, as well as the
effects of mesh refinement studies will be examined. Stresses due only to pressure loads, and
stresses due only to temperature gradients, as well as stresses due to combining these loads
will be given. Stresses due to individual loads are shown because, in a specific application,
loads different than those used for the analysis are likely to be applied. For example, if the
inlet pressure was half of the 50 atm. used, and the pressure distribution was similar, stresses
due to pressure loads would be nearly halved. Two stress components, through thickness and
hoop, will be discussed in detail. The strength of woven CMC materials in the interlaminar
or through thickness direction is substantially lower than the strength in the hoop or fiber
direction. For the majority of cases examined the maximum spanwise stress was less than
the maximum hoop stress. Detailed stress distributions are not given for spanwise stress, but
the maximum spanwise stress is given for each case. The chosen CMC material was N24A.
The properties of N24A were given by Mital et al.[9]. This material is a two dimensional
balanced weave. It was assumed that material properties in the spanwise or radial direction
were the same as those in the hoop direction.
Pressure loads were determined from the vane static pressure distribution. A pressure of
50 atm. was assumed for both the internal pressure, and the external total pressure. Thermal
loads were calculated assuming film cooling and impingement cooling for vane cavities. Heat
transfer in the trailing edge ejection tubes was calculated from a correlation.
The third section of this report discusses thermal analyses for two vane geometries suit-
able for the first HPT stage of a turboshaft engine. Vanes for turboshaft engines are approx-
imately half the size of the vanes for a large commercial aircraft engine. One vane geometry

1$6$&5² 
was a half scale version of the full size vane used for the stress analysis. The other vane ge-
ometry was specified by Dr. Jerry Lang of the NASA Glenn Research Center. This vane is
referred to as the NASA vane. Two cooling configurations were analyzed for each vane. The
first configuration was for radial cooling, where tubes are aligned in the spanwise direction.
Air flowing through these tubes absorbs heat from the mainstream gas. Thermal analysis
was done for radial cooling with and without film cooling. The other cooling configuration
was for a single cavity in the forward part of the vane, and trailing edge ejection tubes to
cool the aft portion of the vane. Two sets of boundary conditions were analyzed. One was
for a gas temperature of 1700◦ K(2600◦ F ) and an inlet total pressure of 17 atm. The other
was for a gas temperature of 2200◦ K(3500◦ F ) and an inlet total pressure of 30 atm. This
temperature and pressure are consistent with NASA’s goals for the Large Civil Tilt Rotor
Aircraft(LCTR) program[10].
One objective of this work was to determine if vane radial cooling was a viable cooling
approach. In the radial cooling approach tubes are oriented in the spanwise or radial direc-
tion. Internal cooling is provided by flow in the radial tubes, and no impingement cooling is
used. This cooling approach is relatively easy to implement, and small size vanes may not
have room for even a single impingement cooling insert. Since the vanes are smaller than
for the full size vane, thermal resistance is lower for HPT vanes of a turboshaft engine than
the thermal resistance of vanes used in a large commercial engine.. The analysis included
film cooling. It was found that radial cooling without film cooling was unable to keep vane
temperatures within acceptable limits. Another objective of this work was to determine if a
vane with a single cavity and having trailing edge ejection tubes could be sufficiently cooled
when the gas temperature was 2200◦ K(3500◦ F ). The procedure used to determine vane
temperatures is discussed in section 4. The same procedure was used to determine vane
temperatures for both the full size vane and for the radial cooled vane.
This work addresses the benefits and challenges associated with using CMC vanes to
replace conventional metallic vanes in the high pressure turbine. The work was undertaken
in consultation with Honeywell International Inc. personnel to insure that it is relevant to
industry needs. It is recognized that the trailing edge thickness of a CMC vane may need to
be greater than the relatively thin trailing edge of this metal vane.
The purpose of this report is to comprehensively present the results of the work done
under the SBIR contract. Previous reports[11-13] covered aspects of the work, but the results
presented here cover aspects of the work not presented in any of the references.

1$6$&5² 
 4.0 - Benefits from CMC vanes and blades

Typical CMC materials have higher temperature capabilities than conventional metallic
vanes, and this leads to less required coolant for the same gas temperature. When only
metallic vanes are replaced by CMC vanes rotor inlet temperature, T41 is unlikely to increase.
Wilcock et al.[14] showed that raising T41 for current HPT blades is unlikely to improve
thermodynamic efficiency because the additional required rotor cooling air negates the gain
in efficiency from raising T41 . Rotor cooling air is unavailable to do work in the stage for
which it is introduced. If current metallic blades are replaced by CMC blades T41 , and
thus thermodynamic efficiency, can increase, since rotor cooling need not increase as T41 is
increased. Even if only current metallic vanes are replaced with CMC vanes there is an
improvement in thermodynamic efficiency, since vane efficiency increases as vane cooling
requirements decrease. Perhaps as importantly, when only the vanes are made from CMC
materials reduced vane coolant is expected to reduce NOx production, since reduced vane
coolant reduces the temperature difference between the combustor outlet, T40 , and the rotor
inlet, T41 . If CMC vanes and rotors are used raise cycle efficiency by raising T41 , CO2
emissions are reduced, but NOx emissions are not reduced.

4.1 - SFC Improvement from CMC vanes.

Less vane coolant is required when metallic vanes are replaced by CMC vanes. Less
vane coolant improves vane aerodynamic efficiency, and stage efficiency further improves
because the amount of coolant air ejected from the vane trailing edge is also decreased.
If the trailing edge ejection air is considered to be chargeable air, a reduction in this air
increases stage output work, with no increase in fuel flow or compressor work. Figure 4.1.1
compares material properties for a selected CMC material and MA754, which is an alloy
suitable for HPT vanes(Halila et al.[8]). The MA754 properties were obtained from the
Special Metals web site[15]. The properties of the CMC material(N24A), a balanced 2D
weave, were obtained from Mital et al.[9]. As discussed by DiCarlo et al.[16] this is a fiber
reinforced SiC/SiC material with an upper use temperature of 1316◦C(2400◦ F ). While the
ratio of the Proportional Limit(PL) strength to the Ultimate Tensile Strength(UTS) is lower
for N24A than for MA754, N24A maintains its strength to a much higher temperature. This
figure shows that for N24A the UTS in the through thickness or interlaminar direction is
nearly a order of magnitude lower than the UTS in the plane of the weave.
Thermal properties and stiffness are given in figure 4.1.2 for both the CMC material and
the reference metallic material. At high temperatures N24A has lower thermal conductivity
and coefficient of thermal expansion than MA754. In the comparative thermal analysis a
thin EBC layer was applied over the CMC vane and was assumed to have a temperature ca-
pability of 1483◦ C(2700◦F ). This EBC layer had a thermal conductivity of 1W/m − K. The
EBC material was described by Lee et al.[17]. For the comparative thermal analysis a Ther-
mal Barrier Coating(TBC) having the same thermal conductivity, but a lower temperature
capability of 1316◦C(2400◦ F ). was applied to the metallic vane.
Figure 4.1.3 shows typical maximum temperatures for the CMC and the protective low
conductivity EBC layer. These temperatures are the midspan outer temperatures of each
layer. A surface distance of zero corresponding to the leading edge stagnation point. Since

1$6$&5² 
CMC materials require an EBC, comparisons were made assuming that both the metal and
CMC vanes had prime reliant 0.25mm(10mil) Thermal Barrier Coatings(TBC). A relatively
thick EBC or TBC layer, reduces the temperature gradients in the vane walls. The desir-
ability of reducing the wall temperature gradient will be shown.
The calculations shown in figure 4.1.3 were done using data from Reiss and Bolcs[18] for
the leading edge region film effectiveness and heat transfer, and the correlation developed
by Boyle and Ameri[19] for film cooling effectiveness. Local film cooling effectiveness was
calculated assuming superposition from upstream cooling rows. The outer temperatures
show a saw tooth pattern, and temperature minimums indicate the location of film cooling
rows. No suction surface film cooling rows were downstream of the vane throat, since doing
so causes excessive aerodynamic losses. The relatively thick barrier coatings resulted in the
outer coatings reaching their maximum temperatures, while most of the vane temperatures
were less than their maximum values. Using a thick barrier coating lessened the vane through
thickness temperature gradients, which lowered vane thermal stresses.
Figure 4.1.4 shows calculated vane coolant fractions, (vane coolant-to-compressor dis-
charge, (wC /w2.5 ) ratio) for both the metal and CMC vanes. Comparisons were made at
constant rotor inlet temperature, T41 . These comparisons show that the reduction in vane
coolant when using CMC vanes is approximately 5% over a wide range of rotor inlet tem-
peratures, T41 . The calculated coolant fractions included both film cooling and cooling using
trailing edge ejection. The portion of the cooling air fraction attributable to trailing edge
ejection was found using internal heat transfer correlations and appropriate pressure losses.
The approximately 5.5% reduction in required cooling air improves cycle efficiency be-
cause vane aerodynamic efficiency increases when vane cooling is reduced. Stage output
increases when vane trailing edge ejection air is reduced. Improved cycle efficiency reduces
CO2 emissions. The improvement in vane aerodynamic efficiency due to reduced coolant
was calculated using the correlation given by Hartsel[1] Calculations using this correlation
showed that a 5.5% reduction in vane coolant air increased the total pressure at the rotor
inlet by 0.86%. Even though this increase in total pressure is less than 1%, there is a re-
sulting increase in turbine output of 0.2%. This 0.2% increase in turbine output results in a
cycle efficiency increase of approximately 0.4%.
Most first stage HPT vane cooling air is non-chargeable air, and is available to do work
in the first stage rotor. However, air ejected from the trailing edge of the first HPT stage
vane may be considered as chargeable air. Even if this air is not considered to be chargeable
air the first stage performance is likely to decrease since trailing edge ejection air causes
increased aerodynamic loss. Assuming first stage vane trailing edge ejection cooling air is
fully chargeable air gives a significant improvement in SFC. Calculations showed that there
was a 5.5% reduction in required vane coolant from using CMC vanes. Trailing edge ejection
air was reduced by 1.2% of mainstream air, or nearly a quarter of the overall 5.5% reduction.
This 1.2% reduction in first stage vane chargeable air improved SFC by 0.6% for a highly
efficient two stage HPT with equal work splits, and an equal work split between the HPT and
Low Pressure Turbine. If trailing edge ejection air is not considered to be chargeable air, it
still benefits stage performance to reduce trailing edge ejection air. Trailing edge ejection air
only partially fills in the vane wake, and there is a large rotor incidence variation when the
rotor passes through the vane wake. The incidence variation decreases rotor, and therefore
stage, performance.

1$6$&5² 
 If using CMC material reduces first stage HPT coolant requirements, cooling can be
eliminated for the second stage HPT vane. This also improves SFC. In the design report
of Halila et al.[8] the second stage HPT vane cooling flow was 1.85%, of which 0.75% was
used for purge air. It is conservative to assume a 1.1% reduction in stage two vane cooling
air due to using CMC vanes. While second stage air is chargeable air, it may come from
an intermediate compressor stage, and thus is not fully chargeable air. On the other hand,
according to Honeywell International Inc. personnel, in small engine applications, it is
sometimes desirable to extract HPT cooling air for both stages from the compressor exit.
A 1.1% reduction in second stage HPT vane cooling air improves cycle efficiency by 0.55%,
when this air is compressor discharge air. It is estimated that the SFC improvement would be
about 0.3% when second stage vane cooling air is extracted from and intermediate compressor
stage. The Hartsel[1] correlation gives only a small reduction in second stage vane pressure
loss due to the elimination of cooling air.

1$6$&5² 
 MA754
100
N24A - In plane - UTS
Strength, ksi

N24A - In plane - PL

10

N24A -Through thickness - UTS

1
0 500 1000 1500 2000 2500
o
Temperature, F

Fig. 4.1.1 Strength properties of CMC material N24A and conventional metallic vane mate-
rial MA754

40

The / m/C
rma it y, W
l co
ndu du ctiv
30 ctiv
ity, l con MA754
W/m r ma CMC
/C The
Material properties

-3
Stiffness, X 10 , ksi
20
Stiffn
ess, X -3
10 , k
si
6 o -1
10 efficient, X 10 , F
Therm al expansion co
6 o -1
Thermal expansion coefficient, X 10 , F

0
0 500 1000 1500 2000 2500
o
Temperature, F

Fig. 4.1.2 Material properties for N24A and MA754

1$6$&5² 
 1600

1500 Max EBC temperature o

2700 F

1400
Max CMC temperature Max TBC temperature o
2400 F
Temperature, C

1300
o

1200
Max metal temperature 2060oF
1100

1000

900
Pressure surface Suction surface
800
-0.08 -0.06 -0.04 -0.02 0 0.02 0.04 0.06 0.08
Surface distance, m

Fig. 4.1.3 Maximum vane and EBC/TBC temperatures for CMC and metallic vanes

0.12
Vane coolant fraction, wC / w2.5

Metal vane

0.08

CMC vane
0.04

0
1400 1500 1600 1700 1800 1900 2000 2100 2200
o
Rotor inlet temperature,T41 C

Fig. 4.1.4 Coolant fractions for CMC and metallic vanes

1$6$&5² 
 4.2 - NOx reduction from CMC vanes.
Because CMC vanes use less cooling air than metal vanes for the same rotor inlet tem-
perature, T41 , the combustor outlet temperature, T40 , is lower when CMC vanes replace
conventional metallic vanes. Since most vane cooling air is non-chargeable air, the turbine
output is nearly the same when CMC vanes replace metallic vanes.
Data show that a small reduction in T40 significantly reduces NOx. A constant specific
heat thermal balance across the vane row gives:

T41 (1 + wC /w40 ) = T40 + wC TC

where wC /w40 is the vane coolant fraction, and TC is the coolant temperature. While T41 and
TC vary depending on the application, a typical difference between these two temperatures
is 1000◦ C(1800◦F ). For this temperature difference a 5% reduction in wC /w40 results in a
50◦ C(90◦ F ) reduction in T40 .
Several references showed that the amount of NOx produced is very sensitive to T40 .
Figure 4.2.1, from Tacina et al.[20], shows the amount of NOx produced per kilogram of
fuel as a function of the flame temperature. NOx is strongly dependent on the combustor
configuration, and the red line data is for the configuration with the lowest NOx at high
temperature. The peak gas temperature, TG , which corresponds to the flame temperature
is over 2000◦K(3140◦ F ) and is expected to be more than 200◦ C(360◦F ) hotter than T40 .
An extrapolation of the curve shows that NOx is reduced by nearly 15% when the flame
temperature is reduced by 50◦ C(90◦ F ).
The straight lines in figure 4.2.2, which is from Harth et al.[21] show that the relative
sensitivity of NOx to the adiabatic combustor temperature is independent of this temper-
ature. This combustor, like the combustor tested by Tacina et al.[20], is for an aircraft
engine application, At the highest combustor pressure and inlet temperature a change of
50◦ K(90◦ F ) in the adiabatic temperature reduces NOx by more than 40%.

1$6$&5² 
 x

Fig. 4.2.1 NOx as a function fo flame temperature for different combustor geometries, from
Tacina et al.[20]

Fig. 4.2.2 NOx as a function of adiabatic temperature for different combustor consitions and
fuel splits, from Harth et al.[21]

1$6$&5² 
 Fiure 4.2.3, from Dusing et al.[22], shows results similar to those in figure 4.2.2, but
for a combustor designed for ground power application. This figure has lines with straight
slopes, and shows that a 50◦ K(90◦ F ) reduction in combustor outlet temperature reduces
NOx by more than 40lines are approximately parallel for two technology levels, (2006 &
2011). Therefore, it is not likely that future combustor designs will have relative NOx sen-
sitivities significantly lower than current combustor designs. The absolute level of NOx is
expected to be lower for future combustors operating at current conditions. However, com-
bustor outlet temperatures and pressures are expected to increase over time, and increases
in either of these parameters raises NOx levels. The benefits of lower T40 from using CMC
vanes provides a means of offsetting the adverse effects of higher temperature and pressure.

