527710

AEROTHERMODYNAMICS OF TURBINE BLADE TRAILING EDGE
COOLING
A THESIS SUBMITTED TO
THE GRADUATE SCHOOL OF NATURAL AND APPLIED SCIENCES
OF
MIDDLE EAST TECHNICAL UNIVERSITY
BY
TUĞBA TUNÇEL
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR
THE DEGREE OF MASTER OF SCIENCE
IN
AEROSPACE ENGINEERING
SEPTEMBER 2018
Approval of the thesis:

COOLING
submitted by TUĞBA TUNÇEL in partial fulfillment of the requirements for the

degree of Master of Science in Aerospace Engineering Department, Middle
East Technical University by,
Prof. Dr. Halil Kalıpçılar

Dean, Graduate School of Natural and Applied Sciences
Prof. Dr. Ozan Tekinalp

Head of Department, Aerospace Engineering
Asst. Prof. Dr. Harika Senem Kahveci

Supervisor, Aerospace Engineering Department, METU
Examining Committee Members:
Prof. Dr. Mehmet Cevdet Çelenligil

Aerospace Engineering, METU
Asst. Prof. Dr. Harika Senem Kahveci

Assoc. Prof. Dr. Oğuz Uzol

Assoc. Prof. Dr. Metin Yavuz

Mechanical Engineering, METU
Asst. Prof. Dr. Sıtkı Uslu

Mechanical Engineering, TOBB ETU
Date:
I hereby declare that all information in this document has been obtained and
presented in accordance with academic rules and ethical conduct. I also declare
that, as required by these rules and conduct, I have fully cited and referenced all
material and results that are not original to this work.
Name, Last Name: TUĞBA TUNÇEL
Signature :
iv
ABSTRACT

COOLING
TUNÇEL, TUĞBA
M.S., Department of Aerospace Engineering
Supervisor : Asst. Prof. Dr. Harika Senem Kahveci
SEPTEMBER 2018, 81 pages
It is known that the thermal efficiency of gas turbines strongly depends on the tur-
bine entry temperature of the working fluid. This has resulted in increased turbine
working temperatures, and peak temperatures in advanced gas turbines have been
well above maximum allowable metal temperatures for quite some time. For turbine
blades to survive while operating beyond these material temperature limits, internal
and external cooling techniques have been developed. Due to structural and aerody-
namic restrictions, improving trailing-edge cooling methods creates a challenge for
the designers. In modern turbine blades, pressure side cutbacks with film cooling
slots stiffened with lands and pin fins embedded in passages are used to cool trailing
edges. In literature, thermal improvements obtained by slots, lands and similar in-
ternal structures have been investigated in detail since the main purpose has been to
promote cooling. But, when the performance of a gas turbine is considered, aerody-
namic enhancements are as important as thermal performance. Regarding that, this
thesis focuses on both aerodynamic and thermal aspects of a turbine blade trailing-
edge section cooling. The internal structure studied consists of staggered arrays of
v
pins, and lands and airfoil-shaped blockages in front of the trailing edge slots right
at the exit. The pins used are of cylindrical, elliptical, and airfoil shape, and have
different sizes. A study using Computational Fluid Dynamics (CFD) was performed
to investigate the flow structure and heat transfer both inside the passage and out-
side in the vicinity of trailing-edge slots. With the goal of choosing an optimal pin fin
configuration that is aerothermodynamically more advantageous for slot film cooling,
this thesis provides a thorough investigation that would be of interest to the turbine
designers.
Keywords: Turbine Blade Cooling,Trailing Edge,Pin Fin Cooling, Film Cooling,

Computational Fluid Dynamics
vi
ÖZ
TÜRBİN KANADI FİRAR KENARI SOĞUTMASININ

AEROTERMODİNAMİĞİ
TUNÇEL, TUĞBA
Yüksek Lisans, Havacılık ve Uzay Mühendisliği Bölümü
Tez Yöneticisi : Dr. Öğr. Üyesi Harika Senem Kahveci
Eylül 2018 , 81 sayfa
Gaz türbinlerinin termal verimliliğinin, çalışma akışkanının türbin giriş sıcaklığına

bağlı olduğu bilinmektedir. Bu durum türbin çalışma sıcaklıklarında artışa ve ge-
lişmiş gaz türbinlerindeki zirve sıcaklıklarının metallerin dayanabileceği maksimum
sıcaklıkları aşmasına neden olmuştur. Türbin kanatlarının malzeme sıcaklık sınırla-
rının ötesinde çalışabilmesi için iç ve dış soğutma teknikleri geliştirilmiştir. Yapısal
ve aerodinamik kısıtlamalar yüzünden firar kenarı soğutma yöntemlerinin geliştiril-
mesi tasarımcılar için zorlu bir görev oluşturmaktadır. Modern türbin kanatlarının
firar kenarlarının soğutulmasında, film soğutması oluklarının bulunduğu kesilmiş ba-
sınç kenarları ve pin fin yapıları kullanılmaktadır. Literatürdeki çalışmalarda oluk,
ada ve benzer kanat içi yapılar sayesinde elde edilebilecek potansiyel termal iyileş-
tirmeler, soğutmanın daha iyi yapılabilmesi amacıyla şu ana kadar detaylı bir şekilde
incelenmiştir. Ancak, gaz türbinlerinin tüm performansı düşünüldüğünde, bu bölge-
deki aerodinamik iyileştirmeler termal iyileştirmeler kadar önem teşkil etmektedir.
Bu durumu göz önünde bulundurarak bu tez, firar kenarı soğutma yapısını aerotermal
açıdan incelemektedir. Çalışılan iç yapı saptırılmış pin dizinleri, adalar ve olukların
vii
önünde bulunan kanat şeklindeki tıkaçlardan oluşmaktadır. Kullanılan pinler dairesel,
eliptik ve kanat şeklindedir. Eliptik ve kanat şeklindeki pinlerin farklı boyutlardaki
versiyonları da bulunmaktadır. Pinlerin akış yapıları ve ısı transferi performanslarını
incelemek için kanal içinde ve firar kenarı yakınlarında Hesaplamalı Akışkanlar Di-
namiği (HAD) kullanan bir çalışma yapılmıştır. Bu tez oluklu film soğutması için
aerotermodinamik açıdan optimum pin şeklini seçmek amacında olan tasarımcılar
için detaylı bir çalışma sunmaktadır.
Anahtar Kelimeler: Türbin Kanat Soğutması, Firar Kenarı, Pin Fin Soğutma, Film
Soğutma, Hesaplamalı Akışkanlar Dinamiği
viii
To My Beloved Ones
ix
ACKNOWLEDGMENTS
First and foremost, I would like to express my most deepest gratitude and thanks
to my supervisor Asst. Prof. Dr. Harika S. Kahveci for her guidance, support and
limitless help she provided me during my academic researches. I will always consider
myself very lucky to be her first graduate student.
I would also like to thank my committee members Prof. Dr. Mehmet Cevdet Çelen-
ligil, Assoc. Prof. Dr. Oğuz Uzol, Assoc. Prof. Dr. Metin Yavuz and Asst. Prof. Dr.
Sıtkı Uslu for their valuable suggestions and insights about this work.
I am so grateful to my friends Metehan Yayla, Ezgi Orbay Akcengiz, Ayşe Usta

Yayla, Alper Açıcı, Berk Gür, Zeynep Ünal, Ayşe Hazal Altuğ Yalçın, Özgür Yalçın,
Özcan Yırtıcı, Ahmet Arda Akay, Tansu Sevine, Derya Kaya, Engin Leblebici, Ali
Büyükkoçak ,and Levent Şahin for helping me survive all the stress and not letting
me give up.
My special thanks to Çağla Ceren Ceyhan,and Melek Dicle Özcan for their invaluable
support, and the moments shared. My special thanks are extended to my oldest friend
Elif Yağmur Tatkan for always being there for me and to Selva Sargın Güçlü for her
cheering and companion. Also I would like to thank my special friend Mehmetcan
Demir for his endless love, support and patience.
Finally, I would like to express my deepest gratitude to my mother, my father and my

brother for their never ending support. Without them, I would not have achieved this
thesis like many of other things that I have in my life.
x
TABLE OF CONTENTS
ABSTRACT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v
ÖZ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . x
TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
LIST OF ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . xix
CHAPTERS
1 INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Objective . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . 5
2 LITERATURE SURVEY . . . . . . . . . . . . . . . . . . . . . . . . 7
2.1 Internal Cooling . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 External Cooling . . . . . . . . . . . . . . . . . . . . . . . . 10
3 METHODOLOGY . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
3.1 Mathematical Modeling . . . . . . . . . . . . . . . . . . . . 15
3.2 Numerical Modeling . . . . . . . . . . . . . . . . . . . . . . 16
xi
3.3 Validation Case . . . . . . . . . . . . . . . . . . . . . . . . 19
3.3.1 Internal Cooling . . . . . . . . . . . . . . . . . . . 21
3.3.1.1 Boundary Conditions . . . . . . . . . 21
3.3.1.2 Geometry and Meshing . . . . . . . . 22
3.3.1.3 Solution Convergence . . . . . . . . . 25
3.3.2 External Cooling . . . . . . . . . . . . . . . . . . 26
3.4 Comparison Cases . . . . . . . . . . . . . . . . . . . . . . . 30
3.4.1 Internal Cooling . . . . . . . . . . . . . . . . . . . 30
3.4.2 External Cooling . . . . . . . . . . . . . . . . . . 35
4 INTERNAL COOLING ANALYSIS & RESULTS . . . . . . . . . . 41
4.1 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.2 Aerodynamic Results From Comparison Cases . . . . . . . . 43
4.3 Thermal Results From Comparison Cases . . . . . . . . . . 50
5 EXTERNAL COOLING ANALYSIS & RESULTS . . . . . . . . . . 55
xii
5.1 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
5.2 Aerodynamic Results From Comparison Cases . . . . . . . . 56
5.3 Thermal Results From Comparison Cases . . . . . . . . . . 65
6 CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
xiii
LIST OF TABLES
TABLES
Table 3.1 Experimental Conditions From Ref [1] . . . . . . . . . . . . . . . . 22
Table 3.2 External Flow Experimental Conditions From Ref [1] . . . . . . . . 26
Table 3.3 Internal Flow Comparison Case Boundary Conditions . . . . . . . 31
Table 3.4 Internal Comparison Case Model Dimensions . . . . . . . . . . . . 32
Table 3.5 Mass Flow Averaged Slot Static Temperatures For Different Pin
Shapes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
xiv
LIST OF FIGURES
FIGURES
Figure 1.1 Variation Of TET And Cooling Methods Over The Years [2] . . . . 2
Figure 1.2 Cooling Methods On A Turbine Blade [2] . . . . . . . . . . . . . . 3
Figure 3.1 Schematic Of Experimental Model [3] . . . . . . . . . . . . . . . . 20
Figure 3.2 Schematic Of Validation Model a) Top View , b) Isometric View . . 21
Figure 3.3 Computational Domain Used For Internal Validation Case . . . . . 23
Figure 3.4 Meshing Used For Internal Validation Case . . . . . . . . . . . . . 24
Figure 3.5 Streamwise Velocity Profiles For Internal Validation Case Grid
Sensitivity Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
Figure 3.6 RMS Residuals For Internal Validation Case . . . . . . . . . . . . 25
Figure 3.7 Computational Domain Used For External Validation Case . . . . . 27
Figure 3.8 Grid Used In External Validation Case . . . . . . . . . . . . . . . 28
Figure 3.9 Streamwise Velocity Profiles For External Validation Case Grid
Figure 3.10 RMS Residuals For External Validation Case . . . . . . . . . . . . 29
Figure 3.11 Computational Models Used In Internal Comparative Study a) Cir-

cular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil
Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
xv
Figure 3.12 Streamwise Velocity Profiles For Internal Comparison Case Grid
Figure 3.13 Near-pin Location Zoomed-in Views Of Grids Used In Compara-

tive Study a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical,
e) Small-Airfoil Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Figure 3.14 RMS Residuals For Internal Comparison Case . . . . . . . . . . . 35
Figure 3.15 Blowing Area Velocity Profiles a) Circular, b) Elliptical, c) Airfoil

Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped . . . . . . . . . . . 36
Figure 3.16 Streamwise Velocity Profiles For External Comparison Case Grid
Figure 3.17 Grid Used In External Comparison Case . . . . . . . . . . . . . . 38
Figure 3.18 RMS Residuals For External Comparison Case . . . . . . . . . . . 39
Figure 4.1 Non-Dimensional Streamwise Velocity Contours a) Experimental

(Fig. 3 of [3] ) , b) Predictions . . . . . . . . . . . . . . . . . . . . . . . 42
Figure 4.2 Streamwise Velocity Contours a) Circular, b) Elliptical, c) Airfoil

Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped . . . . . . . . . . . 43
Figure 4.3 Non-Dimensionalized Streamwise Vorticity Contours a) Circular,

b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
Figure 4.4 Mass Flow-Averaged ψ In Streamwise Direction . . . . . . . . . . 46
Figure 4.5 Entropy Contours a) Circular, b) Elliptical, c) Airfoil Shaped, d)

Small-Elliptical, e) Small-Airfoil Shaped . . . . . . . . . . . . . . . . . . 47
Figure 4.6 Loss Split Per Region . . . . . . . . . . . . . . . . . . . . . . . . 48
Figure 4.7 Mass Flow-Averaged Entropy In Streamwise Direction . . . . . . . 50
xvi
Figure 4.8 Heat Transfer Coefficient Contours a) Circular, b) Elliptical, c)
Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped . . . . . . . . 51
Figure 4.9 Normalized Nusselt Numbers For Pin Arrays a) Experiment (Fig.
4 of [4]), b) Predictions . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Figure 4.10 Loss And Thermal Comparison Of Pin Arrays . . . . . . . . . . . 53
Figure 5.1 Non-Dimensional Streamwise Velocity Contours Downstream Of

Slot Exit a) Experimental (Fig. 3 of [3] ) , b) Predictions . . . . . . . . . 56
Figure 5.2 Streamwise Velocity Contours Downstream Of Slot Exit a) Circu-

lar, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil
Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
Figure 5.3 Streamwise Non-Dimensional Vorticity Contours With Velocity

Vectors At Slot Exit a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-
Elliptical, e) Small-Airfoil Shaped . . . . . . . . . . . . . . . . . . . . . 58

Vectors At 1H Downstream Of Slot Exit a) Circular, b) Elliptical, c) Air-
foil Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped . . . . . . . . . 59



xvii
Figure 5.8 Streamwise Non-Dimensional Vorticity Contours Downstream Of
Slot Exit a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical,
e) Small-Airfoil Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Figure 5.9 Total Temperature Contours Downstream Of Slot Exit a) Circu-

Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
Figure 5.10 Total Temperature Contours With Different Temperature Ranges

for Case (a) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
Figure 5.11 Film Cooling Effectiveness Contours At Breakout Surface a) Cir-

cular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil
Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
Figure 5.12 Film Cooling Effectiveness Contours at Land Side Wall a) Circu-
Shaped . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
Figure 5.13 Laterally-Averaged Film Cooling Effectiveness at Breakout Surface 69
Figure 5.14 X/H Locations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
Figure 5.15 Laterally-Averaged Film Cooling Effectiveness on Land Side Wall . 70
Figure 5.16 Loss And Thermal Comparison Of External Cooling Performance

of Pin Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
xviii
LIST OF ABBREVIATIONS
c Log Layer Constant
D Pin Diameter (mm)
F2 Blending Function
htot Total Enthalpy (m2 /s2 )
H Channel Height (mm)
Hp Pin Height (mm)
HT C Heat Transfer Coefficient
k Turbulent Kinetic Energy
kf Thermal Conductivity of Fluid (W/mK)
L Total Streamwise Length of Internal Cooling Domain (mm)
Nu Nusselt Number
PT Inlet Total Pressure (P a)
PT,l Local Total Pressure (P a)

00
q Wall Heat Flux (W/m2 )
se Mass Flow-Averaged Coolant Exit Entropy (J/kgK)
si Mass Flow-Averaged Coolant Inlet Entropy (J/kgK)
sl Mass Flow-Averaged Local Entropy (J/kgK)
sup Mass Flow-Averaged Upstream Local Entropy (J/kgK)
S Magnitude of Strain Rate
Sij Strain Rate Tensor
Sp Streamwise Distance Between Pins (mm)
Tbulk Bulk Temperature (K)
Tc Coolant Static Temperature at Slot Exit (K)
Ts∞ Main Flow Static Temperature (K)
xix
Twall Wall Temperature (K)
ubulk Bulk Velocity (m/s)
ui Coolant Inlet Velocity (m/s)
us Streamwise Coolant Velocity (m/s)

0
us Non-Dimensional Streamwise Velocity
uτ Friction Velocity (m/s)
U Velocity (m/s)
x Axial Coordinate (mm)
X Axial Distance from Slot Exit (mm)
Xp Lateral Distance Between Pins (mm)
y Distance to Nearest Wall (m)
β Flux Limiter Parameter
∆→
−
r Vector Between Integration Point and Upwind Node
η Film-Cooling Effectiveness
κ Von Karman Constant
ν Kinematic Viscosity (m2 /s)
νt Eddy Viscosity (m2 /s)
ρ Density (kg/m3 )
τ Shear Stress (N/m2 )
ϕ Scalar Variable
Ψ Total Pressure Coefficient
ω Specific Dissipation Rate
ωs Streamwise Vortivity (1/s)

0
ωs Non-Dimensional Streamwise Vorticity
xx
CHAPTER 1
INTRODUCTION
In this chapter, the importance of turbine cooling and conventional cooling techniques
used in modern turbine blades is summarized. Then, the challenges specific to the
trailing-edge cooling, which is the motivation behind this thesis, are explained. Re-
garding the constraints of the trailing-edge cooling, the objectives of this work are
mentioned. Finally, the organization of the thesis is presented.
1.1 Motivation
Gas turbine engines started to be used for power generation in early 1940’s. Since
then the need for improving performance of gas turbines has continued day by day.
This need drives the gas turbine industry and leads to new technological challenges.
One of them is the constant increase of turbine entry temperature (TET) levels. De-
signers prefer to increase TET because this method does not only increase the power
output but also decreases the fuel consumption. The challenge about TET is related
to material limitations. Peak temperatures in advanced gas turbines have been around
2000 K or even more for quite some time, and these levels are well above the maxi-
mum allowable metal temperatures.
To withstand excessive temperatures, different cooling techniques have been devel-

oped and used in modern turbine blades. The increase of TET and the evolution
of cooling techniques over the years is summarized in Figure 1.1. Development of
cooling methods started with smooth cooling holes but this was not enough as temper-
atures continued to rise. With new approaches like turbulated cooling-hole designs,
the aim of cooling blades was achieved, but this time thermal stresses on the blades
1
Figure 1.1: Variation Of TET And Cooling Methods Over The Years [2]
became an issue. This new problem has created new constraints for the designers. As
methods advanced additional complexities were introduced into cooling technologies,
leading to today’s highly sophisticated turbine cooling systems.
Modern turbine blades are equipped with a combination of various cooling tech-
niques to prevent any potential means of failure. In general, these techniques are
divided into two categories as internal cooling and external cooling. Internal cool-
ing is achieved by circulating coolant through cooling holes that are called serpentine
passages. Coolant inside of the blade is blown through small holes on the blade called
film- cooling holes in order to externally cool the blade.
The main aim of internal cooling is to prevent overheating of the blade by creating a
convective heat transfer between coolant and the blade. Mostly used internal cooling
methods are impingement cooling, rib-turbulated cooling, and pin-fin cooling. In
impingement cooling, a high velocity coolant is impinged through the rows of small
holes into the interior surface of the blade. This method is only used in the leading
edge part of the blades due to structural constraints and high centrifugal loads. To
increase heat transfer, various shaped ribs are placed on inner walls of serpentine
2
passages, which is referred to as rib-turbulated cooling. These ribs enhance heat
transfer by both increasing the surface area and the turbulance level of the coolant.
Rib-turbulated cooling is generally implemented on the mid-section of the blades.
The idea behind the pin-fin cooling and the rib-turbulated cooling is the same, but
due to the limited space and the structural constrains of the trailing edge region, pin
fins are used instead of rib turbulators. The internal cooling configurations are shown
in Figure 1.2.
Unlike the internal cooling, the main purpose of the external cooling is to protect
outside of the blade from direct contact with hot gasses. It is achieved by injecting
coolant air used for internal cooling onto the surface of the blade through film cooling
holes located on the blade. Injected coolant creates a protective layer on the surface.
By this way, the heat load on the surface is decreased and since the blade is cooled
from both inside and outside, the thermal stresses generated inside of the blade are
reduced. Figure 1.2 shows the most conventional film cooling hole locations.
Figure 1.2: Cooling Methods On A Turbine Blade [2]
Cooling each region of a blade has its own challenges, but among them the trailing-
edge region is the most difficult to cool down. The difficulty is due to the contra-
dictory requirements of aerodynamic, thermal, and structural aspects. To withstand
high centrifugal loads, the trailing edge region has to be structurally rigid. But, from
an aerodynamic perspective, it should be as thin as possible to reduce mixing losses
downstream of the blade. Regarding the thermal issues, the thin trailing edge causes
high convective fluxes due to low heat capacity, which leads to burns on the blade and
3
eventually the failure. Because of those conflicts, the design of trailing-edge cooling
systems requires a tradeoff between the aerodynamic losses and thermal enhance-
ments.
The most common and effective trailing-edge cooling method is slot film cooling.
The pressure side of the blade is partially cut out along the blade chord, and various
shaped-ribs called lands are placed periodically along the span. Coolant is injected
out of the cutback surface through the openings between the lands, which are called
slots. The sprayed coolant creates a buffer layer between the hot mainstream and the
blade surface [5]. To enhance heat transfer and to counteract the structural weakening
caused by thinning the trailing edge, pin fins are used in the internal cooling passages
at this region.
1.2 Objective
Performance of pin-fin cooling and slot film cooling depends on many variables. For
pin-fin cooling, most important parameters are the pin shape, pin array orientation,
pin spacing, and the channel aspect ratios. Related to slot film cooling, critical factors
are the blowing ratio, lip thickness-to-slot height ratio, land shape, lip shape, and the
lip thickness variation. Those parameters both affect the thermal and aerodynamic
performance of the blade. As stated in the motivation part, trailing-edge cooling is
a trade-off between heat transfer and aerodynamic losses generated by the cooling
system.
In this thesis, aerodynamic and thermal performances of circular, elliptic and NACA
0033 profile shaped pins are studied. The circular pins have been the most-widely
used pin shape in internal cooling, followed by elliptical shapes with a more lim-
ited usage. The airfoil shape is rather more novel shape type that has been studied
scarcely. The blockages in the studied model have the specific airfoil profile NACA
0033 and this profile is mimicked in the whole pin array. The study is divided into two
parts investigating the effects of 1) internal cooling, and 2) external cooling. First part
of the study considers the effects of those three shapes inside of the internal cooling
passages. The major aim in this part is to find the best pin shape and size combina-
4
tion regarding the pressure losses versus heat transfer enhancement. In order to do
this, the aerodynamic loss characteristics of pins and the distribution of heat transfer
coefficients in the passage are investigated. The second part considers the external
trailing-edge cooling. The major aim here is to understand the effect of different inlet
coolant profiles on the slot film-cooling performance. The inlet profiles are generated
by five different pin arrays, and the resulting external flow features downstream of
the slots are studied from the aerodynamic and thermal perspectives. In this way, the
external cooling performance of various internal pin-fin configurations is investigated.
1.3 Thesis Outline
This thesis consists of six chapters. Chapter 1 introduces the need for turbine blade
cooling and gives brief information about conventional cooling technologies specif-
ically used for the blade trailing-edge section. Then, the objectives of this work are
summarized.
Chapter 2 presents the overview of literature that has studied the internal and external
trailing-edge cooling. The internal cooling section particularly focuses on the size,
shape, and the orientation effects of pin-fin cooling. In the external cooling section,
research concerning the slot film cooling is reviewed.
In Chapter 3, the mathematical and numerical modeling approaches applied in this