Fig. 4.2.3 NOx as a function fo flame temperature for different combustor geometries, from
Dusing et al.[22]

1$6$&5² 
 0.52
o o
T41=3500 F (2200 K)
0.5
o o
T41=3320 F (2100 K)
0.48 o o
T41=3140 F (2000 K)
Cycle efficiency

o o
T41=2960 F (1900 K)
0.46
o o
T41=2780 F (1800 K)

0.44 o o
T41=2600 F (1700 K)

0.42

0.4
20 30 40 50 60 70
Overall pressure ratio

Fig. 4.3.1 Typical cycle efficiency, constant coolant fraction, from Boyle and Jones[23]

4.3 - SFC improvements from CMC vanes and blades

The maximum SFC improvement from replacing metallic airfoils with CMC airfoils is
achieved by maintaining the cooling fraction and raising the rotor inlet temperature, T41 .
Wilcock et al.[14] showed that raising T41 for conventional metallic rotor blades does not
improve SFC because of the additional cooling air required for the rotor blades. Without
penalizing the rotor for additional cooling air, raising T41 implicitly assumes both CMC
vanes and rotor blades. Figure 4.3.1, from Boyle and Jones[23], shows cycle efficiency for
a representative set of conditions, and the assumption that rotor cooling flow fractions are
independent of temperature. Higher gas temperatures are achieved by improving the tem-
perature capability of the vane and rotor materials. In reference 24, it was shown that there
is no improvement in cycle efficiency with increasing temperature for an ideal cycle, due
to the specific heat increasing with temperature. However, this is not the case when stage
efficiencies are less than one. At the lowest T41 , efficiency decreases for Overall Pressure Ra-
tio(OPR) beyond 45, but efficiency continues to increase with OPR for higher temperatures.
Analyses were also done for the increase in T41 when using CMC vanes. Maximum
surface temperatures were 2700◦ F (1756◦K) for the CMC vane and 2400◦F (1589◦ K) for the
metal vane. The vane coolant fraction was held constant at 10.6% of compressor discharge
air. Coolant air temperature was also held constant. The rotor inlet temperature for the
metallic vane was calculated to be 2935◦ F (1886◦K). Somewhat surprisingly, the increase in
T41 was greater than the 300◦ F (167◦C) increase in the maximum surface temperature. T41
increased by 525◦ F (292◦C) when the metallic vane was replaced by a CMC vane. Figure
4.3.1 shows that a 300◦ (540◦F ) increase in T41 produces a relative increase in cycle efficiency
of at least 3% at constant OPR when the OPR is between 40 and 50. The minimum relative

1$6$&5² 
increase is at high T41 values. When the increase in T41 is accompanied by an increase in
OPR, as has been the historic trend, a 300◦ increase in T41 yields even greater efficiency
gains. Going from T41 = 1900◦K(2960◦ F ) and an OPR of 40 to a T41 = 2200◦ K(3500◦F )
and an OPR of 50 results in a relative cycle efficiency gain of over 6%. It is conservative to
expect a gain of 5% in cycle efficiency due to replacing metallic vanes and rotor blades with
CMC components.

1$6$&5² 
 4.4 - Conclusions

CMC first stage HPT vanes require 5.5% less vane coolant due to the CMC vanes having
higher temperature capabilities for the same cooling configuration. Just replacing metallic
vanes with CMC vanes can improve cycle efficiency up to 1.5%. An improvement of 0.4% in
efficiency was due to increased first stage vane efficiency. Elimination of second stage vane
coolant was estimated to increase cycle efficiency between 0.3 and 0.5%. The lower increase
assumed that second stage vane cooling was extracted from an intermediate compressor stage.
The higher efficiency increase assumed that second stage vane cooling air was extracted from
the compressor exit. With CMC vanes there is a reduction in the required amount of air
ejected from the vane trailing edge. Trailing edge ejection air is sometimes considered to be
chargeable air in a cycle analysis. If the reduction in trailing edge ejection air, is treated
as chargeable air, the additional cycle efficiency improvement was estimated as 0.6% This
improvement came from replacing metallic vanes with CMC vanes.
When conventional metallic vanes are replaced with CMC vanes, and the rotor inlet
temperature, T41 remains the same, the required vane coolant is reduced by approximately
5% of mainstream air ratio. This is due to the higher temperature capability of the CMC
materials. The 5% reduction in mainstream air used for vane coolant reduces T40 by approx-
imately 50◦ K(90◦ F ). This has a disproportionately large reduction in NOx due to the high
sensitivity of NOx to combustor outlet temperature, T40 . Experimental data from different
sources show a reduction in NOx ranging from 15% to over 40% due to just a 50◦ K(90◦ F )
reduction in T40 .
Replacing both conventional metallic HPT vanes and rotor blades with CMC vanes and
rotors results in the highest gain in cycle efficiency. It was calculated that the rotor inlet
temperature could be raised nearly 300◦C(540◦ F ) when CMC vanes and rotor blades are
used. It was shown that a 300◦ C increase in T41 is expected to raise cycle efficiency by at
least 5%. A 5% efficiency increase represents a 5% decrease in CO2 emissions. However,
NOx emissions will not decrease due to using CMC vanes and blades, since raising T41 raises
T40 .

1$6$&5² 
 5.0 - Stress Analysis of CMC Vane

Steady state aerothermal and structural analyses were done to identify approaches to
achieving acceptable stress distributions for a specified vane external geometry. This section
of the report gives the effects on stress components due to changes in the vane internal
geometry, as well as other factors, such as material properties, boundary conditions, and grid
density. Figure 4.1.1 shows that allowable stresses for the CMC material are directionally
dependent. Consequently, component stresses were calculated. A somewhat arbitrary goal
of this work was to identify configurations having through thickness or interlaminar stresses
below 1.5ksi(10.5MPa), and having hoop and spanwise stresses below 17ksi(117MPa). This
goal was based on the properties of N24A material, and as shown by Engel[25] other CMC
materials have significantly higher interlaminar strengths. Detailed results are presented for
through thickness and hoop stresses. For the assumed orientation of the reference CMC
material(N24A) the allowable stresses in the spanwise or radial direction and in the hoop or
vane circumferential direction were the same. Generally, but not for all cases, the maximum
spanwise stress in the vane midspan region was lower than the maximum hoop stress in
this region. Although a detailed discussion of spanwise stresses is not given, the maximum
component stress in all three directions is discussed.
Pressure and thermal loads were determined from Navier-Stokes Computational Fluid
Dynamics analyses. Pressure and heat transfer coefficient distributions were calculated using
two and three dimensional versions of the Navier Stokes computer analysis codes described
in references 26 and 27. A single vane normalized pressure distribution was used. This dis-
tribution is for a vane exit Mach number of 0.95. Three dimensional analyses of the vane and
flow path of reference vane geometry given by Halila et al.[8] showed little spanwise variation
in either pressures or heat transfer coefficients. Consequently, midspan distributions were
used for each spanwise vane location.
Heat transfer distributions vary with vane total pressure, and were also obtained using
the Navier Stokes analysis. Boundary layer transition was accounted for in determining heat
transfer distributions, when film cooling was not used. Data show that, just the presence
of film cooling holes causes transition to fully turbulent flow immediately downstream of
the film cooling holes. Consequently, when film cooling was used, the flow was assumed to
be fully turbulent. The local heat flux is a function of both the heat transfer coefficient
and the adiabatic wall temperature. Without film cooling the adiabatic wall temperature is
nearly the gas temperature. With film cooling the adiabatic wall temperature approaches the
coolant, or compressor discharge, temperature, when the film effectiveness approaches one.
The distribution of the vane film cooling effectiveness away from the leading edge was found
using the analysis given by Boyle and Ameri[19]. In the leading edge region the film cooling
effectiveness, and the increase in the local heat transfer coefficients, were found using the
results given by Reiss and Bolcs[18] Internal cooling was either by impingement cooling or by
flow through radial cooling tubes. When impingement cooling was assumed, the correlation
of Florschuetz et al.[28] for a staggered array of impingement cooling holes was used. When
radial cooling was used, the tube heat transfer coefficient was determined by the standard
correlation for turbulent flow in a tube[29]. Tube heat transfer, which is a function of the
mass flux, was corrected to account for the change in midspan mass flux when film cooling
was used in conjunction with radial cooling.

1$6$&5² 
 Fig. 5.0.1 Vane with two endplates
Structural analysis was done using the ANSYS code[30]. Component stresses were
calculated using the ANSYS linear-static assumptions. CMC materials have higher strengths
in the fiber directions. At each point within the CMC material stresses were resolved into
three directions. These were the through thickness, hoop, and spanwise directions. The
through thickness direction was generally normal to the internal vane surface. Where the
vane had no internal surface, typically the aft portion of the vane, the through thickness
direction was normal to the vane external surface. The hoop direction was normal to the
through thickness direction. Component stresses were calculated using local coordinate
systems within the vane and the “as calculated” ANSYS option. Properties of the CMC
material, N24A, were also directionally dependent. Even though actual gas turbine vanes
are defined by cylindrical coordinates, Cartesian coordinates were used in the analysis. The
model vane, shown in figure 5.0.1, has all the necessary features of an actual vane. The vane
and end plates show a typical external temperature distribution, where the temperature is
in ◦ F . One of the end plates is fixed, while the other end plate is free to expand in all
directions.
While film cooling was assumed to determine vane temperature distributions, film cooling
holes were not included in the structural analysis. Film cooling holes have closer spacing in
the spanwise direction than in the circumferential or hoop direction. Because of this, hoop
stresses are more sensitive than spanwise stresses to stress augmentation due to an area
reduction from film cooling holes. It will be shown that peak stresses are highly localized.
For most cases there are ample regions where rows of film cooling holes could be located
without causing excessive stresses.
The reference vane is shown in figure 5.0.2 The external vane shape is that of the
vane of Halila et al.8]. This vane was designed under contract as part of NASA’s Energy
Efficient Engine(EEE) program. This figure shows a uniform wall thickness of 2mm(80 mil).
The vane has an axial chord, CX , of 3.38cm(1.33in), and a true chord nearly twice as long at
6.3cm(2.48in). Three red circles are shown in this figure. The largest circle shows the leading
edge region curvature. While the actual leading edge is elliptical in shape, the leading edge
radius is approximately 0.3cm(118mil). A smaller circle of 0.076cm(30mil) radius is shown
where the inner CMC surfaces come together to form the fork region. Where the inner
suction and pressure walls join this radius was used in the stress analysis for the uniform
wall thickness cases.. The third circle with a radius of 0.0535cm(21mil) is at the trailing
edge. The analysis was done using this radius, even though a thicker trailing edge is likely
to be needed for a CMC vane.
1$6$&5² 
 6
0.20cm(80mil)

Pitchwise distance, cm
4

Tru
ec
ho
3

rd
6
.48
2

cm
(2
.55
1

in.
)
0 CX=3.38 cm (1.33 in.)

-6 -5 -4 -3 -2 -1 0 1
Axial distance, cm

Fig. 5.0.2 Reference vane geometry

0.9
Pressure surface
Static pressure ratio, P / PT

0.8

0.7

Suction surface
0.6

0.5

0.4

0.3
0 0.2 0.4 0.6 0.8 1
Axial distance, x/CX

Fig. 5.0.3 Pressure distribution for reference vane

1$6$&5² 
 Pressure loads were evaluated using the external pressure distributions shown in figure
5.0.3. Except for one analysis case, there was an internal pressure equal to the external total
pressure. The largest pressure differential occurs near mid-chord, where the suction surface
pressure is only about half of the inlet total pressure.
Table 5.0.1 gives the list of cases that were analyzed. The case numbering, 1A through
13B, is somewhat arbitrary, but is used to identify cases for comparison purposes. Stresses
were calculated for pressure and temperature loads independently, and when pressure and
temperature loads were combined. The purpose of each case is given in the table. The
majority of cases had a rib connecting the pressure and suction surfaces of the vane, and
for most of these cases the rib was insulated. The majority of cases had a differential wall
thickness, where the the thickness of the pressure surface and in leading edge were half of
the 2mm(80 mil) thickness of the uniform wall thickness cases. Reducing the suction surface
wall thickness resulted in very high stresses just due to pressure loads. For most of the
cases a non-uniform temperature distribution was used. Nearly half of the cases had either
centerline or pressure side trailing edge ejection.
Figure 5.0.4 shows the vane configurations described in Table 5.0.1 This figure show the
vane configurations and temperatures for the CMC vane. Only very close to the trailing
edge does the vane temperature slightly exceed the desired maximum CMC temperature of
1316◦ C(2400◦F ). The region is very small, and is approximately the same in each part of the
figure. Part a shows the baseline case of single large impingement cavity. Part b shows a vane
having a differential wall thickness. The rib is uninsulated, accounting for its relatively cold
temperature. Part c shows temperatures through a slice of the vane showing the trailing
edge ejection tube. As expected, near the entrance of the tube the surface temperatures
are relatively cold. Part d shows temperatures for a single cavity configuration. The end
temperatures are shown through the solid section, rather than through the trailing edge
ejection tube. This single cavity configuration has the size cavity as the forward cavity
shown in part b of figure 5.0.4

1$6$&5² 
 Table 5.0.1 - List of Cases

Cases Load Purpose Rib Wall CMC T.E.

Present Insulated Thickness ΔT Ejection
1A Pres Base line case No N.A. Uniform N.A. No
1B Temp Uniform ΔT No N.A. Uniform Uniform No
1C Temp Non-Uniform ΔT No N.A. Uniform Non-Uniform No
1D Comb Combined loads No N.A. Uniform Non-Uniform No
2A Pres Connecting rib Yes N.A. Uniform N.A. No
2B Temp Connecting rib Yes No Uniform Non-Uniform No
2C Comb Combined loads Yes No Uniform Non-Uniform No
3A Pres Differential wall thickness Yes N.A. Differential N.A. No
3A2 Pres No internal pressure Yes N.A. Differential N.A. No
3B Temp Differential wall thickness Yes No Differential Non-Uniform No
3C Temp Insulating rib Yes Yes Differential Non-Uniform No
3D Comb Insulating rib Yes Yes Differential Non-Uniform No
4A Pres Centerline trailing edge ejection Yes N.A. Differential N.A. Yes/Cl
4B Temp Centerline trailing edge ejection Yes Yes Differential Non-Uniform Yes/Cl
4C Comb Centerline trailing edge ejection Yes Yes Differential Non-Uniform Yes/Cl
5A Comb Revised rib shape Yes Yes Differential Non-Uniform Yes/Cl
5B Comb Reduced coolant Yes Yes Differential Non-Uniform Yes/Cl
5C Comb Presure side trailing edge ejection Yes Yes Differential Non-Uniform Yes/P.S.
5D Pres Revised rib shape Yes N.A. Differential N.A. No
5E Pres Single impingement cavity No N.A. Differential N.A. Yes/P.S.
5F Pres Pressure side trailing edge ejection Yes N.A. Differential N.A. Yes/P.S.
5G Comb Single impingement cavity No N.A. Differential Non-Uniform Yes/P.S.
6A Comb Modulus(E33) change Yes Yes Differential Non-Uniform No
6B Comb Modulus(E33) change Yes Yes. Differential Non-Uniform No
6C Comb Revised temperatures Yes Yes Differential Non-Uniform No
6D Comb Modulus(E33) change Yes Yes Differential Non-Uniform No
6E Comb Modulus(E33) change Yes Yes Differential Non-Uniform Yes/Cl
6F Temp Revised temperatures Yes Yes Differential Non-Uniform No
6G Comb Combined loads Yes Yes Differential Non-Uniform No
7A Comb End plate thickness Yes Yes Differential Non-Uniform No
7B Comb Mesh refinement Yes Yes Differential Non-Uniform No
7C Comb Boundary conditions Yes Yes Differential Non-Uniform No
7D Comb Square trailing edge Yes Yes Differential Non-Uniform No
8A Comb Increased CMC conductivity Yes Yes Differential Non-Uniform No
10A Temp Modified leading edge shape Yes Yes Differential Non-Uniform No
10B Temp Mesh refinement Yes Yes Differential Non-Uniform No
10C Temp Hemispherical rib Yes Yes Differential Non-Uniform No
10D Temp New temperature distribution Yes Yes. Differential Non-Uniform No
11A Pres Square tube - Hemispherical fork No N.A. Differential N.A. Yes/P.S.
11B Pres Meshing variation No N.A. Differential N.A. Yes/P.S.
11C Pres Circular tube No N.A. Differential N.A. Yes/P.S.
11D Pres Circular tube - mesh refinement No N.A. Differential N.A. Yes/P.S.
11E Pres Circular tube - meshing variation No N.A. Differential N.A. Yes/P.S.
11F1 Temp Circular tube - meshing variation No N.A. Differential Non-Uniform Yes/P.S.
11F2 Comb Circular tube - meshing variation No N.A. Differential Non-Uniform Yes/P.S.
11G1 Temp Square tube - meshing variation No N.A. Differential Non-Uniform Yes/P.S.
11G2 Comb Square tube - meshing variation No N.A. Differential Non-Uniform Yes/P.S.
12A Comb E33 variation E33=157 Yes Yes Differential Non-Uniform Yes/P.S.
12B Comb E33 variation E33=77 Yes Yes Differential Non-Uniform Yes/P.S.
13A Temp Revised temperatures Yes Yes Differential Non-Uniform Yes/Cl
13B Temp Insulated fork region Yes Yes Differential Non-Uniform Yes/Cl

1$6$&5² 
 a) Baseline vane - no rib - no trailing edge ejection

b) Differential wall thickness - no trailing edge ejection

1$6$&5² 
 c) Centerline trailing edge ejection

d) Pressure side trailing edge ejection

Fig. 5.0.3 Vane internal configurations

1$6$&5² 
 Table 5.0.2 - List of Comparisons

Page
5.1 - No Trailing Edge Ejection 27
5.1.1 - Effect of Rib 28
5.1.2 - Effect of Non-Uniform Temperature Gradient 31
5.1.3 - Effect of Differential Wall Thickness 33
5.1.4 - Effect of Combined Pressure and Temperature Loads 35
5.1.5 - Effect of Insulating Connecting Rib 36
5.1.6 - Effect of Modifying Rib Shape 37
5.1.7 - Effect of No Internal Pressure 38
5.1.8 - Effect of Revised Temperature Distribution 40
5.1.9 - Effect of Trailing Edge Shape 42
5.2 - Trailing Edge Ejection 43
5.2.1 - Effect of Centerline Trailing Edge Ejection 44
5.2.2 - Effect of Pressure Side Trailing Edge Ejection 47
5.2.3 - Effect of Modified Rib Shape 49
5.2.4 - Effect of Single Impingement Cavity 50
5.2.5 - Effect of Rectangular & Square Ejection Tube Geometry 52
5.2.6 - Effect of Square & Circular Tube Geometry 54
5.2.7 - Effect of Revised CMC Temperature Gradients 57
5.2.8 - Effect of Reduced Tube Heat Transfer Coefficient 59
5.2.9 - Effect of Insulating Fork Region 60
5.3 - Property and Boundary Condition Effects 61
5.3.1 - Effect of Through Thickness Stiffness Modulus 62
5.3.2 - Effect of Doubling Thermal Conductivity 64
5.3.3 - Effect of Reducing End Plate Stiffness 66
5.3.4 - Effect of Boundary Condition Variation 67
5.4 - Mesh Refinement Studies 68
5.4.1 - Effect of Vane & End Plate Mesh Density 69
5.4.2 - Effect of Mesh Shape for Square Ejection Tube 70
5.4.3 - Effect of Mesh for Circular Ejection Tube 71
5.5 - Maximum Component Stresses 72

The four sections of Table 5.0.2 describe the comparisons made to illustrate the effects
on midspan region stresses due to various internal configurations. The first section lists
the comparisons made for configurations without trailing edge ejection. The second section
lists comparisons for cases with trailing edge ejection. The third section lists comparisons
that investigate the effects of CMC property configurations and boundary conditions. The
fourth section lists comparisons that involved changes to the mesh used in the analysis. Also,
included in this table are the pages in which the comparisons are discussed.

1$6$&5² 
 5.1 - No Trailing Edge Ejection Comparisons

There are ten comparisons for cases with no trailing edge. These are the cases listed in
the first section of Table 5.0.2. Trailing edge ejection is commonly used to cool the portion
of the vane suction surface aft of the throat. This is done to avoid large aerodynamic loss
penalties associated with having film cooling rows aft of the vane throat. On the other hand
it may be desirable to avoid trailing edge ejection tubes in CMC vanes. These tubes have
a far greater length-to-hydraulic diameter ratio than are required for film cooling holes. If
trailing edge ejection is not feasible in a particular application, suction surface film cooling
rows downstream of the throat is an alternative. The reduction in vane cooling from using
CMC vanes could still be achieved. However, the aerodynamic efficiency benefits of reduced
coolant may be overwhelmed by the additional loss from having suction surface cooling rows
downstream of the throat. Unless vane efficiency is severely degraded from aft suction surface
film cooling rows, the benefits of reduced NOx would still be achieved.
In the following figures, which show stress contours, the vane orientation may be different
in each part of the comparison. The different orientations arise because maximum stress
locations are different for cash case. The vane in each figure was rotated to show the location
of maximum stress.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No rib - case 1A Rib - case 2A
Fig. 5.1.1.1 - Effect of rib - Pressure load only

5.1.1 - Effect of Rib. Figures 5.1.1.1 through 5.1.1.3 illustrate the effects of using a rib
to connect the suction and pressure surfaces of the vane. Without a rib there is a single hole
for an impingement tube, and with a rib two impingement tubes are needed. Comparisons
are made for just pressure loads, cases 1A and 2A, just thermal loads, cases 1C and 2B, and
when the pressure and thermal loads are simultaneously applied, cases 1D and 2C.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No rib - case 1C Rib - case 2B
Fig. 5.1.1.2 - Effect of rib - Thermal load only

Figure 5.1.1.1 shows through thickness and hoop stresses in the vane midspan region due
only to pressure loads for the baseline cases without a rib(1A), and a case with a rib(2A).
The presence of a rib is very effective in reducing maximum stresses. Without a rib the
maximum tensile stress occurs in the fork region, where the suction and pressure surfaces
join together. With a rib the maximum through thickness and pressure stresses occurs in
the rib region, and are much lower than the maximum through thickness and hoop stresses
for the case without a rib. Although not shown, for each of the six cases the maximum
spanwise or radial stresses in the midspan region were lower than the maximum hoop stress
for pressure, thermal, and combined loads.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No rib - case 1D Rib - case 2C
Fig. 5.1.1.3 - Effect of rib - Combined pressure and thermal loads

Figure 5.1.1.2 shows stresses due to non-uniform thermal loads. With a rib maximum
through thickness and hoop stresses are higher, and occur in the rib region. The thermal
stresses in the fork region do not significantly change due to the presence of a rib. In the
through thickness direction thermal load stresses are significantly greater than pressure load
stresses. When a rib is used the maximum through thickness stress due to thermal loads is
nearly seven times the maximum through thickness stress due to pressure loads. When a
rib is used the maximum hoop stress due to thermal loads is over twice the maximum hoop
stress due to pressure loads
Figure 5.1.1.3 shows stresses due to combined pressure and thermal loads. When no
rib is present high stresses occur in the fork region for both pressure and thermal loads.
Consequently, the maximum stress due to combining loads approaches adding the stress
for both types of loads. The results are different when a rib is present. Even though the
maximum hoop stress due to pressure loads is 17ksi(117MPa) the maximum stress due to
combined loads is only 4ksi(28MPa) greater than the maximum hoop stress due to thermal
loads.

1$6$&5² 
 300
150
250 Pressure surface Suction surface
Vane temperature difference, F
o

200 Uniform temperature difference

100
o
150 C

100
50
50

-50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Surface distance, s/CX

Fig. 5.1.2.1 Temperature difference across CMC vane walls

5.1.2 - Effect of Non-Uniform Temperature Gradient. Since thermal stresses are

greater than pressure stresses, the effect of a uniform temperature gradient across the walls of
the vane were determined. This was done for the case without a rib. Figure 5.1.2.1 shows the
temperature difference across the vane walls. This figure shows the unwrapped vane midspan
temperature difference. Local minimums occur just downstream of rows of cooling holes.
(The presence of cooling hole rows was not included in the stress analysis.) The temperature
difference shown in figure 5.1.2.1 results in the stress distribution previously shown in figure
5.1.1.4. Through thickness and hoop stresses for the uniform and non-uniform temperature
difference cases are shown in figure 5.1.2.2. The uniform temperature difference across the
CMC vane was 83.3◦ C(150◦ F ). This figure shows that a non-uniform temperature difference
increases both maximum through thickness and maximum hoop stress stress by over 50%.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Uniform ΔT - case 1B Non-uniform ΔT - case 1C
Fig. 5.1.2.2 Effect of Non-uniform temperature gradient - Thermal load

Calculations were done where the uniform temperature gradient across the vane wall was
halved. As expected, halving the gradient halves the thermal load stresses. Previous results
for cases 2A and 2B showed that thermal stress generally exceed thermal stresses. While
this comparison is for uniform differential temperatures, later comparisons for a case where
the CMC thermal conductivity was increased show similar stress reductions due to reduced
CMC vane thermal gradients.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Uniform wall thickness - case 2A Differential wall thickness - case 3A
Fig. 5.1.3.1 Effect of Non-uniform wall thickness - Pressure load

5.1.3 - Effect of Differential Wall Thickness. In an effort to reduce the thermal

gradient across the CMC vane wall a case was analyzed where the wall thickness along the
pressure surface and in leading edge region was halved. Reducing the wall thickness from
2mm(80mil) to 1mm(40mil) reduced thermal gradients across the pressure surface. It was
not practical to reduce the suction surface wall thickness below 2mm(80mil), since suction
surface stresses rose very rapidly as the suction surface wall thickness was reduced. Figure
5.1.3.1 shows that the maximum through thickness stress due to pressure loads is higher for
the thinner pressure surface wall. Fortunately, figure 5.1.3.2 shows that maximum through
thickness stress due to thermal loads is lower by nearly 20% for the thinner pressure surface
wall case.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Uniform wall thickness - case 2B Differential wall thickness - case 3B
Fig. 5.1.3.2 Effect of Non-uniform wall thickness - Thermal load

The maximum hoop stress is lowered when the pressure surface is thinner than the
suction surface. While the reduction in maximum hoop stress is only a few percent, stresses
for both pressure and thermal loads are reduced.

1$6$&5² 
 a) Through thickness stress, psi

a) Through thickness stress, psi

Uniform wall thickness - case 2C Differential wall thickness - case 3D
Fig. 5.1.3.3 Effect of Non-uniform wall thickness - Combined pressure and thermal loads

5.1.4 - Effect of Combined Pressure and Thermal Loads. Figures 5.1.1.1 through
5.1.1.3 illustrated the effects of combining pressure and temperature loads on stresses for
uniform wall thickness cases. Figures 5.1.3.1 through 5.1.3.3 illustrate the same effects for
the differential wall thickness case. These two sets of comparisons show behavior typical
of all cases examined. Also, the results are similar for either through thickness or hoop
stresses. The maximum stress for combined loads is significantly lower than the algebraic
sum of the maximum pressure load stress and the maximum thermal load stress. However,
at the location of the maximum tensile stress from one load, which is generally a thermal
load, there is a positive stress from the other load. The maximum stress from combined
loads is slightly greater than the maximum stress from either a pressure or thermal load.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Cooled rib - case 3B Insulated rib - case 3C
Fig. 5.1.5.1 Effect of insulating connecting rib - Thermal load

5.1.5 - Effect of Insulating Connecting Rib. The effect of insulating the rib con-
necting the vane pressure and suction surfaces is to reduce maximum stresses due to thermal
loads. Figures 5.1.5.1 shows that insulating the rib reduces peak through thickness stress
by nearly 20%. Both the magnitude and location of the maximum hoop stress is affected
by insulating the rib. With a cooled rib the maximum hoop stress occurs in the rib region,
while for the insulated rib the maximum hoop stress is in the fork region. The maximum
hoop stress is reduced by over 20% when the rib is insulated. For both cases internal cooling
was assumed to be done using an impingement insert. For an uncooled or insulated rib
there would be no impingement holes facing the rib. Since there are several pressure surface
film cooling rows, cross flows passing over the rib from the internal pressure surface to the
internal suction surface can be avoided.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Small rib radii - case 3A Semi-circular rib radii - case 5D
Fig. 5.1.6.1 Effect of Rib shape - Pressure load

5.1.6 - Effect of Modifying Rib Shape. Figure 5.1.6.1 shows stresses caused by
pressure loads for the reference rib geometry and a semi-circular rib geometry. In the semi-
circular rib case each rib surface used to form the front and rear impingement cavities has
a single radius of curvature. The rib in the other case has four corners with small radii
are only 2.3% of the axial chord, CX . The effect of larger rib radii is to reduce maximum
through thickness stress due to pressure loads by more than half. The semi-circular rib
increased the maximum hoop stress, and changed the location where the maximum hoop
stress occurred. The maximum hoop stress increased by nearly 7%, and moved from the
rib-suction surface region to the inner surface, near the leading edge. The minimum hoop
stress for the semi-circular rib occurred on the exterior surface of the vane. Unfortunately,
this rib configuration is unlikely to significantly affect thermal stresses.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Internal pressure - case 3A No internal pressure - case 3A2
Fig. 5.1.7.1 Effect of no internal pressure - Pressure load only

5.1.7 - Effect of No Internal Pressure If the vane is uncooled, which may be the
case for a HPT second stage CMC vane, there need not be an internal pressure causing
a ballooning stress. When there is no internal pressure there is still a bending load on
the vane, because the average external pressure surface pressure is higher than the average
suction surface external pressure. While a second stage vane has a much lower inlet total
pressure than the 50 atm. total pressure assumed for the first stage vane, the normalized
external pressure ratio, P/PT−IN , distribution is expected to be similar to that for the first
stage vane. Calculations showed that pressure load stresses are proportional to the inlet
total pressure, PT−IN , when the normalized pressure distribution, P/PT−IN , is constant. For
comparative purposes the analysis was done assuming the same inlet total pressure of 50
atm. as for the other cases. Figure 5.1.7.1 shows that pressure load stresses for a case with
no internal pressure are much higher than those where the internal pressure is 50 atm. Both
through thickness and hoop stresses are much higher for the case with no internal pressure.
There were no thermal loads present for the case with no internal pressure, and previous
cases showed that thermal stresses generally exceeded pressure stresses. With no internal
pressure the maximum through thickness stress shown in figure 5.1.7.1a nearly equals the

1$6$&5² 
maximum through thickness combined load stress for any case previously shown. Similarly,
the maximum hoop stress shown in figure 5.1.7.1b exceeds the maximum hoop combined
load stress for any of the previous cases. These results indicate that the stresses for an
uncooled first stage HPT vane may exceed those for a cooled first stage vane. Even for an
uncooled second stage HPT vane, stresses may be as high as for a cooled first stage vane, if
the thermal gradients in the first stage vane are moderate.