thesis are explained. The experimental case used as the validation case is explained,
and the solid model and the boundary conditions used for the numerical modeling are
presented.
Chapters 4 and 5 are devoted to the findings derived from the CFD analysis. The
presentation of results is divided into two parts. In the first part, the aerodynamic
findings are discussed followed by the second part where the thermal findings are
demonstrated.
Chapter 6 summarizes the conclusions drawn from this study and gives suggestions
for future work.
5
6
CHAPTER 2
LITERATURE SURVEY
In this chapter, the type of studies that have been done so far related to trailing-edge
cooling methods are summarized. The studies are divided into two parts as internal
and external cooling. In the internal cooling, the research specific to pin-fin cooling
is reviewed. The external cooling part focuses on the studies that investigate slot film
cooling.
2.1 Internal Cooling
Coolant channel structures, especially circular pin fins, were studied in great detail
in literature since they are being used in many other cooling applications as well.
An early study done by Brigham and VanFossen [6] focused on the pin length-to-
diameter ratio and found that this ratio has a major effect on heat transfer coefficient.
Further, they showed that when this ratio is below 2, Nusselt number depends on the
Reynolds number, but not on the length-to-diameter ratio anymore. Metzger et al.
[4] performed experiments to investigate the heat transfer performance of arbitrary
combinations of circular pins by changing the diameter and the stream-wise spacing
of pins in constant and converging flow areas, separately. Then, they used a superim-
position technique to obtain the Nusselt number for complex combinations of circular
pin-fin arrays. Armstrong and Winstanley [7] performed a review on the turbine cool-
ing applications using a staggered-array of pin fins. Their findings showed that the
research results on pin-fin heat transfer and flow friction were not enough to develop
generalized heat transfer and friction-loss correlations. An experimental investigation
for heat transfer capabilities of partial-length circular pin fins was performed by Arora
7
and Abdel Messeh [8]. As a result of these experiments, it was found that both heat
transfer and pressure loss are inversely proportional to the pin-tip clearance. Ligrani
and Mahmood [9] focused on the effects of the coolant-to-wall temperature ratio on
the Nusselt number and friction factor. They concluded that as this ratio decreased,
the Nusselt number increased while the friction factor decreased. Chyu [10] com-
pared circular pins with and without endwall fillet radius, and he discovered that the
pins without the endwall fillet caused higher heat transfer and lower pressure loss.
Although circular pins are the most-frequently used turbulence promoters due to their
easy manufacturing, in order to further increase heat transfer, the effects of different
pin shapes have been investigated by different researchers. A major study on effects
of pin shape on pressure loss and heat transfer was conducted by Metzger et al. [11].
In this work, experiments were performed for two families of array geometries. Pres-
sure loss and heat transfer coefficients of oblong-shaped pins and circular pins were
compared. It was found that although heat transfer performance of oblong-shaped
pins are 20% higher than that of circular pins, the resulting pressure loss is 100 %
higher, which is a noticeable disadvantage to the use of such pins at regions where
a trade-off exists between heat transfer and aerodynamics. In addition, they studied
the arrangement of circular pin fins, and found that changing it from an inline to a
staggered configuration not only increased the heat transfer, but also decreased the
pressure loss at the same time.
Chyu et al. [12] conducted experiments with cubic and diamond-shaped pins. Since
they result in a high heat transfer enhancement while maintaining a moderate pressure
penalty, cubic pins can be used as an alternative to circular pins. A similar compar-
ison between elliptic and circular pins was done by Li et al. [13]. Elliptical pins
had a major-to-minor axis ratio of 1.78 and the same circumference with that of the
circular pins. Heat transfer and resistance coefficients of both pin shapes were calcu-
lated using experimental measurements. It was found that for the investigated range
of Reynolds numbers between 1000 to 10000, elliptical pins display both higher con-
vective heat transfer performance and less flow resistance.
In another study, Uzol and Camci [14] used elliptical pins having the same frontal
area with that of circular pins, rather than the circumference. They investigated the
8
reasons for the improved aerodynamic performance of elliptical pins. Studying the
flow structures around pins, they concluded that the wakes of circular pins are higher
than those of elliptical pins, which enhance heat transfer but create higher pressure
loss.
Chen et al. [15] proposed drop-shaped pins as an alternative to circular pins. In

their research, naphthalene sublimation technique was used to investigate the flow
field. The results showed that depending on their relative spacing, drop-shaped pins
demonstrate 41 % to 52 % less flow resistance with increased overall heat transfer
performance in the studied Reynolds number range.
More recently, Ling et al. [3] measured the three-dimensional velocity and concen-
tration fields of the flow through the pressure-side cutback region of a NACA-0012
airfoil, with a plenum located inside of the airfoil consisting of an array of four rows
of staggered pin fins. The study focused on the flow characteristics at the slot exit
and on the breakout surface. Comparisons between two trailing edge configurations,
one with thin straight lands and airfoil-shaped blockages and one without, revealed
that the thinner lands of the former airfoil resulted in higher spanwise average sur-
face effectiveness, but in lower coolant concentration uniformity. In a subsequent
study, Ling et al. [16] performed RANS modeling of the same trailing edge slot con-
figuration with and without blockages. They showed that the k-ω SST turbulence
model under-predicts the turbulent viscosity both throughout the pin-fin array and on
the breakout surface resulting in inaccurate velocity calculations and coolant concen-
tration. Likewise, Ames and Dvorak [17] performed experiments searching for the
reasons behind heat transfer enhancement of pin fins. They modeled their experi-
ments using conventional turbulence models. Their findings suggest that all of the
turbulence models they used under-predict both heat transfer and pressure drop.
As in [16] [17], there have been other attempts to approach the problem from a nu-
merical standpoint in addition to experimentation. In their study, Martini et al. [18]
modeled three different internal cooling structures using unsteady detached eddy sim-
ulation (DES). Their results showed that for examining discharge coefficients, steady
RANS models could be used as an alternative to time-consuming unsteady DES cal-
culations. Wang et al. [19] performed both experiments and computations with five
9
different pin geometries with the same cross-sectional areas. While flow structures
showed distinct differences between the data and the predictions, the pressure loss be-
haviors were found to be in good agreement. Their findings suggest that drop-shaped
pins can be used as an alternative to circular pins. More recently, Fernandes et al.
[20] studied the accuracy of the well-known turbulence models in the prediction of
heat transfer rates and pressure distribution on the surface of pin fins. They concluded
that the quadratic realizable k- and the k-ω SST turbulence models estimate the heat
transfer rates the most accurately.
Pin shapes other than circular ones did not get enough attention because manufactur-
ing of different shapes in micro scales was a problem. But with a new manufacturing
technique called "Direct Metal Laser Sintering " (DMLS), it is now possible to pro-
duce complex shapes in micro scales. This new technique, opened a new research
area in pin fin cooling. Nowadays, researchers in this area are trying to find alter-
native shapes that have better aerothermodynamic performance. In their recent work
Ferster et al. [21] conducted experiments for triangular, star shaped and dimpled
spherical-shaped pin fins manufactured with DMLS technique. They investigated the
effects of pin geometry, spacing, and number of pins on heat transfer and pressure
loss performance. Their experiments showed that triangular pins outperformed star
shaped and dimpled spherical pins in heat transfer augmentation. Besides experimen-
tal results were compared with cylindrical pin results from an old study [22], which
proved that their heat transfer and pressure loss performances were similar which
makes triangular pins desirable since weight of triangular pins would be less than
circular ones having the same wetted area.
2.2 External Cooling
Trailing edge external cooling configurations are investigated in great detail due to
their direct impact on turbine stage efficiency. A notable study related to trailing-edge
cooling was done by Cunha and Chyu [23]. They developed analytical solutions for
the temperature distribution of four simple trailing-edge cooling configurations that
are solid wedge shape without discharge, wedge with slot discharge, wedge with dis-
crete hole discharge, and wedge with pressure-side cutback slot discharge. They have
10
also done experiments to understand the details of a more complicated trailing-edge
cooling configuration, which includes an internal cooling channel with pins and an ex-
ternal cooling structure with pressure-side cutback supported with lands. They claim
that with a good selection of the design parameters, cutbacks with lands configura-
tion is a desirable trailing-edge cooling method regarding structural and aerodynamic
performances.
A leading study in slot film-cooling area was done by Kacker and Whitelaw [24]. In
their study, they conducted experiments to find a correlation between adiabatic wall
effectiveness and slot lip thickness-to-slot height ratio. They concluded that those two
parameters are inversely proportional to each other. In a similar study, Sivasegaram
and Whitelaw [25] considered the effects of slot lip thickness-to-slot height ratio and
the injection angle on film cooling effectiveness. Their findings on slot lip thickness-
to-slot height ratio agree with the earlier study. Besides, they showed that film-cooling
effectiveness of an angled slot can be grater than a tangential slot.
More recently, Horbach et al. [26] investigated the aerodynamic and heat transfer
performance of different lip geometries. They conducted experiments by considering
four lip thicknesses and three lip shapes. This study revealed that the lip thickness
affects the mixing of coolant and the main flow at the slot. As the lip thickness
increases, the wake region grows, which increases pressure loss and decreases film-
cooling performance. They also found that the lip shape does not considerably affect
the thermal and aerodynamic performance of film cooling. In their following study,
Horbach et al. [27] used the same setup with three different pin fin configurations
which are circular and streamwise and spanwise-oriented elliptical pins to see the
effect of internal pin fin configuration on the external film-cooling performance. By
taking circular pins as a reference point, the streamwise-oriented pins have higher dis-
charge coefficients and lower heat transfer performance, while the spanwise-oriented
pins have lower discharge coefficients and higher heat transfer performance. Re-
garding film-cooling effectiveness, the streamwise-oriented and circular pins perform
similarly whereas the spanwise configuration displays poor performance.
A very detailed analysis of slot film cooling was performed by Taslim et al. [28].
Their study concerns about the effects of the injection angle, slot lip thickness-to-
11
height ratio, slot width-to-height ratio, blowing ratio, and the coolant-to-mainstream
density ratio on film effectiveness. They applied constrains on parameters consider-
ing the general engine operation conditions and conducted a series of experiments
parametrically. The experiments revealed that film effectiveness is insensitive to the
density ratio and slot aspect ratios, while lip thickness-to-height ratio and the injec-
tion angle have considerable impact on film effectiveness. Based on their results, they
developed a correlation between lip thickness-to-height ratio and film effectiveness,
and they showed that the correlation changes depending on the blowing ratio. They
also found that 8.5 is the optimum injection angle for the best film effectiveness.
A similar research was done by Fiala et al. [29], which investigated how the blow-
ing rate, Reynolds number and the external turbulence affect heat transfer and film-
cooling effectiveness. They focused on the top and side surfaces of letterbox-shaped
lands. According to their results, film effectiveness can be correlated with blowing
ratio. It is directly proportional at land tops while inversely proportional at land sides.
Moreover, they found that the external turbulence does not have a considerable influ-
ence on the land film-effectiveness levels.
Yang and Hu [30] conducted experiments by using Pressure Sensitive Paint Technique
to explore influence of existence of lands on film-cooling performance. Experiments
were performed with and without land configurations for blowing ratios between 0.4
and 1.6. Their results claim that as the blowing ratio increases, the streamwise film
effectiveness increases but the spanwise effectiveness decreases. The existence of the
lands empowers film-cooling performance in the streamwise direction since the lands
enforce coolant to spread in the slot channel. But, at the same time, the lands restrain
the expansion of coolant in the spanwise direction, which worsens film-cooling ef-
fectiveness. Although their findings support that overall heat transfer performance of
design without land case is better, they still recommend using lands to ensure struc-
tural strength of blades.
Although most of the research related to the trailing-edge cooling focus on heat trans-
fer augmentation and film-cooling effectiveness, Uzol et al. [31] studied the aerody-
namic aspects of the trailing-edge cooling. They investigated how cut back length and
rib spacing affect the discharge coefficient at different free-stream Reynolds numbers.
12
For all configurations, they showed that the discharge behavior weakly depend on the
Reynolds number. They also showed that as the spanwise rib-spacing gets smaller,
the aerodynamic performance of the configuration gets better. In their following work
[32], they looked into the flow field near the trailing edge closely to clarify the reasons
of pressure loss characteristics. Besides, they tried to numerically model flow near
the trailing edge to shed light into the details of the flow field that was not resolved by
experimental capabilities. They observed that for the ejection rates of 0 and 3, pres-
sure loss levels were increasing while at the ejection rate 5, loss levels were reduced.
Their computational results showed that the loss mechanisms were mainly ruled by
the shear generated between the coolant and the mainstream.
Most of the research related to the trailing-edge cutback cooling are experimental.
But, as in [32] there have been other attempts to numerically model the flow near
the trailing edge. Holloway et al. [33] conducted experiments to see the unusual
phenomenon related to the effect of increasing the blowing ratio on the film-cooling
effectiveness and modeled the experimental set up numerically to look at the reasons
of this phenomenon closely. Their experimental results showed that the effectiveness
increases until blowing ratio 0.8 but it decreases between 1.0 and 1.25. Further in-
creasing blowing ratio results in an increase in the effectiveness levels. They modeled
the experiment using steady Reynolds Averaged Navier Stokes with RNG k- turbu-
lence model. The computationally-calculated efficiency monotonically increases as
the blowing ratio increases, which was not the case in the experiments. Their conclu-
sion about that was that the unsteadiness in experiments could not be modeled using
steady RANS, and an unsteady simulation was needed. In this direction, in their
subsequent study [34] they focused on the unsteady modeling. Those results revealed
that the interaction between the coolant and the mainstream vortices is responsible for
the phenomenon observed in the first part of the study. Until a blowing ratio of 1, the
mainstream vortices were dominant, but after that point the coolant vortex shedding
became dominant, and they had approximately an equal strength near the blowing
ratio of 1.
As is summarized in this chapter, there are many studies that investigate the coolant
flow through a trailing edge slot geometry from a thermal performance perspective,
while only a limited number of studies have been concerned with the aerodynamic
13
aspects. Among those limited studies, mostly circular pins have been of interest,
while studies on other shapes have been much less in number. In this study, the ef-
fect of internal pin-array structures on the downstream trailing edge slot exit flow
is investigated. Different pin shapes that are of circular, elliptical, and airfoil shape
(NACA 0033 profile) with varying sizes are modeled, presenting a detailed compari-
son. The same airfoil shape is adopted with that of the island blockage of the trailing
edge configuration, resulting in a unique pin-array and blockage combination that has
not been investigated before according to the literature search performed during this
thesis work. The results obtained in the internal cooling structure provide the inlet
coolant profiles to the external cooling structure that consists of the pressure-side cut-
back slot region. The downstream effects of the internal pin-array configurations are
also examined in detail providing information on the slot film-cooling performance.
Resulting internal and external flow field characteristics are investigated both from
aerodynamic and thermal aspects of trailing edge design.
14
CHAPTER 3
METHODOLOGY
In this chapter, the computational method applied in the thesis is explained in detail.
First, the mathematical modeling and numerical modeling approaches are defined.
Then, the cases used for the validation of the numerical model are described. Dis-
cretization of the computational domains and the boundary conditions used in nu-
merical modeling are presented. Finally, the cases used for comparing different pin
shapes are explained with their discretization and boundary conditions.
3.1 Mathematical Modeling
The behavior of fluid flow can be mathematically modeled using governing equations.
The governing equations are obtained by applying the conservation of mass, conser-
vation of momentum and conservation of energy principles on the domain of interest.
The differential forms of these conservation equations are presented below:
Conservation of Mass
∂ρ ∂(ρUi )
+ =0 (3.1)
∂t ∂xi
Conservation of Momentum
∂(ρUi ) ∂(ρUj Ui ) ∂p ∂τij