1$6$&5² 
 300
150
250 Pressure surface Suction surface
Vane temperature difference, F
o

Case 3D
200 Case 6C
100
o
150 C

100
50
50

-50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Surface distance, s/CX

Fig. 5.1.8.1 Two temperature differences across CMC vane walls

5.1.8 - Effect of Revised Temperature Distribution The arrangement of film cool-

ing rows can affect the CMC temperatures to some degree without affecting the amount
of required film cooling. Figure 5.1.8.1 shows two temperature differences across the CMC
vane. Case 6C has lower temperature differences across much of the vane surface. On the
other hand, case 6C has higher streamwise temperature gradients on the suction surface
between s/CX = 1 and s/CX = 1.4. The surface distance, s, is approximately twice the axial
chord, CX .
Figure 5.1.8.2 gives the through thickness and hoop stresses for the temperature differences
shown in figure 5.1.8.1. Stresses are given for combined loads.. The results are somewhat
unexpected, in that the lower temperature difference results in higher maximum through
thickness stress. The increase in maximum through thickness stress is not large, being less
than 8%, but is associated with a smaller temperature difference. Overall through thickness
stresses, though not the maximum stress, are reduced with a lower temperature difference.
This is most evidenced in the fork region where case 3D has stresses in excess of 5000
psi(34474 MPa), and the reduced temperature difference case, 6C, has no stresses in excess
of 4500 psi(31027 MPa).
Both maximum and local hoop stresses are reduced with lower temperature difference.
The reduction in maximum is only 4%. However, figure 5.1.8.2 shows large regions of the
pressure surface with lower hoop stresses when the temperature difference across the CMC
vane walls is lower.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Original temp. dist. - case 3D Revised temp. dist. - case 6C
Fig. 5.1.8.2 Effect of revised temperature distribution - Combined loads

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Circular trailing edge - case 7B Square trailing edge - case 7D
Fig. 5.1.9.1 Effect of modified trailing edge shape - Combined loads

5.1.9 - Effect of Trailing Edge Shape Figure 5.1.9.1 shows that the effect of changing
the shape of the trailing edge from a semi-circle to a squared-off trailing edge is small. This
analysis was done to determine if this geometric modification, which required a different grid
topology, would result in a significantly changed stress distribution near the vane trailing
edge. For the case with a semi-circular trailing edge, 7B, both through thickness and hoop
stresses are low in the trailing edge region, relative to the rest of the vane.

1$6$&5² 
 5.2 - Trailing Edge Ejection
Typical metallic HPT first stage vane have some form of trailing edge ejection,
whether along the vane centerline or at the rear of the pressure surface. From a fabrication
point alone it would be convenient if trailing edge ejection could be avoided for CMC vanes.
However, analysis showed that to maintain the suction surface of the vane either film cooling
rows had to be placed downstream of the vane throat, or substantially more cooling air was
required. Placing film cooling rows downstream of the vane throat, which is close to the
midpoint of the vane surface, causes a substantial vane efficiency penalty. To cool the aft
portion of the suction surface without trailing edge ejection requires high film effectiveness
at a relatively long distance from the last cooling row. To achieve the high film effectiveness
substantially more cooling air is required, since film effectiveness decreases with downstream
distance from the location of the film row, but increases with increased amounts of film
cooling. Trailing edge ejection is an effective cooling approach because there is a relatively
small thermal resistance between the outer vane surface and the trailing edge ejection tube.
In effect, trailing edge ejection redirects film cooling air from the last one or two rows of
pressure side cooling holes, and uses this air in a conventional duct cooling approach before
ejecting the air at the vane trailing edge.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No trailing edge ejection - case 3A Centerline trailing edge ejection - case 4A
Fig. 5.2.1.1 Effect of centerline trailing edge ejection - Pressure load

5.2.1 - Effect of Centerline Trailing Edge Ejection From a thermal standpoint

centerline trailing edge ejection is desirable because the distances between the trailing edge
ejection tube and the vane walls are minimized. Figures 5.2.1.1 through 5.1.1.3 show through
thickness and hoop stress when there are pressure, thermal, and combined loads. In the trail-
ing edge ejection case the trailing edge ejection tube is a converging duct with a uniform
height in the spanwise or radial direction. For all three loading conditions there are substan-
tial increased in maximum through thickness or hoop stress when there is a tube connecting
the aft cavity and the vane trailing edge. For pressure loads the maximum through thick-
ness stress increased by nearly a factor of four, while the maximum hoop stress increase by
nearly a factor of 1.5. For thermal loads the maximum through thickness stress increased
by a factor of 3.5. The increase in maximum hoop stress was less, being only 20%. As
in previous cases combined load stresses were heavily influenced by thermal load stresses.
The maximum combined load through thickness stress with trailing edge ejection was over
a factor of three greater than the maximum stress without trailing edge ejection. For hoop
stresses the increase in maximum stress was nearly 30%.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No trailing edge ejection - case 3C Centerline trailing edge ejection - case 4B
Fig. 5.2.1.2 Effect of centerline trailing edge ejection - Thermal load

While the maximum stresses are very high with trailing edge ejection, the high stresses
are confined to very small regions. For pressure load and combined load through thickness
stresses the maximum and minimum stresses are located near each other, and the maximum
thermal load stress is located near the tube exit, while the minimum through thickness stress
is on the inner pressure surface wall, close to the tube entrance. It may seem somewhat
surprising that the location for the maximum through thickness stress changes between
the thermal and combined load cases. The highly localized nature of the maximum stress
obscures the fact that there was a high, but not the highest, stress at the same location as
the maximum combined load stress.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

No trailing edge ejection - case 3D Centerline trailing edge ejection - case 4C
Fig. 5.2.1.3 Effect of centerline trailing edge ejection - Combined loads

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Centerline trailing edge ejection - case 4A Pressure side trailing edge ejection - case 5F
Fig. 5.2.2.1 Effect of pressure side trailing edge ejection - Pressure load

5.2.2 - Effect of Pressure Side Trailing Edge Ejection Because using a centerline
trailing edge ejection tube resulted in high maximum stresses, changing the location of the
trailing edge ejection tube was investigated. In this trailing edge configuration the tube
exited the vane on the its pressure side, The shape of the tube was also changed, to be easier
to fabricate. The pressure side trailing edge ejection tube had a constant rectangular cross
section, and was 0.86 mm(34 mil) high, and 0.25 mm(10 mil) wide.
Figures 5.2.2.1 and 5.2.2.2 show stresses for the centerline and pressure side trailing edge
ejection cases. While maximum through thickness stresses are still high, there was a substan-
tial reduction in maximum stresses when pressure side trailing edge ejection is substituted
for centerline trailing edge trailing edge ejection. The maximum through thickness stress is
reduced by approximately 25% for both the pressure load and combined load cases. While
the maximum hoop stress is also reduced by nearly 25% for the pressure load case, the
maximum hoop stress is nearly the same for centerline and pressure side ejection for the
combined load case.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Centerline trailing edge ejection - case 4C Pressure side trailing edge ejection - case 5C
Fig. 5.2.2.2 Effect of pressure side trailing edge ejection - Combined loads

The locations of maximum and minimum through thickness stresses are approximately
the same for pressure side and centerline trailing edge ejection. For combined loads the
location of maximum through thickness stress is unchanged between centerline and pressure
side trailing edge ejection. For the combined load case the location of minimum through
thickness stress is near the tube entrance for centerline ejection, and near the tube exit for
pressure side ejection. The location for maximum hoop stress due to pressure loads moves
away from the tube when there is pressure side ejection. The hoop stress in the rib region
reaches almost the same level for both forms of trailing edge ejection. For combined loads
the location of maximum hoop stress is the same for both forms of trailing edge ejection.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Small rib corner radii - case 4C Semi-circular rib - case 5A
Fig. 5.2.3.1 Effect of semi-circular rib with trailing edge ejection - Combined loads
5.2.3 - Effect of Modifying Rib Shape It was hoped that modifying the rib shape
would reduce maximum stresses, even if they occur away from the rib region. Figure 5.1.6.1
showed that a semi-circular rib was beneficial for cases without trailing edge ejection. Figure
5.2.3.1 shows stresses for cases with centerline trailing edge ejection and two rib geometries.
The rib geometries are the same as those shown in figure 5.1.6.1. One is a rib with four
small radii where the rib joins the pressure and suction surfaces. The other geometry has
the rib sides formed by constant radius arcs. The radius of the rib side facing the aft cavity
is smaller than the radius of the rib side facing the forward cavity.
Because maximum through thickness and hoop stresses occur in the tube region, the
rib shape had no significant effect on maximum stresses. In the rib region the results for
combined pressure and thermal loads shown in figure 5.2.3.1 are different from the results
shown in figure 5.1.6.1 for just pressure loads. For combined loads the semi-circular rib sides
show higher through thickness stresses with a semi-circular rib, but for pressure loads the
stresses are lower for a semi-circular rib. For combined loads the semi-circular rib shows
lower hoop stresses, especially where the rib joins the vane pressure surface forming the
forward cavity. For pressure loads figure 5.1.6.1 shows only a small decrease in stresses in
the rib region for the semi-circular rib geometry.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Two impingement cavities - case 5F Single impingement cavity - case 5E
Fig. 5.2.4.1 Effect of single impingement cavity - Pressure load

5.2.4 - Effect of Single Impingement Cavity Since semi-circular ribs were generally
preferable to ribs with four small radii corners, a case was analyzed where the radius near
the trailing edge ejection tube was maximized. This resulted in a single cavity configuration.
Figure 5.2.4.1 and 5.2.4.2 shows result for two cavity and single cavity configurations. Both
cases have pressure side trailing edge ejection tubes. Figure 5.2.4.1 shows results for pressure
loads, and figure 5.2.4.2 shows results for combined loads. The single cavity configuration
reduces maximum through thickness stress by over 60% for pressure load cases, and by over
25% for the combined load cases. For maximum hoop stresses the single cavity configuration
was reduced by nearly 25% for the combined load case, but only by 5% for the pressure
load only case. Other than a lower maximum through thickness stress, which is definitely
beneficial, the through thickness stress distribution within the trailing edge ejection tube is
similar for the single and two cavity cases. The location of maximum hoop stress moves
away from the entrance to the trailing edge ejection tube for the single cavity cases. For the
pressure load case the location of maximum hoop stress moved from the suction side of the
rib to the leading edge region.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Two impingement cavities - case 5C Single impingement cavity - case 5G
Fig. 5.2.4.2 Effect of single impingement cavity - Combined loads

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Rectangular tube - case 5E Square tube - case 11A
Fig. 5.2.5.1 Effect of trailing edge ejection tube geometry - Pressure load

5.2.5 - Effect of Rectangular & Square Ejection Tube Geometry The effect of
the tube geometry was found to significantly affect stresses. The maximum through thickness
stress increased by more than a factor of four when the tube geometry was changed from a
0.86mm X 0.25 mm(34 mil X 10mil) rectangle to a 0.76mm(30mil) square. Figure 5.2.5.1 for
pressure loads shows that the maximum through thickness stress increases by more than a
factor of four when the rectangular tube is replaced by a square tube. The maximum hoop
stress increases by less than 9%, but the location of maximum hoop stress moves from the
leading edge region to the entrance of the square tube.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Square tube - case 11B Circular tube - case 11C
Fig. 5.2.6.1 Comparison of square and circular tube geometries - Pressure load

5.2.6 - Effect of Square & Circular Tube Geometry Figures 5.2.6.1 through 5.2.6.3
shows stresses for square and circular trailing edge ejection tube with pressure, thermal, and
combined loads. The diameter of the circular tube, 0.76mm(30mil), is the same as the width
of the square tube. Both have the same hydraulic diameter, but the cross sectional area of
the square tube is greater by just over 20%. The difference in maximum stresses in either the
through thickness or hoop directions between the square and circular trailing edge ejection
tubes is small. While the maximum through thickness stress is higher for the circular tube for
thermal loads, the maximum stress is lower for pressure or combined loads. The maximum
hoop stress is higher for the circular tube for pressure loads, but the maximum hoop stress
is lower for thermal or combined loads.
Comparing these results with those in figure 5.2.4.2 shows that for combined loads the
maximum through thickness or hoop stress is nearly the same for all three trailing edge tube
geometries. On the other hand figure 5.2.4.1 shows that for pressure loads the rectangular
tube geometry has a much lower maximum through thickness stress, and also a reduced
maximum hoop stress.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Square tube - case 11G1 Circular tube - case 11F1
Fig. 5.2.6.2 Comparison of square and circular tube geometries - Thernal load

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Square tube - case 11G2 Circular tube - case 11F2
Fig. 5.2.6.3 Effect of trailing edge ejection tube geometry - Thermal load

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Square tube - case 11G2 Circular tube - case 11F2
Fig. 5.2.6.3 Effect of trailing edge ejection tube geometry - Combined loads

1$6$&5² 
 300
150
250
Vane temperature difference, F
o

Case 4B
200 Case 13A
100
o
150 C

100
50
50

0
Pressure surface Suction surface
-50
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5 3
Surface distance, s/CX

Fig. 5.2.7.1 Temperature difference across CMC vane walls

5.2.7 - Effect of Revised CMC Temperature Gradients Figure 5.2.7.1 shows two
temperature difference distributions for the vane. Case 13A has a much lower average tem-
perature difference across the CMC vane than does case 4B. To achieve the lower tempera-
ture difference case 13A required addition vane cooling. While the multiple film cooling row
cause large variations in the temperature difference for both cases, case 13A has an average
temperature difference only about half of the average temperature difference for case 4B.
Figure 5.2.7.2 shows stresses for both cases due to thermal loads. Surprisingly, the lower
average temperature difference case, 13A, resulted in higher maximum through thickness
stress. In the rib region the lower average temperature case, 13A, shows lower through
thickness stresses. Near where the maximum through thickness stress occurs, the streamwise
temperature gradient on the pressure surface are of opposite sign. Between s/CX = −1.4 and
-1.6 on the pressure surface case 4B has a negative streamwise pressure gradient, and case
13A has a positive streamwise pressure gradient. For case 13A temperatures are increasing
with pressure surface distance, and for case 4B temperatures are decreasing with pressure
surface distance. These results imply that streamwise temperature gradients, as well as
temperature gradients across the vane significantly affect thermal stresses.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Reference case temperature gradients - case 4B Lowered CMC temperature gradients -
case 13A
Fig. 5.2.7.1 Effect of changing CMC temperature gradient - Thermal load

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

◦
hTUBE = 8000W/m − K - case 5A hTUBE = 4000W/m −◦ K - case 5B
Fig. 5.2.8.1 Effect of varying ejection tube heat transfer coefficient - Thermal load

5.2.8 - Effect of Reduced Tube Heat Transfer Coefficient The heat transfer coef-
ficient(s) within the ejection tube can be varied in a number of ways. Reducing the amount
of trailing edge ejection coolant reduces the average tube heat transfer coefficient, and the
reduction in tube flow rate could be balanced by an increase in film cooling. Increasing the
tube cross sectional area for the same flow rate would also reduce the average tube heat
transfer coefficient. An analysis was done to determine the effects of halving the ejection
tube heat transfer coefficient. Figure 5.2.8.1 shows the results of this analysis. Maximum
through thickness stresses are reduced by 18% when the tube heat transfer coefficient is
halved. The location of maximum through thickness stress is unchanged. Hoop stresses are
not affected by reducing the average heat transfer coefficient.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Uninsulated fork region - case 13A Insulated fork region - case 13B
Fig. 5.2.9.1 Effect of insulating fork region - Thermal load

5.2.9 - Effect of Insulating Fork Region Because high stresses were often seen at
the entrance of the trailing edge ejection tube, an analysis was done to determine if these
stresses would be less if the fork region at the entrance to the trailing edge ejection tube
was insulated. Figure 5.2.9.1 shows the results of this analysis for a case with thermal loads.
The thermal boundary condition on the cavity surface in the fork region had no noticeable
effect on either through thickness or hoop stresses.

1$6$&5² 
 5.3 -CMC Material Property and Boundary Condition Effects

In this section four comparisons are examined. The first comparison is for the effects of
varying the through thickness stiffness modulus, E33 , This parameter is expected to experi-
ence significant variations due to even small changes in material composition. The second
comparison is for variations in thermal conductivity. It is desirable to increase CMC ther-
mal conductivity, since this decreases the thermal gradients across the vane walls. The third
comparison examines the effects of changing the stiffness of the two vane end plates. This
was done by halving the thickness of each end plate. This comparison helps to answer the
question of whether increasing the rigidity lowers stresses in the midspan region. The fourth
comparison examines the effects of changing the boundary conditions on midspan stresses.
The boundary conditions are applied to the edges of each end plate.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

E33 = 157 GPa(22.8 M psi) - case 6A E33 = 77 GPa(11.2 M psi) - case 6B
Fig. 5.3.1.1 Effect of through thickness stiffness- Combined loads

5.3.1 - Effect of Through Thickness Stiffness Modulus The through thickness

stiffness modulus, E33 , was varied to determine the sensitivity of stresses to changes in this
parameter. When E33 was varied Poisson’s ratio was held constant at 0.17 for the through
thickness direction properties. Figure 5.3.1.1 and 5.3.1.2 show stresses for two values of E33
for pressure and thermal loads combined. Figure 5.3.1.1 compares cases without trailing
edge ejection, and figure 5.3.1.2 compares cases with pressure side trailing edge ejection
through a rectangular ejection tube. Stresses are reduced as E33 is reduced, but by less than
proportionately. For cases without trailing edge ejection, and when E33 is nearly halved
the maximum through thickness stress is reduced by only about 20%. Nearly halving E33
reduces the maximum hoop stress by 12%. Figure 5.3.1.2 shows similar results for trailing
edge ejection cases. Nearly halving E33 reduces the maximum through thickness stress by
28%, and reduces the maximum hoop stress by 11%.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

E33 = 157 GPa(22.8 M psi) - case 12A E33 = 77 GPa(11.2 M psi) - case 12B
Fig. 5.3.1.2 Effect of through thickness stiffness- Combined loads

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

kCMC = 20 W/m-K - case 6G kCMC = 40 W/m-K - case 8A
Fig. 5.3.2.1 Effect of doubling CMC thermal conductivity - Combined loads

5.3.2 - Effect of Doubling Thermal Conductivity Increasing the CMC thermal

conductivity decreases the temperature difference across the CMC vane. The difference is
not proportional to the increase in thermal conductivity. The overall thermal resistance is
determined by the internal and external heat transfer coefficients, the thermal resistance
of the EBC protective layer, as well as the thermal resistance of the vane itself. When the
CMC thermal conductivity is increased less vane coolant is required, so that the temperature
differences across the vane also change. Under these circumstances the chief benefit for stress
reductions may be reductions in streamwise temperature gradients. Since the analysis was
done with less cooling for the higher thermal conductivity case, there is also a cycle benefit
to higher CMC conductivity.
Figure 5.3.2.1 shows a comparison for cases with combined loads and no trailing edge
ejection. Maximum through thickness stress is reduced by more than 25% when the CMC
thermal conductivity is doubled from 20W/m −◦ K. Maximum hoop stress is reduced by less
than 9%. The location of maximum stress changed when the CMC conductivity was doubled.
When the CMC thermal conductivity doubled, the location of the maximum through
1$6$&5² 
thickness stress moved from the suction side of the fork region to forward of the rib on the
pressure surface. The location of maximum hoop stress remained in the fork region, but
shifted from the suction side to the pressure side as the conductivity doubled.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Original end plate thickness- case 6G End plate thickness halved - case 7A
Fig. 5.3.3.1 Effect of reducing end plate thickness - Combined loads

5.3.3 - Effect of Reducing End Plate Stiffness The stiffness of the end plate has
a significant effect on vane stresses, even in the midspan region, away from the end plate.
Previous results were calculated where the two end plates were 4mm(160 mil) thick, and
had the same material properties as the CMC vane. Figure 5.3.3.1 shows stresses when
the thickness of both end plates is halved. Results are shown for combined loads and cases
where there is no trailing edge ejection. This allows greater vane deflections, especially in
the midspan region. The maximum through thickness stress increased by nearly 45% when
thickness of both end plates was halved, and the maximum hoop stress increased by 36%.
The location of maximum stresses also changed. The maximum through thickness stress
moved from the suction side of the fork to just forward of the rib on the pressure surface
when the end plate thicknesses were halved. The maximum hoop stress moved from the
suction side of the fork to the leading edge region when the end plate thicknesses were
halved. These results show the desirability of using stiff end plates with CMC vanes.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Original boundary condition- case 7B Modified boundary condition - case 7C
Fig. 5.3.4.1 Effect of modifying plate boundary condition - Combined loads

5.3.4 - Effect of Boundary Condition Variation The effect of changing the end
plate boundary conditions was examined. Figure 5.3.4.1 shows results for two cases with
different end plate boundary conditions. Case 7B has the standard boundary conditions,
and case 7C has the modified boundary conditions. For the standard boundary conditions
the hub end plate is held in the axial and spanwise directions, and along its leading edge in
the pitchwise direction. The other end plate is free to move in all directions. For case 7C
there is an additional constraint along the edge of the hub end plate. The hub end plate
is fixed at the pressure side corner, where the axial and pitchwise edges meet. When the
additional constraint was added to the end plate, the maximum through thickness stresses
decreased by 12%, and the location of maximum stress remained the same. The maximum
hoop stresses increased by 12%, and the maximum hoop stress location moved from the
pressure side of the fork to the leading edge region.

1$6$&5² 
 5.4 - Mesh Refinement Studies
This section gives the results of four comparisons for results with different computational
meshes. The first comparison examines the effects on midspan region stresses due to a
variation in the mesh density used for each end plate. The second comparison examines the
effects of changing the mesh density in the leading edge region. The last two comparisons
examine the effects on stresses for changes in the mesh used around the trailing edge ejection
tube. One case is for a square trailing edge ejection tube and the other is for a circular tube.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

“Coarse” mesh - case 7A Finer mesh - case 7B
Fig. 5.4.1.1 Effect of increasing end plate mesh density - Combined loads

5.4.1 - Effect of Vane and End Plate Mesh Density Figure 5.4.1.1 shows stresses
for two cases with different mesh density. Comparisons are for combined thermal and pressure
loads. This comparison is for thinner end plates of 2mm(80mil) thickness, and not the
standard end plate thickness of 4mm(160mil). Case 7A has typical mesh of 30000 total
elements for cases without trailing edge ejection. Case 7B has a more dense mesh of 50000
total elements. When the number of end plate elements increased, there was also an increase
in the number of vane elements. Only in the region of maximum stress does increasing the
number of elements have an effect. It is not unexpected that an increase in the number
of elements would increase the maximum stress. Since maximum stresses are so highly
localized, the magnitude of the stress gradient is very high at the location of maximum
stress. Increasing the number of elements positions nodes closer to the theoretical location
of maximum stress, and a higher maximum stress is expected as the mesh density increases.
Figure 5.1.1,1a shows that the maximum through thickness stress increased by 17%. As
expected the location of maximum through thickness stress did not change. Somewhat
unexpectedly figure 5.4.1.1b shows that the maximum hoop stress decreased by 5% as the
total number of elements increased. The response of the maximum hoop stress to the increase
in the number of elements occurred because the location of maximum hoop stress changed
as the number of elements increased. The maximum hoop stress moved from leading edge
to the pressure side of the fork region.
1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

Hexagonal mesh elements - case 11A Hexagonnal & tetrahedral elements - case 11B
Fig. 5.4.2.1 Effect of varying mesh type for square tube - Pressure load

5.4.2 - Effect of Mesh Shape for Square Ejection Tube Figure 5.4.2.1 shows the
effects of mesh element shape for squared tube cases. For case 11A a hexahedral mesh was
used 11A with 130,000 elements. Typically, hexahedral elements were used for the ANSYS
analysis. The number of elements was significantly increased when trailing edge ejection
was used because a mesh was needed for each of the ten trailing edge ejection tubes. Case
11B used a combination of hexahedral and tetrahedral mesh elements. Transitional pyramid
elements were used in the fork and trailing edge regions. Case 11B had 230,000 elements.
Through thickness stresses are unaffected by the shape of the mesh elements. Only the
magnitude of the maximum hoop stress is affected by the shape of the mesh elements.
While the increase of 28% in maximum hoop stress is large, the increase is confined to a
very small region. Part of the increase in maximum hoop stress may be due to a larger
number of mesh elements. Also, hexahedral elements are preferred to tetrahedral elements
for achieving accurate results, Brauer[31]. However, tetrahedral elements are required when
the trailing edge ejection tube is circular.

1$6$&5² 
 a) Through thickness stress, psi

b) Hoop stress, psi

“Standard” mesh density case 11C Reduced mesh density - case 11E
Fig. 5.4.3.1 Effect of varying mesh density for circular tube - Pressure load

5.4.3 - Effect of Mesh for Circular Ejection Tube Due to complexity of circular
ejection tube model, tetrahedral shaped elements were used in case 11C. This case had
720,000 elements. Case 11E had a combination of hexahedral and tetrahedral elements,
Transitional pyramid elements were used in the fork and trailing edge regions. Case 11E
had 350,000 elements. Figure 5.4.3.1 shows that the only effect of using fewer elements is a
reduction in maximum stress. The maximum through thickness stress is reduced by 19%,
but the location at the tube entrance is unchanged. The maximum hoop stress is reduced
by 27%, and the location of maximum stress changed between the two cases. For case 11C
the maximum hoop stress occurred more than half way down the ejection tube, and for case
11E the maximum hoop stress occurred in the leading edge region.

1$6$&5² 
 5.5 - Maximum Component Stresses
Even though maximum stresses are often confined to a very small region of the vane,
comparisons of maximum stress in the three component directions is useful. These compar-
isons help to show a path to reduced stresses. by different vane internal geometries, and also
the effects of modeling assumptions.. Tables IIa through IId show the maximum component
stresses, and the location where these stresses occur. The maximum through thickness or
interlaminar, the maximum hoop, and the maximum spanwise or radial stress is given for
pairs of cases. Comparisons are given for pairs of cases to illustrate the change in maximum
stresses for a specific effect. Table IIa shows results for cases without trailing edge ejection,
and Table IIb shows results for cases with centerline and pressure side trailing edge ejec-
tion. Table IIIc shows results for property and boundary condition variations, while Table
IId shows results of mesh sensitivity studies. While no figures were given for the spanwise
stresses in the midspan region, these tables include the maximum spanwise stress in the
midspan region, The location of the maximum spanwise stress is also given.
Table IIa shows that, except for one case, the maximum spanwise stress is lower than
the maximum hoop stress. The assumed orientation of the reference material, N24A, was
that the material properties in the hoop and span directions were the same. Since spanwise
maximum stresses were almost always lower than the maximum stress in the hoop direction,
hoop stresses were of greater interest. Table IIa shows that the location of maximum hoop
stress was generally different from the location of maximum spanwise stress.
Table IIa shows that the case of no internal pressure, 5.1.7, results in very high maximum
hoop and spanwise stresses. For this case the maximum through thickness stress is compa-
rable to the maximum through thickness stress for cases with combined loads. Without an
internal pressure no vane thermal stresses are expected to occur.
Maximum thermal stresses are generally higher than maximum pressure stresses. Table
IIa shows that maximum thermal stresses are very sensitive to variations in temperature
distributions. There was over a 50% increase in maximum thermal load stresses when a
uniform CMC temperature differential was replaced by wall temperatures consistent with a
film cooled vane. There was over a 20% reduction in maximum thermal load stresses when
the rib connecting the pressure and suction surface was insulated. Table IIb shows similar
results for cases with trailing edge ejection. Changing the CMC temperature gradients
resulted in maximum through thickness and spanwise stresses increasing by more than 30%.
Table IIb shows that maximum stresses increase significantly when there is a trailing
edge ejection tube. Pressure side, as differentiated from centerline, trailing edge ejection
is sometimes favored for vane efficiency reasons, since it allows for a thinner trailing edge
thickness. The results in Table IIb show that it is also preferable from a stress standpoint.
For pressure loads and combined loads the reduction in maximum through thickness stress
is over 25%. For combined loads the reduction in maximum spanwise stress is nearly 40%.
Maximum hoop loads are less affected by going from centerline to pressure side ejection.
In contrast to the results without trailing edge ejection, maximum spanwise stresses for
several cases exceed maximum hoop stresses. However, maximum spanwise stresses vary
over a wide range. For the single impingement cavity case, which has the lowest combined
load maximum through thickness stress, the maximum spanwise stress is within 15% of
the maximum hoop stress. The 5.2.7 comparison shows that the maximum spanwise stress

1$6$&5² 
decreased by over 10% when the trailing edge ejection tube was halved.
Table IIc shows that reducing the through thickness modulus of elasticity, E33 , or in-
creasing the CMC thermal conductivity is effective in reducing maximum through thickness
stresses. Using thinner end plates, which is desirable from a weight standpoint, results in
higher maximum stresses in the midspan region of the vane. Halving the thickness of the
end plates caused the maximum through stress to increase by more than 40%. The other
maximum stresses also increased.
Table IId shows that maximum stresses are somewhat sensitive to the type of mesh used.
Since all other comparisons were done with the same mesh type, the relative change between
cases is expected to be accurate. The values in Table II are maximum values, and near the
maximum values the component stresses change very rapidly. Therefore, it is not surprising
that as the number of elements in the finite element analysis increase, maximum stresses also
increase. The results for comparisons 5.4.1 and 5.4.3 show that more elements give higher
maximums in regions of high gradients. Using a mixture of hexahedral and tetrahedral mesh
element resulted in higher maximum hoop and spanwise stresses compared with just using
hexahedral elements.

1$6$&5² 
 Table IIa. Comparisons for midspan region maximum calculated stress, ksi,(MPa)
5.1 - No Trailing Edge Ejection

Case Load Through thickness Hoop Span

Value Location Value Location Value Location
5.1.1 - Effect of Rib
1A Press 4.62(31.9) Fork 31.4(217) Fork 13.3(92.0) T.E.
2A Press 1.57(10.8) Rib 17.0(117) Rib 13.3(92.0) T.E.
% change -66 -46 0
1C Temp 6.91(47.6) Fork 31.1(214) L.E. 28.5(197) Fork
2B Temp 10.5(72.4) Rib 41.5(286) Rib 34.3(236) P-aft of rib
% change +52 +33 +20
1D Comb 10.8(75.4) Fork 55.5(383) Fork 29.1(201) Fork
2C Comb 10.6(73.1) Rib 45.6(314) Rib 36.9(254) P-aft of rib
% change -2 -18 +27
5.1.2 - Effect of Non-Uniform Temperature Gradient
1B Temp 4.44(30.6) L.E. 20.4(141) L.E. 17.7(122) Fork
1C Temp 6.91(47.6) Fork 31.1(214) L.E. 28.5(197) Fork
% change +56 +53 +61
5.1.3 - Effect of Differential Wall Thickness
2A Pres 1.57(10.8) Rib 17.0(117) Rib 13.3(92.0) T.E.
3A Pres 2.40(16.5) Rib 16.7(115) Rib 13.8(94.8) P-L.E.
% change +53 -2 +4
2B Temp 10.5(72.4) Rib 41.5(286) Rib 34.3(236) P-aft of rib
3B Temp 8.52(58.7) Rib 38.2(263) Rib 31.2(215) Fork
% change -20 -8 -9
5.1.4 - Effect of Combining Pressure and Temperature Loads
3A Pres 2.40(16.5) Rib 16.7(115) Rib 13.8(94.8) P-L.E.
3C Temp 6.61(45.6) Rib 29.4(203) Fork 26.3((182) P-aft of rib
3D Comb 7.28(50.2) Fork 36.6(252) Fork 27.2(188) P-aft of rib
5.1.5 - Effect of Insulating Rib
3B Temp 8.52(58.7) Rib 38.2(263) Rib 31.2(215) Fork
3C Temp 6.61(45.6) Rib 29.4(203) Fork 26.3(182) P-aft of rib
% change -22 -23 -16
5.1.6 - Effect of Modified Rib Shape
3A Pres 2.40(16.5) Rib 16.7(115) Rib 13.8(94.8) P-L.E.
5D Pres 1.11(7.65) Rib 17.9(123) L.E. 15.4(106) P-for. of rib
% change -54 +7 +12
5.1.7 - Effect of No Internal Pressure
3A Pres 2.40(16.5) Rib 16.7(115) Rib 13.8(94.8) P-L.E.
3A2 Pres 10.9(75.2) Rib 115.(793) Rib 59.8(412) P-for. of rib
% change +354 +589 +333
5.1.8 - Effect of Revised Temperatures
3D Comb 7.28(50.2) Fork 36.6(252) Fork 27.2(188) P-aft of rib
6C Comb 7.82(53.9) Fork 35.1(242) Fork 40.2(277) P-for. of rib
% change +7 -4 +48
5.1.9 - Effect of Trailing Edge Shape
7B Comb 10.8(74.2) Rib 38.1(263) Fork 38.4(264) P-for. of rib
7D Comb 10.8(74.2) Rib 39.2(270) L.E. 38.4(264) R-for. of rib
% change 0 +3 0