+ =− + (3.2)
∂t ∂xj ∂xi ∂xj
Conservation of Energy
∂T
∂(ρ(htot )) ∂p ∂(ρUi htot ) ∂(Ui τij ) ∂(λ ∂xj )
− + = + (3.3)
∂t ∂t ∂xi ∂xi ∂xj
15
3.2 Numerical Modeling
The analytical solutions of coupled governing equations exist for some simple cases.
For solving the flow field of complex flow cases, numerical methods are developed.
Mostly-used numerical approaches are the finite difference, finite volume, finite el-
ement, and spectral methods. The common property of numerical methods is that
the differential forms of the governing equations are discretized and approximated
as algebraic equations that can be solved computationally. To apply discretization
methods on the flow domain, firstly the physical flow space need to be divided into
small parts called grids. Scales of grids depend on the smallest length scales present
in the flow domain. The smallest grid should be smaller than the smallest length scale
of flow so that the flow filed is fully resolved. This requirement becomes a problem
according to the available computational power, especially for high Reynolds flows
since the length scale gets smaller as Reynolds Number is increased.
Thanks to high performance of modern supercomputers, a direct solution of the dis-

cretized governing equations are possible for simple flows with Reynolds numbers
up to order 105 . This method is called Direct Numerical Simulation (DNS) and it
gives the most accurate results related to the flow domain. Other than DNS, to predict
the effects of turbulence with less computational effort, different turbulence models
were developed. In these methods, flow field variables are divided into average and
fluctuating parts, and governing equations are reconstructed with average and time-
dependent values. This method is called Reynolds Averaging. The related Reynolds-
Averaged Navier-Stokes (RANS) equations are presented below:
∂ρ ∂(ρUi )
+ =0 (3.4)
∂t ∂xi
∂(ρUi ) ∂(ρUj Ui ) ∂p ∂(τij − ρui uj )

+ =− + (3.5)
∂t ∂xj ∂xi ∂xj
∂T
∂(ρhtot ) ∂p ∂(ρUj htot ) ∂(Ui (τij − ρui uj )) ∂(λ ∂xj − ρuj h)
− + = + (3.6)
∂t ∂t ∂xj ∂xi ∂xj
16
RANS equations contain additional flux terms. These are Reynolds stresses ρui uj and
Reynolds fluxes ρuj h . Reynolds stresses and fluxes contain fluctuating velocity com-
ponents that are unknowns, which means that Reynolds-Averaging brings additional
unknowns to the closed system of equations formed by Navier-Stokes equations. This
problem is referred to as the ”Closure Problem ” in fluid dynamics. To overcome this
problem, additional terms have to be modeled. Modeling of turbulence is an active
research area and a large variety of turbulence models were developed. Main cate-
gories of turbulence models are first order closures, second order closures, and Large
Eddy Simulation.
In this work for numerical modeling, a commercial software ANSYS CFX was used
and the fluid flow was modeled using the 3D Reynolds-Averaged Navier-Stokes (RANS)
equations. For discretization, CFX uses a dual median finite-volume method. A
high-resolution advection scheme was used for evaluating the discretized equations.
Formulation of scheme is given in equation 3.7 where ϕ means any scalar variable.
Integration point value of variable is calculated using upwind node value (ϕupw ), con-
trol volume gradient of variable ϕ (∇ϕ), vector between integration point and upwind
node (∆→ −
r ) , and flux limiter parameter β. This scheme switches from the second or-
der to the first order depending on the value of β. This parameter changes between 0
and 1 according to the boundedness principle described by Barth and Jespersen [35].
The methodology is explained via the equations 3.8,3.9 , and 3.10. Application of this
method requires the computation of minimum and maximum values of ϕ in the up-
wind node and its adjacent nodes. Since the integration point value should be between
the minimum and maximum values, to eliminate overestimation or underestimation,
the β formulation given in equation 3.10 is used.
ϕip = ϕupw + β∇ϕ∆→

−
r (3.7)
ϕmax
i = max{ϕupw , ϕmax
i }andϕmin
i = min{ϕupw , ϕmin
i } (3.8)
ϕmin
i < ϕip < ϕmax
i , ∀i (3.9)
17