1$6$&5² 
 Table IIb. Comparisons for midspan region maximum calculated stress, ksi,(MPa)
5.2 - Trailing Edge Ejection

Case Load T.E. Through thickness Hoop Span

Ejection Value Location Value Location Value Location
5.2.1 - Effect of Centerline Trailing Edge Ejection
3A Pres No 2.40(16.5) Fork 16.7(115) Fork 11.1(76.5) P-front
4A Pres Cl 9.04(62.3) Tube 20.2(139) Fork 16.5(114) Fork
% change +277 +21 +41
3C Temp No 6.61(45.6) Rib 29.4(203) Fork 26.3(181) P aft of rib
4B Temp Cl 23.1(159) Tube 35.4(244) Tube 74.4(513) Tube
% change +249 +20 +183
3D Comb No 7.28(50.2) Fork. 36.6(252) Fork 27.2(178) P aft of rib
4C Comb Cl 22.5(155) Tube. 49.3(340) Tube 83.4(575) Tube
% change +209 +35 +207
5.2.2 - Effect of Pressure Side Trailing Edge Ejection
4A Pres Cl 9.04(62.3) Tube 20.2(139) Fork 16.5(114) Fork
5F Pres PS 5.60(38.6) Tube 17.8(123) Rib 17.1(118) Fork
% change -38 -12 +4
4C Comb Cl 22.5(155) Tube 49.3(340) Tube 83.4(575) Tube
5C Comb PS 16.7(115) Tube 49.4(340) Fork 51.2(353) Fork
% change -26 0 -39
5.2.3 - Effect of Revised Rib Shape
4C Comb Cl 22.5(155) Tube 49.3(340) Tube 83.4(575) Tube
5A Comb PS 22.5(155) Tube. 48.2(333) Fork. 83.4(575) Tube
% change 0 -2 0
5.2.4 - Effect of Single Impingement Cavity
5F Pres PS 5.60(38.6) Tube 17.8(123) Rib 17.1(118) Fork
5E Pres PS 2.19(15.1) Tube 16.9(117) L.E. 15.8(109) Pres. surf.
% change -61 -5 -8
5C Comb PS 16.7(115) Tube 49.4(340) Fork 51.2(353) Fork
5G Comb PS 12.2(84.4) Tube 37.9(261) P.S. 43.5(300) Tube
% change -27 -23 -15
5.2.5 - Effect of Trailing Edge Ejection Tube Geometry
5E Pres PS 2.19(15.1) Tube 16.9(117) L.E. 15.8(109) Pres. surf.
11A Pres PS 9.29(64.1) Tube 18.4(127) Fork 18.5(128) Pres. surf.
% change +324 +9 +17
5.2.6 - Effect of Square & Circular Tube Geometry
11A Pres PS 9.29(64.1) Tube 18.4(127) Fork 18.5(128) Pres. surf.
11C Pres PS 8.94(61.6) Tube 13.8(95 ) L.E. 94.0(648) Pres. surf.
% change -3 -25 +408
11G1 Temp PS 10.3(71.0) Tube 32.9(227) P.S. 69.7(480) Tube
11F1 Temp PS 11.0(76.1) Tube 30.6(211) P.S. 74.2(512) Tube
% change +7 -7 +6
11G2 Comb PS 14.8(102) Tube. 34.4(237) P.S. 71.8(495) Tube
11F2 Comb PS 13.2(91.1) Tube 33.9(233) S.S. 75.5(521) Tube
% change -11 -1 +5
5.2.7 - Effect of Revised CMC Temperature Gradients
4B Temp Cl 23.1(159) Tube 35.4(244) Tube 74.4(513) Tube
13A Temp Cl 30.7(212) Tube 35.4(244) Tube 102(701) Tube
% change +33 0 +37
5.2.8 - Effect of Reduced Tube Heat Transfer Coefficient
5A Comb PS 22.5(155) Tube. 48.2(333) Fork. 83.4(575) Tube
5B Comb PS 18.4(127) Tube. 49.0(338) Fork. 74.2(512) Tube
% change -18 +2 -11
5.2.9 - Effect of Insulating Fork Region
13A Temp Cl 30.7(212) Tube 35.4(244) Tube 102(701) Tube
13B Temp Cl 30.7(212) Tube 34.4(244) Tube 102(701) Fork
% change 0 0 0

1$6$&5² 
 Table IIc. Comparisons for midspan region maximum calculated stress, ksi,(MPa)
5.3 - Property and Boundary condition effects

Case Load T.E. Through thickness Hoop Span

Ejection Value Location Value Location Value Location
5.3.1 - Effect of Through Thickness Stiffness Modulus
6A- E33 = 157 GPa Comb No 6.87(47.4) Fork 33.9(234) Fork 24.4(167) P-aft of fork
6B E33 = 77 GPa Comb No 5.40(37.2) Fork 29.9(206) Fork 24.5(169) P-L.E.
% change -22 -12 0
12A - E33 = 157 GPa Comb PS 13.7(94.9) Tube 46.6(321) Fork 47.8(329) Tube
12B - E33 = 77 GPa Comb PS 9.81(67.7) Tube 41.6(287) Fork 46.9(324) Tube
% change -29 -11 -2
5.3.2 - Effect of CMC Thermal Conductivity
6G Comb No 6.41(44.2) Fork 29.5(204) Fork 32.4(224) P-aft of fork
8A Comb No 4.73(32.6) Fork 27.0(186) Fork 29.7(205) P-for. of rib
% change -26 -8 -8
5.3.3 - Effect of End Plate Stiffness
6G Comb No 6.41(44.2) Fork 29.5(204) Fork 32.4(224) P-aft of fork
7A Comb No 9.24(63.7) Fork 40.0(276) L.E. 38.3(264) P-for. of rib
% change +44 +36 +18
5.3.4 - Boundary Conditions
7B Comb No 10.8(74.5) Rib 38.1(263) Fork 38.4(265) P-for. of rib
7C Comb No 9.49(65.4) Rib 42.5(293) L.E. 38.8(268) P-for. of rib
% change -12 +12 +1

Table IId. Comparisons for midspan region maximum calculated stress, ksi,(MPa)
5.4 - Mesh Refinement
Case Load T.E. Through thickness Hoop Span
Ejection Value Location Value Location Value Location
5.4.1 - End Plate
7A Comb No 9.24(63.7) Fork 40.0(276) L.E. 38.3(264) P-front
7B Comb No 10.8(74.5) Rib 38.1(263) Fork 38.4(265) P-front
% change +17 -5 0
5.4.2 - Square Trailing Edge Ejection Tube
11A Pres PS 9.29(64.1) Tube 18.4(127) Fork 18.5(128) Pres. surf.
11B Pres PS 9.27(64.0) Tube 23.4(161) Tube 20.3(140) Pres. surf
% change 0 +27 +10
5.4.3 - Circular Trailing Edge Ejection Tube
11C Pres PS 8.94(61.6) Tube 13.8(95 ) L.E. 94.0(648) Pres. surf.
11E Pres PS 7.22(50.0) Tube 17.5(121) L.E. 29.4(122) Pres. surf.
% change -19 +27 -69

1$6$&5² 
 5.6 - Conclusions

The results of the stress analysis for the full size EEE vane showed that stresses from
expected thermal loads are greater than stresses expected from pressure loads even though
the inlet total pressure was 50 atm. All components of stress, through thickness, hoop, and
spanwise, were higher due to thermal loads compared with pressure loads. Maximum stresses
occurred either in the fork region, where the pressure and suction surfaces meet, or in the
rib connecting these two surfaces. When there was trailing edge ejection, maximum stresses
often occurred in the trailing edge ejection tube. These maximum stresses often occurred in
the fork region near the tube entrance.
The reference material, N24A, was a two-dimensional woven material. The desired max-
imum stress in the trough thickness, or interlaminar, direction was 10.3 MPa(1.5 ksi). The
desired maximum stress in the hoop and spanwise directions was 17ksi(117 MPa). Maximum
calculated combined load through thickness stress was more than four times greater than the
desired maximum through thickness stress. The maximum combined load hoop stress and
the maximum spanwise stress were about double the desired stress for the better low stress
cases examined. Maximum stresses were very localized. In many cases close to 98% of the
vane cross section had stresses less than one half of the maximum component stress. Also, for
cases with trailing edge ejection maximum and minimum stresses were located close to each
other within the ejection tube. This suggests that a creep analysis of vane stresses, which
is expected to be more applicable in the engine environment, may result in lower maximum
component stresses. Kaufman[xx] showed a significant reduction in maximum stress for a
metal vane when creep was taken into account.
The analysis was done for thermal loads, pressure loads, and thermal and pressure loads
combined. Stresses due to thermal loads were greater than those due to pressure loads. The
location of maximum pressure load stress was not coincident with the location of maximum
thermal load stress. The maximum combined load component stress was generally only
slightly greater than the the maximum thermal load component stress.
The stress analysis results indicate that not including stress augmentation due to rows
of film cooling holes is not a short coming of the analysis. Hoop stresses are most affected
by the presence of a row of film cooling holes, due to their relatively close spacing in the
spanwise direction. Many cases showed large regions of the vane with low stresses, where
rows of film cooling holes could be placed without exceeding desired maximum hoop stress.
Stress augmentation in the spanwise direction due to rows of film cooling holes is much less
than in the hoop direction due to the wider spacing between rows than between holes in a
row.

1$6$&5² 
 6.0 - Radial Cooled Vane
Current cooling schemes rely on impingement and film cooling to achieve acceptable ma-
terial temperatures. These schemes are well suited for commercial turbofan and military
engine applications. Impingement cooling schemes rely on an insert to distribute the vane
cooling air. For small size vanes, such as for use in turboshaft engine applications, impinge-
ment cooling may not be practical. Radial cooling has been proposed for small CMC vanes,
such as would be found in turboshaft engine applications. With radial cooling the impinge-
ment scheme is replaced by using several cooling tubes, oriented in the spanwise or radial
direction. No impingement insert is required, and the diameter of the radial cooling holes
can be small, approaching the diameter of film cooling holes if necessary. Similar to film
cooling, many small diameter radial tubes, as opposed to one or two maximum diameter
tubes, facilitate heat removal from the radial tubes. However, pressure drop considerations
favor somewhat larger diameter tubes, since radial cooling tubes are much longer than film
cooling tubes. An additional reason for investigating the feasibility of radial cooling for tur-
boshaft engine applications is that as the vane size decreases the thermal resistance within
the vane material also decreases. Vane wall thickness decreases as the vane size decreases.
The decrease in vane size is generally coupled with an increase in freestream and coolant side
convective resistances. The convective resistances increase as the combustor outlet pressure
decreases, and historically the combustor outlet pressure is lower for smaller size engines.
It is easier to fabricate CMC vanes with radial cooling tubes and no film cooling, than
to fabricate CMC vanes with radial cooling tubes from which film cooling is extracted. On
the other hand, the heat flux absorbed by the air flowing through the tubes is less when film
cooling is used. Without film cooling, air flowing through the radial tubes absorbs the entire
heat flux due to the gas-to-wall temperature difference. With film cooling, air is extracted
from the tubes and is used to reduce the external heat flux by insulating the wall with a film
of cooler air. For the same coolant amount, when film cooling air is extracted from the tubes
the average tube heat transfer coefficient is lower than when no film cooling is used. Vane
thermal analyses were done for vanes cooled only by radial cooling, and for vanes cooled by
radial cooling coupled with film cooling. The configuration where trailing edge ejection is
used in conjunction with radial cooling is also analyzed.
The thermal and aerodynamic performance was determined for two vane external con-
figurations. One was a half scale version of the vane used for the full size vane structural
analysis. The other was a vane shape specified by Dr. Jerry Lang at the NASA Glenn
Research Center. Both vanes had acceptable aerodynamics. The calculated profile loss was
similar for the two vanes.
The analysis showed that neither vane could be effectively cooled using only radial cool-
ing. Film cooling or trailing edge ejection was required to be used in conjunction with radial
cooling. With film and radial cooling the aggressive future engine goals of NASA’s Large
Civil Tilt Rotor(LCTR)[10] program could be achieved. These goals are a gas temperature
of 1927◦ C(3500◦F ), and a coolant temperature of 594◦ C(1100◦F ). The gas temperature
includes a pattern factor to account for temperature variations at the combustor outlet. The
coolant temperature is significantly warmer than ambient due to a takeoff pressure of 30
atm. The thermal and aerodynamic analysis results were transmitted to Dr. Lang to be
used in his structural analysis of the LCTR vane.

1$6$&5² 
 Thermal and aerodynamic results will be given for the half scale vane of Halila et al.[8]
and for the NASA vane. For each vane results will be given with and without film cooling,
and when trailing edge ejection is used in conjunction with film cooling. Current engines
have lower combustor outlet temperatures, and the engine Overall Pressure Ratio(OPR)
is also lower. Results will be given for engine gas temperatures of 1427◦ C(2600circF ) and
1927◦ C(3500◦F ). For the lower gas temperature the analysis assumed a coolant temperature
of 376◦C(708◦ F ), which is consistent with an engine OPR of 17.. For the gas temperature of
1927◦ C(3500◦F ) coolant temperature is 594◦ C(1100◦ F ), and the engine OPR was 30. Heat
transfer coefficients changed as the gas temperature changed, primarily due to changes in
the vane inlet total pressure.
Figure 6.0.1 shows the two vane shapes. The half size vane is the vane given by Halila
et al.[8] with the dimensions reduced by a factor of 2. This vane is designated as the half
scale EEE vane. The vane shape designated as the NASA vane is the one specified by NASA
personnel. The NASA vane has a longer axial chord, but its true chord-to-axial chord ratio
is lower than the true chord-to-axial chord ratio for the half scale EEE vane. The true chord-
to-axial chord ratio for the NASA vane is 1.727. Most of the gas turning for the NASA vane
occurs in the last half of the vane, and this results in an aft loaded pressure distribution.
Figure 6.0.2 compares the pressure distributions for the two vanes. The pressure distri-
butions were calculated for the same axial chord-to-pitch(solidity) ratio. Since the NASA
vane has a longer chord than the half scale EEE vane, there would be fewer NASA vanes in
an engine with the same vane inner and outer diameters. The aft vane loading tends to de-
lay suction surface transition at low Reynolds numbers. The aft loading results in favorable
pressure gradients for more of the suction surface, and this tends to reduce boundary layer
growth.

1$6$&5² 
 1

Tangential distance, cm 0
NASA vane

-1

-2
Half scale EEE vane

-3

-4
-1 0 1 2 3 4
Axial distance, cm
Fig. 6.0.1 Comparison of vane eometries

0.9 Pressure surface

Pressure surface
Static pressure ratio, P / PT

0.8
Suction surface
0.7

Suction surface
0.6

0.5

Half scale EEE vane

0.4 NASA vane

0.3
0 0.2 0.4 0.6 0.8 1
Axial distance, x/CX

Fig. 6.0.2 Pressure distribution for both vanes

1$6$&5² 
 2500

Transitioning
Fully turbulent
2000
Nusselt number, hCX / k

Pressure surface Suction surface

1500 PT=30 atm.
PT=30 atm.
PT=17 atm.
1000

500

Pressure surface PT=17 atm.

0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.0.1 Midspan Nusselt number distributions

6.1 - Half scale EEE vane

Figure 6.1.0.1 shows four Nusselt distributions for the half scale EEE vane. These dis-
tributions are shown for vane inlet total pressures of 17 and 30 atm., and for two transition
modeling assumptions. If there is no vane film cooling, transition does not occur close to the
leading edge at these low Reynolds numbers. The Reynolds number is almost proportional
to the inlet pressure and surface distance. Both of these quantities are lower in turboshaft
engine applications than in commercial engines in the CF6 size class. The transition analysis
accounted for an inlet turbulence of 10%. Prior to the flow becoming fully turbulent, the
effective viscosity included a freestream turbulence effect. This effect is the reason that the
transitioning Nusselt numbers are higher than the fully turbulent Nusselt numbers along the
forward portion of the pressure surface. Where the suction surface heat transfer approaches
zero, at s/CX near 1.0, the boundary layer is nearly separated. This corresponds to the
adverse pressure gradient region seen in figure 6.0.2. Figure 6.1.0.1 shows that when the flow
is assumed fully turbulent, there is a heat transfer decrease in this region, but the boundary
layer does not approach separation. Data, such as Arts[32], show that flows downstream of
film cooling rows are turbulent, even at low Reynolds numbers.

1$6$&5² 
 2 radial holes 4 radial holes

Fig. 6.1.1.1 Two radial cooling hole configurations.

6.1.1 - Radial cooling tubes - no film cooling. Figure 6.1.1.1 shows the half scale
EEE vane with two and four cooling tubes. The two tube configuration has two 2.54mm(100mil)
diameter cooling tubes. In the four tube configuration the two forward tubes are also
2.54mm(100mil) in diameter, and the third and fourth tubes have a smaller diameter in
order to fit within the CMC shell. The third tube is 2.0mm(78mil) in diameter, and the
fourth tube is 1,27mm(50mil) in diameter. The shell for the half scale vane was assumed
to be 1mm(40mil) thick, and that the EBC protective layer was 0.127mm(5mil) thick. The
thermal analysis assumed that the space inside the CMC shell was filled with a CMC material
having the same 20W/m −◦ C thermal conductivity as the CMC shell.
Figure 6.1.1.2 through 6.1.1.5 show outer surface temperatures for both the CMC layer
and the EBC layer. Figures 6.1.1.2 and 6.1.1.3 show the effects of having four instead of
two radial cooling holes. These results are for a gas temperature, TG , of 2600◦ F (1700◦K),
and a total pressure of 17 atm. Even though this relatively low gas temperature is below
the desired maximum EBC temperature, CMC temperatures exceed the maximum desired
CMC temperature of 2400◦ F (1589◦K). The four tube configuration has less of the vane
seeing excessive CMC temperatures than does the two tube configuration. Aft of where
there is insufficient space for a cooling tube, heat can not be effectively removed, and CMC
temperatures are excessive.

1$6$&5² 
 2800
EBC Limit
o
TG=2600 F
2600

CMC Limit
2400

Tem[erature, F
o 2200
EBC Temp.
2000

1800
CMC Temp.

1600

1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.1.2 CMC and EBC temperatures - two tubes - P=17 atm. - TG = 2600◦F - no film
cooling

2800
EBC Limit
o
TG=2600 F
2600

CMC Limit
2400
Tem[erature, F
o

2200
EBC Temp.
2000

1800
CMC Temp.
1600

1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.1.3 CMC and EBC temperatures - four tubes - P=17 atm. - TG = 2600◦ F - no film
cooling
Figures 6.1.1.4 and 6.1.1.5 show temperatures for cases similar to that shown in figure
6.1.1.3, except that the gas temperature was increased to 3500◦F (2200◦K), and the total
pressure was increased to 30 atm. Increasing the total pressure increases both the coolant
temperature, and the internal and external heat transfer coefficients. For comparison pur-
poses the internal heat transfer coefficient, hI , was set to 5000W/m2 −◦ K in figure 6.1.1.4.
This is the same value as was used for the lower total pressure cases. Figure 6.1.1.5 shows
results where the internal heat transfer coefficient is increased to 8000W/m2 −◦ K.

1$6$&5² 
 3600
o
3400 TG=3500 F

3200
EBC Temp.

Tem[erature, F
3000

o
CMC Temp.
2800
EBC Limit
2600
CMC Limit
2400

2200

2000
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.1.4 CMC and EBC temperatures - four tubes - P=30 atm. - TG = 3500◦F -
hI = 5000W/m2 −◦ K - no film cooling

3600
o
3400 TG=3500 F

EBC Temp.
3200
Tem[erature, F

3000
o

2800
EBC Limit
2600
CMC Limit
2400

2200 CMC Temp.

2000
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.1.5 CMC and EBC temperatures - four tubes - P=30 atm. - TG = 3500◦F -
hI = 8000W/m2 −◦ K - no film cooling

These two figures show that, even away from the trailing edge region, radial cooling
alone is insufficient to maintain EBC and CMC temperature at acceptable values. Away
from the trailing edge region EBC temperatures are unacceptably high. While a higher
internal heat transfer coefficient allows acceptable CMC temperatures in the forward part
of the vane, EBC temperatures are unacceptably high. Reducing the EBC thickness below
0.127mm(5mil) would lower the maximum EBC temperatures. However if this was done
to lower maximum EBC temperature to acceptable levels, the maximum CMC temperature
would be unacceptable.

1$6$&5² 
 Figures 6.1.1.4 and 6.1.1.5 illustrate the importance of the internal film cooling heat
transfer coefficient. Because the radial tubes have a small diameter, they may not be hy-
draulically smooth. Tube heat transfer coefficients increase as surface roughness increases.
The penalty for increased surface roughness is increased tube pressure losses. Kays and
Crawford[26] discuss smooth and rough tube heat transfer rates in terms of Nusselt num-
bers, Nu. Nu = hD/k, where h is the tube heat transfer coefficient, D is is the tube
diameter, and k is the thermal conductivity. For a smooth tube with fully developed flow
Nu is given by:

Nu = 0.022P r 0.5Re0.8
P r is the Prandtl number, and Re is the tube Reynolds number. It is convenient to express
the Reynolds number in terms of the tube Mach number, M.
P 
Re = M γRT D
RT μ
where P is pressure, R is the gas constant, T is the absolute temperature, μ is the dynamic
viscosity, and γ is the specific heat ratio. μ and k are both proportional to T n , where n is
approximately 0.7. Consequently Nu is almost independent of temperature, but h increases
somewhat with temperature. Also, h increases slightly at the diameter, D, decreases. For P
equal to 17 atm., T equal to 650◦ K(710◦F ), M equal to 0.3, and D equal to 2.54mm(100mil),
Nu is 196. The tube heat transfer coefficient, h is 3725W/m2◦K.
When tubes are not hydraulically smooth, roughness can increase tube heat transfer by
up to a factor of 2.5(Kays and Crawford[26]). They also state that this increase comes at
the expense of a four fold increase in tube friction factor, and thus a four fold increase in
tube pressure drop.
The expected length-to-diameter,(L/D), ratios for radial cooling tubes is between 10
and 20. Even at an L/D = 20, the average smooth tube Nu will be somewhat greater than
the Nu for fully developed flow. This is due to an entrance effect, and the degree of heat
transfer augmentation is highly dependent on the specific entrance configuration. Overall,
the consequences of possible tube roughness and entrance effects indicate that a tube heat
transfer coefficient, hI of 5000W/m2 −◦ K for a pressure of 17 atm., and 8000W/m2 −◦ K for
a pressure of 30 atm. is conservative. However, hI will be lower when coolant is extracted
from the tubes for either film cooling or trailing edge ejection.
The amount of cooling flowing through the tubes was calculated to be approximately 7%
for the four tube configuration. This calculation assumed a coolant-to-inlet Mach number
ratio of 2, an inlet-to-coolant temperature ratio of 2.25, a span-to-axial chord ratio of 1.5,
and a vane pitch-to-axial chord ratio of 1.4.

1$6$&5² 
 6.1.2 - Radial cooling & film cooling. It was shown by Arts[32] that just the pres-
ence of film cooling holes causes laminar boundary layers to become turbulent. When a
laminar boundary layer is tripped and becomes turbulent there is a rapid increase in the
heat transfer coefficient. If the flow upstream of a cooling row is already turbulent, there is
only a small increase in heat transfer coefficients downstream of the film cooling row. Heat
transfer in the leading edge region is always laminar like because of strong favorable pressure
gradients, which increase as the leading edge diameter decreases. In this region heat transfer
coefficients increase dramatically in the presence of film cooling rows as shown by Reiss and
Bolcs[18], and by Ou and River[33]. Augmentation factors as great as a factor of four have
been reported. However, augmentation factors are relative to a solid leading edge with low to
moderate freestream turbulence. The high turbulence at the combustor outlet is expected to
increase solid vane leading edge region heat transfer coefficients by 50% or more. However,
because of low Reynolds numbers associated with vanes for turboshaft engine applications,
turbulence alone will not cause transition to turbulent flow in this region.
Figure 6.1.2.1 shows CMC and EBC temperatures for the vane with two assumptions
regarding transition in the leading edge region. Figure 6.1.2.2 shows the Nusselt number dis-
tributions corresponding to these two assumptions, and also shows results for a transitioning
flow model with a vane inlet turbulence intensity of 10%. The first assumption is that flows
are fully turbulent, and the second delays transition to locations downstream of the leading
edge. The second assumption was made to better account for the heat transfer augmentation
factors that are present in the Reiss and Bolcs[18] leading edge film cooling model. Figure
6.1.2.2 shows significantly different leading edge heat transfer coefficients between the fully
turbulent and delayed transition models. However, figure 6.2.2.1 shows that the effect of
these two model assumptions on both EBC and CMC temperature is relatively small. With
film cooling CMC temperatures are acceptable throughout the entire vane. However, EBC
temperatures are excessive in the leading edge region. Temperatures may be over estimated
because any heat transfer within the CMC vane to the cooling holes was neglected. Leading
edge region film holes are closely spaced, and are relatively long because these holes are
steeply angled in the spanwise direction. An indication that heat transfer into the film cool-
ing holes should be considered in a future analysis is that the temperatures in the leading
edge region were nearly the same with and without film cooling.
Only three rows of film cooling holes were used in addition to the leading edge film
cooling. Two rows were on the pressure surface, and one on the suction surface. The coolant
fraction was calculated to be 16% of mainstream flow. Only three rows of film cooling holes
were sufficient because the minimum cooling hole size is independent of the vane size. The
smaller the vane, the fewer rows of film cooling holes are required.

1$6$&5² 
 6.1.3 - Single cavity & trailing edge ejection. Film cooling is very effective away
from the vane leading edge region. Nusselt numbers are lower in the leading edge region if
there is no film cooling. Trailing edge ejection was effective in the trailing edge region. A
thermal analysis was done for a solid vane with trailing edge ejection. At most there would
be room for only a single cavity in the half scale EEE vane. The configuration analyzed
here is geometrically similar to the single cavity configuration discussed in section 5.2.4.
The configuration analyzed has no film cooling. The internal heat transfer coefficients will
vary significantly, depending on the internal cooling arrangement. For comparison purposes
the internal heat transfer coefficient was assumed to be 5000W/m2◦ K. The trailing edge
ejection tube heat transfer coefficient was determined based on the assumed inlet-to-exit
pressure ratio.
Figure 6.1.3.1 compares CMC and EBC outer surface temperatures for the four tube
radial cooling configuration with those for the single cavity with trailing edge ejection tube
configuration. Neither configuration employs film cooling. The single cavity configuration
has acceptable CMC temperatures over much of the vane surface. However, on both the
suction and pressure surfaces CMC temperatures exceed the temperature limit by more
than 180◦ F (100◦ C). EBC temperatures exceed the desired maximum EBC temperature over
much of the vane surface. The single cavity configuration yields lower EBC temperatures
than does the four radial tube configuration.
Figure 6.1.3.2 shows that decreasing the pitch-to-diameter ratio of the trailing edge ejec-
tion tube is an effective means of maintaining CMC temperatures less than the maximum
allowable CMC temperature. Unfortunately, decreasing the pitch-to-diameter ratio has little
effect on the EBC outer surface temperatures.
The outer surface temperature of the EBC can be lowered by reducing the thickness of the
EBC coating. However, reducing the EBC thickness raises the outer surface temperature
of the CMC vane. Figure 6.1.3.3 shows the effects of varying the EBC thickness. For
illustration purposes the EBC coating has a uniform thickness. These results were calculated
for a trailing edge ejection tube pitch-to-diameter ratio of 1.67. Unfortunately, near the aft
portion of the vane, where EBC temperatures are most in excess of the EBC temperature
limit, reducing the EBC thickness does not significantly reduce EBC temperatures. In the
leading edge region EBC temperatures also exceed the EBC temperature limit. In this region
reducing the EBC thickness is somewhat more effective than in the trailing edge region.

1$6$&5² 
 3600
o
3400 Fully Turbulent TG=3500 F
3200 Delayed Forced Transition

3000
EBC Temp.
Tem[erature, F

2800 EBC Limit

o

2600
CMC Limit
2400
2200
2000
1800
CMC Temp.
1600
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.2.1 CMC and EBC temperatures four tubes - P=30 atm. - TG = 3500◦ F - hI =
5000W/m2 −◦ K - film cooled

2500

Transitioning
Fully Turbulent
2000 Delayed Forced Transition
Nusselt number, hCX / k

Delayed Forced
1500 Transition
Transitioning
Fully Turbulent
1000

500

Pressure surface Suction surface

0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.2.2 Midspan Nusselt number distributions - P=30 atm.

1$6$&5² 
 3600
o
3400 TG=3500 F
EBC Temp.
3200
3000
Tem[erature, F

2800 EBC Limit

o

2600
2400 CMC Limit

2200
2000
CMC Temp.
1800
4 Radial Cooling Tubes
1600 Single Cavity with Trailing Edge Ejection
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.3.1 Comparison of radial tube and single cavity CMC and EBC temperatures -
P=30 atm. - TG = 3500◦F - hI = 5000W/m2 −◦ K

3600
o
3400 Suction surface TG=3500 F Pressure surface
EBC Temp.
3200
3000
Tem[erature, F

2800 EBC Limit

o

2600
2400 CMC Limit

2200
2000
CMC Temp.
1800 Tube pitch = 5.0D
Tube pitch = 2.5D
1600
Tube pitch = 1.67D
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.3.2 Effect of trailing edge ejection tube pitch-to-diameter ratio on CMC and EBC
temperatures - P=30 atm.

1$6$&5² 
 3600
o
3400 Suction surface TG=3500 F Pressure surface
3200
EBC Temp.
3000

Tem[erature, F
2800 EBC Limit
o
2600
2400 CMC Limit

2200
2000
tEBC = 10 mil CMC Temp.
1800
tEBC = 5 mil
1600 tEBC = 2.5 mil
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.1.3.3 Effect of EBC thickness on CMC and EBC temperatures - P=30 atm.

1$6$&5² 
 2500

Transitioning
Fully turbulent
2000
Nusselt number, hCX / k

PT=30 atm.
1500
PT=30 atm. PT=17 atm.
1000

PT=17 atm.
500

Pressure surface Suction surface

0
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Surface distance, s/CX

Fig. 6.2.0.1 Midspan Nusselt number distributions - NASA vane

6.2 - NASA vane

Figure 6.2.0.1 shows Nusselt number distributions for the NASA vane. The profile and
pressure distributions for this vane were shown in figures 6.0.1 and 6.0.2. The Nusselt number
distributions for this vane can be compared with those shown in figure 6.1.0.1 for the half
scale EEE vane. The higher maximum Nusselt numbers for the NASA vane compared with
the maximum Nusselt numbers for the half scale vane do not represent higher heat transfer
coefficients. As shown in figure 6.0.1, the NASA vane has a longer axial chord, CX , and h
is proportional to Nu divided by CX . Suction surface transition is very different between
the NASA vane and the half scale EEE vane. Figure 6.1.0.1 for the half scale EEE vane
showed that increasing the inlet total pressure from 17 to 30 atm. caused suction surface to
move forward from 2/3 to 1/2 of the surface distance. Figure 6.2.0.1 shows that the NASA
vane at a pressure of 17 atm. has suction and pressure surface transition occurring close
to the trailing edge. At 30 atm., however, suction surface transition is relatively close to
the leading edge, and after transition the maximum Nusselt number exceeds the maximum
fully turbulent Nusselt number. Figure 6.0.2 shows that the NASA vane has a less favorable
suction surface pressure gradient. Consequently, at a pressure of 30 atm. suction surface
transition occurs further forward on the NASA vane compared with the half scale EEE vane.