ϕmax −ϕupw



min{1, i
ϕi −ϕupw
} ϕi − ϕupw > 0

β = min 1 ϕi − ϕupw = 0 (3.10)
∀i 

min{1, ϕmin −ϕupw

} ϕi − ϕupw < 0
 i
ϕi −ϕupw
To satisfy closure, the Reynolds stresses are modeled using the k − ω based Shear
Stress Transport (SST) [36] that is a two-equation Eddy Viscosity turbulence model.
This model behaves like the k − model outside of the boundary layer and the k − ω
model inside of the boundary layer with the help of the blending function. Addition-
ally, in order to restrain the over-prediction of eddy-viscosity, a limiter was used in its
formulation. For this model Eddy Viscosity is formulated with blending function F2
and magnitude of strain rate tensor Sij as follows:
5/9k
νt = (3.11)
max(5/9ω, SF2 )
√
2 k 500ν 2
F2 = tanh([max( , )] ) (3.12)
0.009ωy y 2 ω
p
S= 2Sij Sij (3.13)
Since the use of wall function is not encouraged for the separated and recirculating
flows, the automatic wall function provided by CFX for the k − ω based models was
active. This wall function automatically switches from a low-Reynolds to a near-wall
formulation when the boundary layer is highly resolved [37]. In this method specific
dissipation rate ω and friction velocity uτ formulated combining both viscous sub-
layer and logarithmic region formulations. Their formulations are given in equations
below where ∆y is THE wall normal distance between the first and the second nodes.
s
µ ∆U
uvis
τ = | | (3.14)
ρ ∆y
U
ulog
τ = √ τw (3.15)
ρ
∆y
1/κ ln ( ν
)+c
18
q
4 log 4
uτ = (uvis 4
τ ) + (uτ ) (3.16)
6ν
ωvis = (3.17)
0.075(∆y)2
q √
(uvis 4
τ ) + ( 0.3k)
4
ωlog = q (3.18)
0.3κ∆y τρw
r
ωlog 2
ωω = ωvis (1 + ( )) (3.19)
ωvis
3.3 Validation Case
Validation case selected for this study is taken from the experimental study of Ling
et al. [3]. Magnetic resonance imaging is used for measuring the 3D velocity field
and the coolant concentration fields at the downstream of the trailing edge pressure-
side film-cooling slots. NACA-0012 airfoil with 283 mm blunted trailing edge chord
and 76 mm span are used in the experiments. Inner section of the airfoil includes a
coolant channel with four rows of pin fins and NACA-0033 profile shaped blockages.
Last 42.1 mm of the blade is cut out from the pressure side and is stiffened with 7
mm-wide lands. The coolant flows through the breakout surface from the slots with 5
mm height and 17 mm length. The details of the geometry are given in Figure 3.1.
19
Figure 3.1: Schematic Of Experimental Model [3]
The validation case is divided into two parts for this work. First, the flow inside the
cooling channel is investigated. Then, by using the results of the cooling channel,
the external film cooling part that includes the trailing-edge breakout surface is in-
vestigated. For the first part, the whole cooling channel from the coolant inlet to the
slot exit is modeled. Internal section model is shown as blue in Figure 3.2. Results
of the first part showed relatively uniform flow for all three slots and this situation is
observed in the experiments as Ling et al. [1] mentioned. Due to the symmetry of
the geometry and the uniformity of the flow structure at the slot exit, the symmetry
boundary conditions were imposed for the external cooling validation. The flow do-
main that includes the half of the slot and half of the land is modeled for the second
part. The modeled slot and land geometry is shown between two solid lines in part (a)
of Figure 3.2. The region of interest for external and internal cooling is shown with
green and purple colors respectively, while the yellow region in part (b) of Figure
3.2. is a common part of the solution domains for the internal and external validation
cases.
a)
20
b)
Figure 3.2: Schematic Of Validation Model a) Top View , b) Isometric View
3.3.1 Internal Cooling
3.3.1.1 Boundary Conditions
Boundary conditions of the numerical model is set to match with those of the exper-
iments of Ling et al. [1]. The typical velocity-inlet and static pressure-outlet type
of boundary conditions were applied on the model. The details of the experiment
include mostly the main flow inlet conditions and the turbulence intensity, but there is
no information about the coolant flow conditions at the coolant inlet. For this reason,
the inlet velocity is determined through iterations so that the average coolant velocity
at the slot exit and the coolant mass flow rate given in Table 3.1 are matched with that
of the experiment. There was no explanation about the coolant inlet turbulence char-
acteristics and the velocity profile. However, in Ling et al. [1]’s computational work,
it is observed that the predictions were insensitive to the coolant inlet conditions [1].
Regarding this observation, inlet velocity is taken to be uniform and the turbulence
21
intensity was assumed to be 5% for the computations. The end walls, pins, slots,
and the airfoil-shaped blockages were modeled with no-slip and fixed temperature
surfaces.
Table 3.1: Experimental Conditions From Ref [1]
Average Bulk Velocity at Slot Exit [m/s] 0.39

Average Coolant Velocity at Slot Exit [m/s] 0.30
Coolant Temperature [0 C] 20
Working Fluid Density [kg/m3 ] 998
Coolant Mass Flow Rate [l/min] 4.41
The coolant used in the experiments is water. In the experiments, copper sulfate is
mixed with water to enhance signal to noise ratio and effectively use MRV. Density
and viscosity of used solution are negligibly different than pure water. Because of
this reason, in this work coolant is used as pure water. The fluid domain is modeled
as a constant-property pure water at 1 atm reference pressure and at 25 0 C reference
temperature with a specific heat capacity of 4181.7 J/kg.K at constant pressure and
a dynamic viscosity of 8.899x10−4 kg/m.s.
3.3.1.2 Geometry and Meshing
A schematic of the flow domain used in the internal cooling validation study is shown
in Figure 3.3. Coolant enters the domain through a circular opening on the side
and the triangle-shaped domain turns the flow through the pins. Coolant leaves the
domain through the slots located downstream of the airfoil-shaped blockages.
The computational domain was meshed using ICEM CFD software. Due to the com-
plexity of the geometry, the domain was discretized using tetrahedral elements. Prism
elements were used around the pins, airfoil-shaped blockages, slots, and at the end
walls. The thickness of the first prism cells were selected such that maximum y +
is equal to 1. This is recommended for accurate heat transfer predictions [37]. To
satisfy this, near-wall spacing was set as 0.006 mm .The growth of prism layers was
set as exponential with constant rate of 1.2. 19 prism elements are used to represent
22
Figure 3.3: Computational Domain Used For Internal Validation Case
the boundary layer. Figure 3.4 shows the grid used in validation study and zoomed-in
view which includes the last row of pins and the trailing edge exit region with a land
and a blockage.
A grid sensitivity study was conducted by using different sizes of three grids with the
number of elements varying from 1.1M to 5.6M. The sensitivity study was performed
by monitoring the velocity profiles along the centerline of the pin array across the top
and bottom walls of the computational domain. Figure 3.5 shows the velocity profiles
along the centerline for three different grid solutions.
23
Figure 3.4: Meshing Used For Internal Validation Case
Figure 3.5: Streamwise Velocity Profiles For Internal Validation Case Grid Sensitivity
Study
24
The average percentage of differences between coarser, medium and finer grids were
calculated by using Equation 3.20 where ζfi iner is the solution of finer grid and ζ i is
the solution of the grid of interest at the ith data point and N is the total number of
data points.
PN ζfi iner −ζ i
i=1 ζfi iner
Average % Dif f erence = (3.20)
N
The average percentage of differences between 1.1M and 2.4M was calculated as 11.3
% while 2.4M and 5.6M was calculated as 9.1 %. Due to the computational cost of
further refinement, the grid with 2.4 M elements was selected for the validation study.
3.3.1.3 Solution Convergence
Figure 3.6 shows the convergence levels obtained.
Figure 3.6: RMS Residuals For Internal Validation Case
25
3.3.2 External Cooling
Boundary conditions of the numerical model is again set to match with those of the
experiments of Ling et al. [1]. The typical velocity-inlet and static pressure-outlet
type of boundary conditions were applied on the model. The slot region is modeled
as an opening and the velocity profile information is imported from the results of the
internal cooling validation part. In the internal cooling part, the slot region boundary
condition was set as atmospheric pressure. To be consistent with the internal cooling
part, outlet pressure in external cooling part is adjusted to give atmospheric pressure
at the slot region. Since the fluid used is water at low velocity, setting the outlet static
pressure to match the atmospheric pressure gave only 0.02 % difference in the slot
static pressure, which is considered to be negligible.
Table 3.2: External Flow Experimental Conditions From Ref [1]
Main Flow Inlet Velocity[m/s] 0.19

Inlet Turbulence Intensity[%] 1
Inlet Turbulent Length Scale [mm] 1
Main Flow Temperature [0 C] 20
Main Flow Mass Flow Rate[l/min] 123
A schematic of the flow domain used in the external cooling validation study and
the domain names are shown in Figure 3.7. Coordinates of the test section and the
dimensions of the blade are taken from Ling et al. [1]. Main flow enters the domain
through the inlet and the coolant enters the domain through the section named as
blowing area. This presented domain is the outlet of the internal validation case.
26
Figure 3.7: Computational Domain Used For External Validation Case
The main flow inlet was placed 46.4 mm slot-height upstream of the blade leading
edge and the outlet was placed 114.6 mm slot-height downstream of the blade trailing
edge. To take advantage of symmetry and to decrease computational effort, only half
of the slot and half of the land are included in this validation study.
For meshing, the domain is discretized using unstructured tetrahedral elements and
prism elements are used in near-wall regions. First layer thicknesses around the walls
are adjusted so that y + is kept around 1. It is proved that the tetrahedral mesh with
27
prism layers for capturing the boundary layer can be used in order to predict laterally-
averaged film effectiveness downstream of the coolant holes [38]. Since outer domain
is not far from the blade, prism elements are used for tunnel walls, too. Since the
main flow and the coolant are mixing downstream of the slot, the density of the grid
is increased in that region. Figure 3.8 shows the external geometry meshing and a
close-up look of the slot region.
Figure 3.8: Grid Used In External Validation Case
To satisfy y + around 1, near wall spacing was set as 0.022 mm around the blade,
slot, and the land region, and 0.035 mm around the tunnel walls. The growth of the
prism layers was set as exponential with constant rate of 1.2. 20 layers for tunnel
walls and 18 layers for other walls were implemented. For grid sensitivity, prism
layer parameters are kept constant. This study was performed for the grids with 1.3
M, 3.8 M and 8.4 M elements. The methodology explained in Section 3.3.1.2 is
applied. Streamwise velocity profiles on the mid section of the plane 2.5 mm above
the breakout region are presented in Figure 3.9. This location is selected since the
region of interest in this study is the breakout region. According to these results, the
28
grid with 3.8 M elements were selected since the average percentage of the difference
between the medium and the fine mesh is 2.1% while the difference between the
coarse and the medium mesh is 1.5%.
Figure 3.9: Streamwise Velocity Profiles For External Validation Case Grid Sensitiv-
ity Study
Figure 3.10: RMS Residuals For External Validation Case
29
3.4 Comparison Cases
For the comparative studies, five different cases are prepared. In the internal cooling
part, the aerothermodynamic performances of three different pin shapes that are of
circular, elliptic ,and NACA-0033 profile airfoil are investigated. Besides, in order to
see the effect of pin size, smaller-sized elliptic and airfoil-shaped pins are analyzed.
In the external cooling part, the film-cooling performances of different pin shapes
using the results from the internal cooling part are investigated.
3.4.1 Internal Cooling
For comparison cases, the boundary conditions are taken from the experimental mea-
surements that were obtained by Hylton et al. [39] for the C3X cascade vane. The
boundary conditions are different from the validation case. This change is necessary
since the validation case is performed with water and with conditions that do not rep-
resent the real engine environment. In their study, Benson et al. [40], who performed
experiments at the same test facility with the same basic flow configuration as the val-
idation case, concluded that even if the Reynolds number range that was tested with
water was significantly less than that in real engine environment, it represented a fully
turbulent flow where the main flow characteristics were not significantly affected with
an increase in Reynolds number. Therefore, this agreement obtained between predic-
tions and data provides a validation for the subsequent comparative studies that were
performed with air as an ideal gas.
The computations are performed for the selected boundary conditions that reflect the
actual engine operation conditions. The experimental data sets presented in [39] con-
tain various parameters like total temperature, total pressure, Mach number, Reynolds
number, wall-to-gas temperature ratio at various locations. Cases 4421 and 4422 are
selected for this study. The necessary boundary conditions to perform the compu-
tations were either adopted directly from [39] or were calculated if they were not
readily available. A summary of these conditions is given in Table 3.3. The coolant
30
inlet velocity is calculated using the main inlet conditions such that the blowing ratio
is equal to 0.8 as a typical value. The coolant inlet velocity was taken to be uni-
form. The inlet turbulence intensity was kept at 5% as in the validation case. All
solid walls were modeled using the no-slip boundary conditions with a constant tem-
perature of approximately 614 K. The approach for thermal boundary conditions are
taken from [18]. The fluid domain is modeled as a calorically perfect ideal gas with a
specific heat capacity of 1004.4 J/kg.K at constant pressure and a dynamic viscosity
of 1.831x10-5 kg/m.s.
Table 3.3: Internal Flow Comparison Case Boundary Conditions
Coolant Inlet Velocity[m/s] 31.1