1$6$&5² 
 0.01 0.01

0 0
Tangential distance, m

Tangential distance, m
-0.01 -0.01

-0.02 -0.02

-0.03 -0.03

-0.04 -0.04
-0.01 0 0.01 0.02 0.03 0.04 -0.01 0 0.01 0.02 0.03 0.04
Axial distance, m Axial distance, m

2 radial holes 4 radial holes

Fig. 6.2.1.1 NASA vane with two radial cooling hole configurations.

6.2.1 - Radial cooling tubes - no film cooling. Figure 6.2.1.1 shows two and four
hole configurations for the NASA vane. Since cooling of the trailing edge is an issue, the two
radial tube configuration has the aft tube further aft than the two hole configuration of the
half scale EEE vane shown in figure 6.1.1.1. The radial cooling tubes are 2.54mm(100 mil)
in diameter.
Figure 6.2.1.2 shows outer EBC and CMC temperatures for the two and four radial tube
configurations. The results are for a gas temperature of 1700◦K(2600◦ F ), and an EBC thick-
ness of 0.127mm(5 mil). Temperatures were calculated for a heat transfer coefficient within
the tubes of 5000W/m2◦ K, which at a pressure of 17 atm. may be somewhat optimistic.
Even at this relatively low gas temperature and external heat transfer coefficients that are
mostly laminar, CMC temperatures exceed the design temperature for N24A towards the
rear of the vane. Unless the CMC temperature limit is raised, radial cooling without film
cooling is insufficient at this low gas temperature.
Figure 6.2.1.3 shows EBC and CMC temperatures for a case with four radial cooling
tubes. Here the gas temperature is 2200◦ K(3500◦F ), and the inlet total pressure is 30
atm. Increasing the inlet total pressure from 17 to 30 atm. causes the suction surface heat
transfer to change from mostly laminar to mostly turbulent. There is also an increase in
coolant temperature with an increase in inlet total pressure. Temperatures are shown for
two values of the tube internal heat transfer coefficient, hI . The lower value of 5000W/m2 −◦
K is conservative at this pressure, since hI increases with pressure. Going from 5000 to
8000W/m2 −◦ K represents the expected change in hI due to the total pressure increasing from
17 to 30 atm. The EBC temperatures exceed the EBC temperature limit of 1756◦ K(2700◦F )
almost everywhere, even for the higher hI value. Along the entire suction surface CMC
temperatures exceed their limit, as they do for the majority of the pressure surface.

1$6$&5² 
 2800
EBC Limit
o
TG=2600 F
2600

CMC Limit
2400
Tem[erature, F
o

EBC Temp.
2200

2000

1800

Pressure surface Suction surface

1600
Fore & Aft Cooling Tubes CMC Temp.
4 Cooling Tubes
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Surface distance, s/CX

Fig. 6.2.1.2 Effect of number of cooling tubes on CMC and EBC temperatures - P=17 atm.

3600
o
3400 TG=3500 F

3200
EBC Temp.
Tem[erature, F

3000
o

2800
EBC Limit
2600
CMC Temp.
CMC Limit
2400 2o
hI = 5000 W/m K
2o
2200 hI = 8000 W/m K
Suction surface
Pressure surface
2000
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Surface distance, s/CX

Fig. 6.2.1.3 Effect of tube heat transfer coefficient, hI , on CMC and EBC temperatures -
P=30 atm.

1$6$&5² 
 3600
3400 o
TG=3500 F
3200
Pressure surface Suction surface
3000
EBC Temp.

Tem[erature, F
2800 EBC Limit
o
2600
CMC Limit
2400
2200
2000
1800 CMC Temp.
2o
hI = 3000 W/m K
1600 2o
hI = 5000 W/m K
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Surface distance, s/CX

Fig. 6.2.2.1 Effect of tube heat transfer coefficient, hI , on CMC and EBC temperatures -
with film cooling - P=30 atm.

6.2.2 - Radial cooling & film cooling. EBC and CMC outer surface temperatures
for the four radial cooling tube configuration with film cooling are shown in figure 6.2.2.1.
These results are for a gas temperature of 2200◦ K(3500◦K), and an inlet total pressure of 30
atm. In addition to leading edge film cooling there are three film cooling rows on the pressure
surface, and one row on the suction surface. Tube heat transfer coefficients are lower when
film cooling air is extracted from the tube. Therefore, results are shown for lower tube heat
transfer coefficients of 3000 and 5000W/m2 −◦ K. The effects of changing hI are relatively
small. However, the higher value for hI is sufficient to maintain the CMC suction surface
temperatures near the leading edge within the CMC limit temperature.

1$6$&5² 
 3600
3400 o
Pressure surface TG=3500 F Suction surface
3200
3000
EBC Temp.
Tem[erature, F

2800 EBC Limit

o

2600
CMC Limit
2400
2200
2000 CMC Temp.

1800
Trailing edge ejection tube pitch = 5D
1600 Trailing edge ejection tube pitch = 1.67D
1400
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2
Surface distance, s/CX

Fig. 6.2.3.1 Effect of trailing edge ejection tube heat transfer coefficient, hI , on CMC and
EBC temperatures - no film cooling - P=30 atm.

6.2.3 - Single cavity & trailing edge ejection. Figure 6.2.3.1 shows temperatures for
a single cavity configuration similar to the configurations discussed in sections 5.2.4 and 6.2.3.
This figure shows that acceptable CMC temperatures can be achieved with a single cavity
configurations as long as there is trailing edge ejection. Except at the rear of the pressure
surface, CMC temperatures are less than the CMC limit temperature of 1583◦ K(2400◦F ).
Previous results showed that only a small amount of pressure surface film cooling would
reduce CMC temperatures at the rear of this surface to acceptable values.
The difference in temperatures between results with a trailing edge ejection spanwise
pitch-to-tube diameter,D, ratio of 1.67 and 5 shows the sensitivity of temperatures to the
amount of heat removed via trailing edge ejection. The average heat transfer coefficient for
the trailing edge ejection tube was calculated from the correlation for fully developed flow
in a tube. The hydraulic diameter of the cavity is much larger than the hydraulic diameter
of the radial cooling tubes, and the average inter heat transfer coefficient decreases as the
hydraulic diameter decreases. Fortunately, the decrease is not linear, but only to the 0.2
power, because the radial flow is turbulent in either the cavity or radial tubes. While the
assumed value of 5000W/m2◦ K may be somewhat optimistic, it is not excessively so. CMC
temperatures are acceptable, except near the trailing edge. However, EBC temperatures
exceed the EBC temperature limit in the forward part of the suction surface as well as the
aft portion of the pressure surface.

1$6$&5² 
 6.3 - Conclusions

Results were similar for both vanes analyzed for turboshaft engine applications. Pressure
distributions, and consequently the heat transfer distributions, were different between the
two vane geometries. Even so, both vane geometries required film cooling or trailing edge
ejection to achieve acceptable temperature limits. Radial cooling alone provided insufficient
cooling to avoid excessive CMC and EBC temperatures, even when the gas temperature was
only 1700◦ K(2600◦F ).
When the gas temperature was 2200◦K(3500◦ F ) film cooling and trailing edge ejection
provided sufficient cooling to keep CMC temperatures generally below 1589◦ K(2400◦F ).
Even when the EBC thickness was 0.127mm(5 mil), EBC temperatures were generally in
excess of their limit of 1756◦ K(2700◦F ) at locations where the CMC temperatures were
close to their limit value. Either raising the EBC temperature limit or increasing EBC
conductivity would help to alleviate this problem. The EBC thermal conductivity was
assumed to be 1W/m − K.

1$6$&5² 
 7.0 - References
1. Hartsel, J.E., 1972, “Prediction of Effects of Cooling Mass Transfer on the Blade Row
Efficiency of Turbine Airfoils,” AIAA paper 72-11.
2. Verrilli, M., Calamino,, A., Robinson, R.C., and Thomas, D.J.,2004, “Ceramic Matrix
Composite Vane Subelement Testing in a Gas Turbine Environment, ASME Paper GT2004-
53970.
3. Vedula, V., Shi, J., Jarmon, D., Ochs, S., Oni, L., Lawton, T., Green, K., Prill, L., Schaff,
J., and Linsey, G., 2005, “Ceramic Matrix Composite Vanes for Gas Turbine Engines, ASME
Paper GT2005-68229
4. Brewer, D.N., Verrilli, M., and Calomino, A., 2006, “Ceramic Matrix Composite Vane
Subelement Burst Testing”, ASME paper GT2006-90883.
5. Watanabe, K.-I., Suzumura, N., Nakamura, T., Murata, H., Araki, T., and Natsumura,
T., 2008, “Development of CMC Vane for Gas Turbine Engine”, Ceramic Engineering and
Science Proceedings, Volume 24, Issue 4, (eds W. M. Kriven and H.-T. Lin), John Wiley &
Sons, Inc., Hoboken, NJ, USA. doi: 10.1002/9780470294826.ch87
6. Nakamura, T., Murata, H., Takahashi, A., and Okita, Y., 2010, “Development of a CMC
Turbine Vane”, in Proceedings of the 7th International Conference on High Temperature
CMCs(HT CMC7), pp. 559-565.
7. Marshall, D.B., and Cox, B.N., 2008, “Integral Textile Ceramic Structures,” Annual
Review of Materials Research, Vol. 38, pp. 425-443.
8. Halila, E.E., Lenahan, D.T., and Thomas, T.T., 1982, “High Pressure Turbine Test
Hardware Detailed Design Report, NASA CR-167955, General Electric Company report
R81AEG284
9. Mital, S.K., Bednarcyk, B.A., Arnold S.A., and Lang, J., 2009, “Modeling of Melt-
Infiltrated SiC/SiC Composite Properties,” NASA TM-2009-215806.
10. Snyder, C.A., and Thurman, D.R., 2010, “Effects of Gas Turbine Component Perfor-
mance on Engine and Rotary Wing Vehicle Size and Performance,” NASA TM 2010-216907.
11. Boyle, R., Miller, I., Halbig, M., DiCarlo, J., Bhatt, R., and Jaskowiak, M, 2012, “Design
Concepts for Cooled Ceramic Matrix Composite Turbine Vanes,” 36th Annual Conference
on Composites, Materials and Structures, Cocoa Beach, FL
12. Boyle,, R., Parikh, A., Nagpal V., Halbig, M., DiCarlo, J., and Bhatt, R., 2013, “Design
Concepts for Cooled Ceramic Matrix Composite Turbine Vanes,” 37th Annual Conference
on Composites, Materials and Structures, Cocoa Beach, FL
13. Boyle, R.J., Parikh, A.H., Halbig, M.C., and Nagpal, V.K., 2013, “Design Considerations
for Ceramic Matrix Composite Vanes for High Pressure Turbine Applications,” ASME paper
GT2013-95104
14. Wilcock, R.C. Young, J.B., and Horlock, J.H., 2005. “The Effects of Turbine Blade
Cooling on the Cycle Efficiency of Gas Turbine Power Cycles,” ASME Journal of Engineering

1$6$&5² 
for Gas Turbines and Power, Vol. 127, pp 109-120.
15. www.specialmetals.com
16. DiCarlo, J.A,, Yun, H.-M., Morsher, G.N., and Bhatt, R.T., 2004, “SiC/SiC Composites
for 1200 ◦ C and Above,” NASA TM-2004-213048
17. Lee, K.N., Fox, D.S., Eldridge, J.I., Zhu, D., Robinson, R.C., Bansal, N.P., and Miller,
R.A., 2002, “Upper Temperature Limit of Environmental Barrier Coatings Based on Mullite
and BSAS,” NASA TM-2002-211372
18. Reiss, H., and Bolcs, A., 1999, “Experimental Study of Showerhead Cooling on a
Cylinder Comparing Several Configurations Using Cylindrical and Shaped Holes,” ASME
paper 99-GT-123.
19. Boyle, R.J., and Ameri, A.A., 2010, “A Correlation Approach to Predicting Film Cooled
Turbine Vane Heat Transfer,” ASME paper GT2010-23597.
20. Tacina, R., Wey, C., Laing, P., and Mansour, A., 2002, “A Low NOx Lean-Direct
Injection, Multipoint Integrated Module Combustor Concept for Advanced Aircraft Gas
Turbines,” NASA TM-2002-211347, Presented at the Conference on Technologies and Com-
bustion for a Clean Environment, Oporto, Portugal, July 2001.
21. Harth, S., Zarzalis, N., Bauer, H.-J., and Turrini, F., 2013, “Evaluation of a Piloted
Lean Injection System in Terms of Emission Performance and Flame Structure at Elevated
Pressure,” ASME paper GT2013-94371.
22. Dusing, K.M., Ciani, A., Benz, U., Eroglu, A., and Knapp, K., 2013, “Development of
GT24 And GT26 (Upgrades 2011) Reheat Combustors, Achieving Reduced Emissions and
Increased Fuel Flexibility,” ASME paper GT2013-95437.
23. Boyle, R.J., and Jones, S.M., 2009, “Effects of Precooling Turbine Cooling Air on Engine
Performance,” ASME paper GT-2009-60120
24 Anon., 2007, “Performance Prediction and Simulation of Gas Turbine Engine Operation
for Aircraft, Marine, Vehicular, and Power Generation,” NATO RTO Technical Report ATV-
036.
25. Engel, T., 2013, “High-Temperature Interlaminar Tension Test Method Development
for Ceramic Matrix Composites,” ASME paper GT2013-94095.
26. Giel P. W., VanFossen, G. J., Boyle, R. J., Thurman, D. R., and Civinskas, K. C.,
1999, “Blade Heat Transfer Measurements and Predictions in a Transonic Turbine Cascade”,
ASME 99-GT-125, June, 1999, also NASA/TM-1999-209296.
27. Boyle, R. J. and Jackson, R., 1997, “Heat Transfer Predictions for Two Turbine Nozzle
Geometries at High Reynolds and Mach Number”, ASME Journal of Turbomachinery, vol.
119, no. 2, pp. 270-283.
28. Florschuetz, L.W., Truman, C.R., and Metzger, D.E., 1981, “Streamwise Flow and Heat
Transfer Distributions for Jet Array Impingement with Crossflow,” Journal of Heat Transfer,

1$6$&5² 
Vol. 103, pp. 337-342.
29. Kays, W.M., and Crawford, M.E., 1980, Convective Heat and Mass Transfer,, 2nd ed.
McGraw-Hill Book Co. New York, NY.
30. Anon., 2009, “ANSYS Mechanical APDL, Release 12.1,”
31. Brauer, J.R., ed., 1993, it What Every Engineer Should Know About Finite Element
Analysis, Marcel Decker Inc. New York, NY.
32. Arts, T., 1995, “Thermal Investigation of a Highly Loaded Transonic Turbine Film
Cooled Guide Vane”, 1st European Conf. on Turbomachinery - Fluid Dynamic and Ther-
modynamic Aspects, Erlanger, Germany, also VKI preprint 1995-11.
33. Ou, S. and River, R., 2001, “90◦ Skew Leading Edge Film Cooling Effectiveness, Heat
Transfer, and Discharge Coefficients for Cylindrical Film Holes at High Free Stream Tur-
bulence,” NATO pub. RTO-MP-069(I), presented at RTO-ATV Symposium on “Advanced
Flow Management: Part B Heat Transfer and Cooling in Propulsion and Power Systems,
held in Loen, Norway, 7-11 May 2001,

1$6$&5²