Coolant Inlet Temperature[K] 478.2
Wall Temperatures [K] 614.2
Outlet Static Pressure [kPa] 332.1
For comparisons, the internal-cooling validation domain is simplified and the coolant
flow is introduced into the passage directly in the streamwise direction. This made
it possible to reduce the size of the computational domain as the flow through the
trailing edge internal geometry becomes symmetrical in addition to the symmetry of
the geometry itself. Using a plane of symmetry through the centerline of the model,
the grid size was halved. This simplification was justified by experiments [1]. In the
experiments the coolant was fed into the domain radially and the pin-fin geometry
orients the flow uniformly at the slot exit. Since this work considers only the geomet-
ric section of the pin-array, the coolant was fed directly in the streamwise direction to
the domain.
The reference pin shape used is circular and it has 5 mm diameter. Elliptical and
airfoil-shaped pins are sized such that the minor axes of the elliptical pins and the
maximum thicknesses of the airfoil-shaped pins have the length of a circular pin di-
ameter. This way, frontal areas of all shapes were kept the same rather than their
circumferences, as was done in [14], guaranteeing the same amount of flow blockage
31
Table 3.4: Internal Comparison Case Model Dimensions
Streamwise Length [mm] 79.82

Height [mm] 5.00
Diameter of Circular Pins [mm] 5.00
Major Axis Length of Elliptical Pins [mm] 14.07
Chord of Airfoil-Shaped Pins [mm] 14.07
Minor Axis Length of Small Elliptical Pins [mm] 1.78
Maximum Thickness Length of Airfoil-Shaped Pins [mm] 1.78
area. Major axis length of the elliptical pins is selected such that it is the same as
the chord of the airfoil-shaped pin. To see the size effect, small elliptical and small
airfoil-shaped pins are created such that the major axis of elliptical pin and the chord
of the airfoil-shaped pin are equal to the diameter of the circular pin. The major axis-
to-minor axis ratio of the small elliptical pins is matched to that of the larger elliptical
pins.
The dimensions of the models used in the comparative study are summarized in Ta-
ble 3.4. The base model is a constant 5 mm-height-section that has a 79.82 mm-
streamwise length. The inlet of the domain coincides with the end of the smooth
contraction from the coolant inlet into the constant-height pin section. Four rows of
pin fins are placed along the section, with straight lands and airfoil-shaped blockages
at the end of the section. An airfoil shape is used for the benefit of separation reduc-
tion in the wake. The flow domains used in the internal cooling comparison study is
shown in Figure 3.11.
Similar to the validation studies, the prism elements were used around the pins,
airfoil-shaped blockages, slots, and at the end walls. The thickness of the first prism
cells were selected such that the maximum y + is equal to 1. To satisfy this, near wall
spacing was set as 0.0008 mm. The growth of prism layers were set as exponential
with constant rate of 1.2, and 30 prism elements are used to refine the boundary-layer
region.
32
a) b)
c) d)
e)
Figure 3.11: Computational Models Used In Internal Comparative Study a) Circular,

A grid sensitivity study is performed as explained in Section 3.3.1.2 for the grids with
650 K, 1.6 M and 3.4 M elements.Figure 3.12 shows the streamwise velocity profiles
along the centerline of the pin array across the top and bottom walls of the computa-
tional domain. The average percentage of the difference between the coarse and the
medium grid was observed to be 9.7%, and the difference between the medium
33
Figure 3.12: Streamwise Velocity Profiles For Internal Comparison Case Grid Sensi-
tivity Study
a) b)
c) d)
e)
Figure 3.13: Near-pin Location Zoomed-in Views Of Grids Used In Comparative

Study a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-
Airfoil Shaped
34
and the finer grid was observed to be 9.1% for the circular-pin case. As a result, the
case with 1.6 millions of elements is decided to be used in the rest of the comparison
study. Other configurations are meshed with the same parameters selected for the
circular-pin geometry, and the corresponding grids of similar sizes are used for those
cases. The mesh in the pin vicinity for each case is shown in Figure 3.13
Figure 3.14: RMS Residuals For Internal Comparison Case
3.4.2 External Cooling
For comparison cases, the boundary conditions are taken from the experimental mea-
surements that were obtained by Hylton et al. [39] for the C3X cascade vane. The
typical velocity-inlet and static pressure-outlet type of boundary conditions were ap-
plied on the model. The slot region is modeled as an opening. The velocity profile
and the static temperature is used as the boundary condition at the blowing area. This
35
information is imported from the results of the comparison cases of internal cool-
ing. Velocity profiles of different pin shapes are presented in Figure 3.15 and static
temperatures are presented in Table 3.5.
Table 3.5: Mass Flow Averaged Slot Static Temperatures For Different Pin Shapes
Circular[K] 504.2
Elliptical[K] 505.5
Airfoil Shaped [K] 505.1
Small-Elliptical [K] 498.6
Small-Airfoil Shaped [K] 498.8
a) b)
c) d)
e)
Figure 3.15: Blowing Area Velocity Profiles a) Circular, b) Elliptical, c) Airfoil

Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped
36
The inlet velocity is calculated as 95.32 m/s from the inlet Mach number given in
the data set [39] and this same value is used for all five comparison cases. The inlet
turbulence level is set as 8.3 % and inlet total temperature is set as 782 K. The outlet
pressure is calibrated such that the average static pressure at the slot exit is equal to
the boundary condition supplied at the internal comparison case, giving a value of
307.8 kPa. All solid walls are modeled using the no-slip boundary conditions and
they are treated as adiabatic. The approach for thermal boundary conditions is taken
from [18].
In the comparison study, the same airfoil and tunnel geometry presented in 3.3.2.2 are
used. The only difference is that the slot region was connected to the breakout surface
with a 3.6 degree-angle in the experimental setup and therefore this was mimicked in
the validation case, but in the comparison case this angle is taken as zero.
The same meshing strategy explained in Section 3.3.2.2 is applied. First cell height
of all walls are taken as 0.0011 mm. For the grid sensitivity study, meshes with 2.7
M, 6.0 M and 14.2 M elements are used. Figure 3.16 shows the streamwise velocity
Figure 3.16: Streamwise Velocity Profiles For External Comparison Case Grid Sen-
sitivity Study
37
profiles along the same line used for the external validation case that was previously
described. The average percentage of the difference between the medium and the
fine mesh is 1.7% and the difference between the coarse and the medium mesh is
5.2% . This is why the medium case with 6.0 M elements is selected for the external
comparison study.
Figure 3.17: Grid Used In External Comparison Case
Figure 3.17 shows the external geometry meshing and a close-up look of the slot
region. This mesh is used for all five comparison cases with the same meshing pa-
rameters.
38
Figure 3.18: RMS Residuals For External Comparison Case
39
40
CHAPTER 4
INTERNAL COOLING ANALYSIS & RESULTS
In this chapter, the analysis performed and the results obtained in the internal cool-
ing study are presented. First, the applied convergence criterion is explained. Then,
the results for the validation study are shown and the validation of the model is pre-
sented. The comparison study results are divided into two parts as aerodynamic and
thermal. The chapter is concluded with the comparison of aerodynamic and thermal
performances of five different pin-fin geometries.
4.1 Validation
The comparison between the predictions and the experiment is shown in Figure 4.1
in the form of contours of streamwise velocity component through the pin arrays and
the slot exit of the test configuration. The velocity is non-dimensionalized with the
average experimental pressure-side main flow velocity, where ubulk is given in Table
3.1, at the plane of the slot exit [3]. The contours are shown on a plane that is 2 mm
above the bottom surface as was done by Ling et al.[3] for data demonstration. The
non-dimensionalization method is given by Equation 4.1 where us is the local coolant
velocity:
0 us
us = (4.1)
ubulk
Because the coolant is injected from the side of the domain at an upstream location,
there is an observable asymmetry in the flow field entering the domain. Figure 4.1
presents the comparison in a portion of the domain demonstrated in Figure 3.3 where
41
a) b)
Figure 4.1: Non-Dimensional Streamwise Velocity Contours a) Experimental (Fig. 3

of [3] ) , b) Predictions
the plenum section contracts into the constant-height region containing the pin array.
According to the orientation of Figure 4.1, the flow enters the domain from the top
and from a further upstream location out of the boundaries of the figure, and moves
towards left where the slots are located. The effect of this injection is reflected as an
increase in the local velocity, both in the contours of the predictions and the data.
The velocity measurement at the location corresponding to the blockage-island max-

imum width is 0.51 m/s [1]. This value was predicted as 0.52 m/s, resulting in an
over-prediction of approximately 2%. In fact, the velocity predictions are found to be
in good agreement with the data over the whole domain of Figure 4.1. The separa-
tion bubbles are observed to form immediately behind every pin fin within the blue
regions in a similar fashion to what the data suggests. These wakes extend towards
downstream pins, and the velocity is recovered across the remaining sections of each
passage. In the wake of the islands, the flow separation is not as obvious. The data
contours show that the flow is redistributed through the pin array; and this is observed
in the prediction contours of Figure 4.1 as well. The only significant difference oc-
42
curs on the right side of the contours that are closer to the manifold entrance where
the coolant is introduced into the test section. The assumed velocity profile and tur-
bulence intensity are likely to cause a difference in the predictions at the immediate
vicinity of this part of the domain. However, as was mentioned before, the computa-
tional studies performed in [1] showed that the coolant inlet conditions did not have
a significant impact on the predictions.
4.2 Aerodynamic Results From Comparison Cases
Figure 4.2 shows the streamwise velocity contours at the mid-plane. Similar to the
validation case in Figure 4.1 as the flow proceeds through the passage, the regions of
reverse flow are formed behind every pin fin. Since the inlet flow is introduced in the
streamwise direction and is assumed to be uniform, the flow distribution reflects this
inlet condition in the downstream region.
a) b)
c) d)
e)
Figure 4.2: Streamwise Velocity Contours a) Circular, b) Elliptical, c) Airfoil Shaped,

d) Small-Elliptical, e) Small-Airfoil Shaped
43
The top side of each contour plot is a physical wall while the bottom side is a symme-
try wall. Therefore, the local flow features along these two sides are not necessarily
symmetric due to the endwall effects occurring along the top side only. On the other
hand, the flow field is symmetric with respect to the symmetry wall, but the other half
of the domain is not shown in the figure.
Although the velocity gradients through the circular pin array are mild, they become
more significant for the pin shapes of ellipse and airfoil profile. The separation re-
gions are intensified at the last row of pins for these cases. This is mostly due to the
relatively short distance between the pin and the blockage surfaces, as well as the lack
of another follow-up pin row. While the circular pin rows are totally isolated from
each other across the array, elliptical, and airfoil-shaped pin rows are not. Hence, the
reduced flow area in between the trailing edge of one row and the leading edge of its
downstream row causes an increase in the local velocity.
On the other hand, the wakes of the airfoil-shaped pins across the pin array of case
are not obvious, since an airfoil shape will have a reduced separation zone compared
to a blunt object such as cylinder or ellipse. For the small elliptical and small airfoil-
shaped cases, since the pins are of a smaller size having the same row extents with
those of circular case, the interaction between each row is not as significant anymore.
The velocity increases in the flow direction in general, reaching its maximum through
the blockage area. Large velocity gradients are observable in this region towards the
slot exit both in streamwise and spanwise (along the section width) directions. Due
to the overlaps between the rows of cases (b) and (c), the interaction between the pins
contribute to the velocity gradients as well. In other words, if the rows are separated
from each other, the velocity gradients diminish as is shown in cases (d) and (e). For
those cases, a significant velocity gradient is observable only through the blockage
area.
44
a) b)
c) d)
e)
Figure 4.3: Non-Dimensionalized Streamwise Vorticity Contours a) Circular, b) El-

liptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped
In order to have more information on the flow patterns, the streamwise vorticity con-
tours in the mid-plane are examined in Figure 4.3. Vorticity is non-dimensionalized
by the air inlet velocity and the section height. The non-dimensionalization method
is given by Equation 4.2, where H is the channel height, ui is the inlet velocity, and
the subscript s defines the streamwise direction:
0 ωs H
ωs = (4.2)
ui
The neighborhood of each pin across an array is dominated by the vortical flow struc-
tures. The blunt shape of the circular pin is the one producing the largest wake region
where the flow separation occurs. In addition, there are horseshoe vortices occurring
along the surfaces of the pins, coinciding with the strong vorticity regions. The wake
of the airfoil-shaped pin (c) is significantly smaller than those of (a) and (b), meaning
a reduced aerodynamic penalty. With the reduced size of the pins (d) and (e), this
penalty is even lower.
45
Flow features can be examined by looking at the total pressure coefficient, which
gives a measure for the total pressure drop in streamwise direction via Equation 4.3:
PT − PT,l
Ψ= (4.3)
ρi u2i /2
Figure 4.4: Mass Flow-Averaged ψ In Streamwise Direction
In Figure 4.4, this coefficient is mass flow-averaged across the width of the domain,
and is shown as a function of the normalized distance from the inlet. To do so, a
series of vertical cut planes at different downstream locations were used. Then, the
mass flow-average values of Ψ were calculated on these planes. Since there are large
velocity variations in the lateral direction, mass flow-averaging was preferred rather
than spanwise-averaging to better quantify the parameters of interest, such as loss and
entropy generation.
Starting off with the same inlet total pressure, all three configurations cause a con-
sistent increase in Ψ due to the loss accumulation across the domain. In Figure 4.4,
the four vertical dashed lines show the locations of the pin centers, which are the
same for all five configurations. The configuration with the elliptical pins seems to
have the highest amount of loss, followed by the circular pins and the airfoil-shaped
46
pins. Although this seems contrary to the findings of [14] regarding the performance
of elliptical and circular pins, the elliptical pins in this study have significantly larger
wetted areas resulting in higher viscous dissipation and hence higher losses. However,
even if the airfoil-shaped pin has also a large wetted area, the lessened aerodynamic
penalty for this shape is remarkable. For the small size elliptical and airfoil-shaped
pins, the aerodynamic penalty is significantly less due to the shrinkage in the surface
area.
a) b)
c) d)
e)
Figure 4.5: Entropy Contours a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-

Elliptical, e) Small-Airfoil Shaped
More insight into the loss mechanism can be gained if the entropy generation across
the domain is analyzed. Figure 4.5 demonstrates the top view of the section mid-plane
with the entropy contours. The general trends show a consistent entropy generation
in the streamwise direction for all five configurations. The top side of the domain
is a solid wall while the bottom side is a symmetry wall. Substantial losses in the
domain occur by the walls due to the effects of viscosity, which is indicated by the red
color. When the flow reaches the trailing edges of the airfoil-shaped pin and blockage
47
surfaces, losses add up due to the local acceleration and mixing of the surface flows.
This is more significant at the blockage trailing edges. The entropy generation is the
largest for the elliptical pins, followed by the circular and airfoil-shaped pins, due
to the large wake regions behind the pins. It is hard to distinguish the contours of
the small elliptical and small airfoil-shaped pins. Hence, their loss characteristics are
very similar.
Figure 4.6: Loss Split Per Region
In order to quantify the loss amount for each configuration in more detail, the com-
putational domain is split into six regions, as in Figure 4.6. Region A represents the
inlet section, Region B is the pin array, and Region C consists of the remaining portion
with the lands and the blockages. The layout given here shows the five configurations
superimposed on top of each other. The split planes are located approximately a one-
pin diameter (1D) away from the row centerlines. The first and last planes are further
48
away from the nearby centerlines, approximately 1.5D and 2D, respectively, in or-
der to include the extensions of the larger pin shapes. Region B is further split into
row-by-row sections. Due to the dimensions of the shapes, the circular pins remain
between the planes, while the leading edge of the large elliptical pins and the trailing
edge of the large airfoil-shaped pins run over the planes. However, with this approach
the loss quantification can be simplified reasonably.
For the loss audit, the mass flow-averaged entropy value for each cutting plane (sl, l
for local) was compared with the mass flow-averaged entropy value of the previous
plane (sup, up for upstream), according to Equation 4.4:
sl − sup
T otalLoss% = × 100 (4.4)
se − si
where the difference is normalized by the total increase between the inlet and exit
planes of the domain.
The loss amount does not accumulate much in Region A, as this is the shortest region
and there are no cross-pin effects due to the absence of the pin structures. As the flow
starts going through the pin array, a significant rise occurs in the loss for all config-
urations, and the loss stays at similar levels across the pin rows. All pins experience
elevated levels of entropy generation in Region B4. This region is where the wakes
of pins are strongly experienced due to a combined effect of the absence of another
downstream row of pins and the interaction with downstream lands and blockages
that are located in close proximity. Finally, the largest amount of losses for all five
configurations occurs in Region C.
Figure 4.6 provides information on the loss split across the domain for each config-
uration, but it does not compare the loss amounts between configurations. For this,
the maximum mass flow-averaged entropy on the exit split plane across all five con-
figurations is used to normalize the local mass flow-averaged entropy on the planes
splitting the domain. This is defined as the entropy ratio on the y-axis of Figure 4.7,
and its variation is given as a function of x/L. The vertical dashed lines are attached
to the plot to identify the pin locations.
49
Figure 4.7: Mass Flow-Averaged Entropy In Streamwise Direction
Figure 4.7 presents the mass flow-averaged entropy values, with similar trends to Ψ
variation of Figure 4.4. Comparing the general trends in these figures, it is clear that
the overall loss generation is greater for the elliptical pins, leading the circular and
airfoil-shaped pins by a small difference of around 2%. The small pins generate less
penalty. It can be concluded from this discussion that losses could be reduced if the
downstream slot section was kept further apart from the last row of pins for the given
pin dimensions relative to the domain. Since the rows were not totally isolated for the
elliptical and airfoil-shaped pins, the interaction between the rows also contributes to
the overall loss mechanism. The losses are further reduced if the pins are of smaller
size. The difference between the two small pins is almost negligible, and their curves
are almost identical. They provide aerodynamics savings of approximately 7-8%
compared to their larger size partners.
4.3 Thermal Results From Comparison Cases
The convection heat transfer coefficient, HTC, was evaluated on the upper wall of the
domain. For the calculations, the bulk fluid temperature, Tbulk, the wall temperature,
50
00
Twall, and the wall heat flux, q , were used as shown in Equation 4.5:
00
q = HT C × (Twall − Tbulk ) (4.5)
a) b)
c) d)
e)
Figure 4.8: Heat Transfer Coefficient Contours a) Circular, b) Elliptical, c) Airfoil

Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped
The bulk temperature was calculated as the mass flow-averaged temperature on the
spanwise planes in order to take into account the streamwise temperature of the
coolant. The HTC contours give the highest levels at the stagnation locations on the
pin surfaces as well as on the downstream blockage frontal areas. The horseshoe vor-
tices rolling up around the pin surfaces coincide with the traces of high heat-transfer
regions. The levels are reduced significantly for the smaller size pins, hinting that the
heat transfer performance of these pins will not be as good.
51
Additionally, Nusselt number was calculated using Equation 4.6:
HT C × D
Nu = (4.6)
kf
where D is the pin diameter, and kf is the thermal conductivity of air as the working
fluid that has a value of 0.0261 W/m.K. The Nusselt number calculations are com-
pared to the data set provided by Metzger et al. [4]. This data is presented in Figure
4.9.
a)
b)
Figure 4.9: Normalized Nusselt Numbers For Pin Arrays a) Experiment (Fig. 4 of
[4]), b) Predictions
52
In this study, Metzger et al. [4] defines a row-averaged Nusselt number (Nu) and an
array-averaged Nusselt number (N¯u). The shaded band consists of the data from cir-
cular pins with Reynolds number ranging from 2310 to 51740. Figure 4.9 shows the
prediction results with respect to this band. The pin-array structures had Xp /D = 2.4,
Sp /D = 2.4, and Hp /D = 1. Here, Xp is the lateral distance between the pins, Sp is
the streamwise distance between the pins, and Hp is the pin height.The flow Reynolds
number was calculated to be 52375. Considering that these values are within close
range of the data of Figure 4.9, a comparison can be performed with the predictions
of this study. The prediction results reasonably fell into or around the shaded band.
Considering that the data represents the circular pins, the observable mismatch of the
predictions for the elliptical and airfoil-shaped pins is understandable.
Figure 4.10: Loss And Thermal Comparison Of Pin Arrays
Figure 4.10 summarizes the findings of the internal comparison study. The bars filled
with diagonal stripes presented in the figure represents the ratio of total pressure loss
along section B to the average total pressure loss of all five pin arrays. The black
bars represent the ratios of the array averages to the overall average of all five pin ar-
rays for Nusselt number. According to this comparison, the airfoil-shaped pins bring
aerodynamics savings due to their significantly-reduced wake region, compared to
the pins of similar size. For the two small-size pins, the differences between the aero-
53
dynamics and thermal characteristics are not distinguishable. Although the pressure
loss is significantly reduced, the heat transfer performance is reduced as well.
54
CHAPTER 5
EXTERNAL COOLING ANALYSIS & RESULTS
In this chapter, the analysis and results of the external cooling study are presented.
First, the validation of model is presented and the results for the validation study
are discussed. Later, the details of the comparison study are explained. Here, the
comparison study results are again demonstrated both from an aerodynamics and a
thermal perspective. The results obtained from five different pin-fin geometries are
compared.
5.1 Validation
The comparison between the predictions and the experiment is shown in Figure 5.1
in the form of contours of non-dimensional streamwise velocity component through
the pin arrays and the slot exit of the test configuration. The non-dimensionalization
method is given by Equation 4.1 where us is the local coolant velocity. The details of
the validation of the internal cooling case was explained in Section 4.1. Here, the val-
idation of the flow field downstream of the slot exit is performed and the predictions
are compared with the experiments of Ling et al. [3].
According to the orientation of Figure 5.1, the coolant enters the domain from the
left side where the slot exit is located and then it mixes with the main flow. The
shown section is the plane 2 mm above the breakout surface and provides a top view
of the downstream flow field. The black region represents the solid land. The top
and bottom sides are the symmetry walls. The use of the symmetry walls has enabled
to cut down the computation time significantly. From Figure 5.1 it can be seen that
the low-momentum wake behind the airfoil-shaped blockages and pin fins extend far
55
Figure 5.1: Non-Dimensional Streamwise Velocity Contours Downstream Of Slot
Exit a) Experimental (Fig. 3 of [3] ) , b) Predictions
downstream in the contours for the predictions. This under-prediction behavior with
k − ω SST turbulence model is reasoned with the insufficient turbulent viscosity [1].
Although the mixing process of the coolant with the mainstream flow seems to occur
slower in the predictions compared to the experiment, the velocity predictions are
generally found to be in good agreement with the data over the whole domain.
5.2 Aerodynamic Results From Comparison Cases
Figure 5.2 shows the streamwise velocity contours at the mid-plane above the break-
out surface for all five configurations.Reverse flow regions due to mixing are seen
at the land tip and locally at the breakout region. Further downstream of the slots,
the velocity increases in the flow direction. Flow patterns for all cases are the same
downstream of the breakout region but shows minor differences at the breakout right
at the slot exit.
To see the development of the flow on the breakout surface, a more detailed analysis
is pursued. Planes perpendicular to the streamwise direction are created at every 5
mm (1H) distance starting from the slot exit, where H is the slot exit height. Vorticity
contours with velocity fields are presented in Figures 5.3, 5.4, 5.5, 5.6, 5.7. The left
56
sides of the figures are symmetry walls, while the right sides are solid land walls.
Vorticity is non-dimensionalized by the air inlet velocity and the section height as
described in Section 4.2.
a)
b)
c)
d)
e)
Figure 5.2: Streamwise Velocity Contours Downstream Of Slot Exit a) Circular, b)

Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil Shaped
Figure 5.3 shows the non-dimensional streamwise vorticity contours at the slot exit.
Regions represented by contour colors smaller than zero indicate clockwise rotation
(blue color) while contour colors greater than zero indicate counterclockwise rotation
(red color). Circular and elliptical cases show similar vortical structures while the
airfoil-shaped case generates reversed vortices near the symmetry wall. A change in
the size decreases the strength of this structure in both airfoil-shaped and elliptical
pin cases. The large vortical structures observed become smaller for the small size
pins. The reason for that is the longer distance between the slot opening and the last
row of small pins than the first three cases. The weaker vortices behind the small pins
find enough room to diminish in the flow direction. It can be said that for small pins,
the flow structure at the slot is mostly generated by the airfoil-shaped blockage just
before the slot exit.
The similarity between the circular and elliptical, and the small elliptical and small
airfoil-shaped cases is obvious in the velocity boundary conditions given in Figure
3.15 as well. In Figure 5.3, for all cases, the velocity vectors are directed to the sym-
57
metry wall. This is expected since the domain is continuous here and the flow moves
in this direction due to the existence of the airfoil-shaped blockages right upstream of
the slot exit in the internal cooling channel.
a) b)
c) d)
e)
Figure 5.3: Streamwise Non-Dimensional Vorticity Contours With Velocity Vectors

At Slot Exit a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-
Airfoil Shaped
Figure 5.4 shows the non-dimensional streamwise vorticity contours at 1H down-
58
stream of the slot exit. For all cases, the velocity vectors have changed their direc-
tions, and the flow moves upward instead of traveling towards the symmetry wall.
a) b)
c) d)
e)

At 1H Downstream Of Slot Exit a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-
This is due to the interaction between the main flow and the coolant blown out from
59
the slot exit. At this plane, due to the mixing of the main flow with the coolant,
the strength of both clockwise and counterclockwise vortices have increased. The
similarity between the circular and elliptical cases as well as the small pins still exist
in this plane.
a) b)
c) d)
e)

60
stream of the slot exit. In this plane, it is observed that the strength of clockwise
vortices (blue regions ) have decreased, while the counterclockwise vortices have
gained strength. Considering the direction of the velocity vectors, it can be said that
the mixing process still continues.
a) b)
c) d)
e)

61
stream of the slot exit. Starting from this plane, the change in the vorticity contours
becomes less obvious.For all cases, a notable difference between 5.5 and Figure 5.6
is the change in the direction of the flow in the plane. Other than that, the velocity
a) b)
c) d)
e)

62
vectors and the vorticity contours look similar. This indicates that the effect of the
inlet conditions applied at the slot opening has spanned a distance of 3H downstream
of the slot exit, and beyond that the main flow conditions suppress the impact of the
coolant flow and dominate the flow region.
Figure 5.7 proves that the effect of coolant has disappeared. There are small differ-
ences in the velocity field and the vorticity contours between 3H and 4H downstream
of the slot exit. The differences are generally seen near the wall sides, which seem to
be the effect of the boundary layer.
The development of the flow field at the breakout surface is summarized in 5.8. For
all five cases, the vorticity contours at the first 25 mm distance from the slot opening
are presented with 5 mm intervals. The first five of six planes are explained in Figures
5.3, 5.4, 5.5, 5.6, 5.7. The last plane is generated to show that the flow field becomes
stable. By looking at the big picture presented in Figure 5.8, the pin shape effect
is dominant till 3H downstream of the slot, but further downstream, the main flow
conditions for all cases are the same and the mixed coolant does not reflect the effects
of the pin shapes anymore.
63
a) b)
c) d)
e)
Figure 5.8: Streamwise Non-Dimensional Vorticity Contours Downstream Of Slot

Exit a) Circular, b) Elliptical, c) Airfoil Shaped, d) Small-Elliptical, e) Small-Airfoil
Shaped
64
5.3 Thermal Results From Comparison Cases
a) b)
c) d)
e)
Figure 5.9: Total Temperature Contours Downstream Of Slot Exit a) Circular, b)

65
Figure 5.9 presents the total temperature contours on some cross-sectional planes at
the downstream of the slot exit along the streamwise direction. For all five cases,
the total temperature contours at the first 25 mm distance from the slot opening are
presented with 5 mm intervals. At the slot exit, coolant temperature levels are low
and and they increases in the flow direction. This is because as soon as the coolant
exits the slot, it starts mixing with the surrounding hot mainstream. These tThermal
mixing patterns of the coolant and the main stream flow show similaritiesy for all
cases. Total temperature levels are much lower for the small- sized cases since the
coolant temperatures are lower.
a) b)
Figure 5.10: Total Temperature Contours With Different Temperature Ranges for
Case (a)
The differences between the cases over the whole breakout region are distinguishable
with the temperature range used in Figure 5.9. On the other hand, the temperature dis-
tribution on the cut planes can be resolved better if a temperature range that includes
the highest temperature value is used. On the last plane, the temperature reaches up to
726 K. Temperature gradients due to mixing of the coolant and the mainstream flow
can be seen especially on the last plane of Figure 5.10 part (b).
The performance of film cooling can be demonstrated with the use of film-cooling
effectiveness. It is calculated with Equation 5.1 where Ts∞ is the main flow static
temperature of 777.5 K , Twall is the wall temperature calculated by the CFD anal-
66
ysis, and Tc is the coolant temperature calculated as the mass flow averaged static
temperature at the slot exit that was obtained from the calculations of internal cool-
ing.
Ts∞ − Twall
η= (5.1)
Ts∞ − Tc
a) b)
c) d)
e)
Figure 5.11: Film Cooling Effectiveness Contours At Breakout Surface a) Circular,

In order to display the effects of different pin shapes, the film-cooling effectiveness
contours for all five cases are presented at the breakout surface in Figure 5.11. At
the slot exit, the effectiveness levels are high and they diminish in the flow direction.
For the small elliptical and small airfoil-shaped pins, the effectiveness shows a rela-
tively slow decrease and a similar behavior regarding the film-cooling performance
of these pins. This is related to the internal cooling performance of the small pins.
Due to their lower heat transfer performance in the internal channel, the coolant tem-
perature did not increase too much. Hence, the heat capacity of the coolant results in
higher film effectiveness for external cooling. Similarly, the coolant temperature at
the slot exit for the circular pin case is a little bit less than that of the elliptical and
67
the airfoil-shaped cases, which increased the film cooling capacity. Since the coolant
temperature for the elliptical case is the highest one among all cases, film-cooling
effectiveness levels for this pin shape are found to be the lowest.
a) b)
c) d)
e)
Figure 5.12: Film Cooling Effectiveness Contours at Land Side Wall a) Circular, b)
As the coolant passes between the two lands, it cools not only the breakout surface
but also the side walls of the lands. Therefore, studying the film-cooling effectiveness
also on the land side wall will give important clues about the performances of different
pin shapes. Figure 5.12 shows the film-cooling effectiveness contours on the side wall
of a land. The coolant is ejected over the breakout surface right at the slot exit, while
the mainstream flow partially fills this section that is split with the slot lip located
above the slot exit. The effect of this mainstream flow is shown with the blue region.
The thermal patterns of the coolant flow and the mainstream mixing are similar for
all cases. As in the breakout surface film-cooling effectiveness, the land side film-
cooling effectiveness levels are directly related to the slot exit temperatures of the
coolant.
68
Figure 5.13: Laterally-Averaged Film Cooling Effectiveness at Breakout Surface
The contour plots of film-cooling effectiveness reveal how the effectiveness dimin-
ishes at the breakout surface. To compare overall film-cooling performances of dif-
ferent pin shapes, the film-cooling effectiveness along the breakout surface is aver-
aged in the lateral direction. Figure 5.13 shows the laterally-averaged film cooling
effectiveness at the breakout surface. The vertical axis shows the laterally-averaged
film-cooling effectiveness, while the horizontal axis shows the ratio of the axial dis-
tance from the slot exit (X) to the slot height (H). X/H=0 is the slot exit, and X/H=8.4
is the exit of the breakout surface.
Figure 5.14: X/H Locations
69
A few sample X/H plane is shown in Figure 5.14 All five cases show similar trends
until the location of X/H=7. After this point, the film-cooling effectiveness for the
circular case drops more drastically compared to the other cases. At the end of the
breakout surface, the film-cooling effectiveness of the airfoil-shaped case stays higher
among all cases.
Figure 5.15: Laterally-Averaged Film Cooling Effectiveness on Land Side Wall
Figure 5.15 shows the laterally-averaged film-cooling effectiveness on the side wall
of the land. Here, averaging is done over the height of the side wall. As was observed
at the breakout surface, on the side of the land elliptical, small elliptical, and small
airfoil-shaped pins perform similarly across the region. But in contrast to the breakout
surface, the circular pins exhibit the best thermal performance while the airfoil-shaped
pins present the lowest average film-cooling effectiveness levels. At x/H=5.5, the
film-cooling effectiveness for all cases starts increasing. This location is where the
breakout surface opens up in the lateral direction. It is likely that this change in the
geometry causes the lower-momentum fluid to provide a better coverage on the side
walls.
70
Figure 5.16: Loss And Thermal Comparison Of External Cooling Performance of Pin
Arrays
Figure 5.16 summarizes the findings of the external comparison study. The bars filled
with diagonal stripes represent the ratio of total pressure loss between the slot exit and
the end of the blade (end of the breakout surface) to the average total pressure loss of
all five pin arrays. The black bars represent the ratio of the film-cooling effectiveness
average for each array to the overall average of all five pin arrays at the breakout
region. According to this comparison, the airfoil-shaped and small-sized pins show
nearly the same aerothermodynamic performances. For elliptical and circular pin
cases, the pressure loss levels are slightly above the average value, while their thermal
performances are similar to other three cases. This comparison shows that the overall
aerodynamic and thermal performances of film cooling at the breakout surface of the
trailing edge region are weakly dependent on the flow structure introduced at the slot
exit region.
71
72
CHAPTER 6
CONCLUSIONS
This thesis has focused on pin fin and slot film cooling that are commonly used
as the means of cooling technology implemented into designs that are working at
high-temperature environments. The fluid flow was modeled using the 3D Reynolds-
Averaged Navier-Stokes (RANS) equations. The computations were performed to
determine the most advantageous combination of shape and size of pin arrays for
trailing edge cooling from an aerodynamics and a thermal perspective. For the inves-
tigation, the base model with circular pins was taken from the experimental work of
Ling et. al [3] and was used for validation purposes. Three different pin shapes of
circular, elliptical, and NACA 0033 airfoil profile, and two different pin sizes were
considered, resulting in five different models. The flow features, losses, and heat
transfer coefficients were compared both inside of the trailing edge section across the
pin array and also on the external surface downstream of the slot exit. In defining the
pin shapes, the frontal areas and the maximum thicknesses of the shapes were kept
the same in order to establish the same amount of blockage area for aerodynamic
comparisons. The other option studied was to keep the major axes and the diameter
the same. The blockages by the slot exit were of NACA 0033 profile shape, and the
airfoil-shaped pins adopted the same profile.
The study is divided into two parts. In the first part, the aerothermodynamic and
thermal performances of the pin fins inside the internal cooling channel were inves-
tigated. The velocity, pressure, loss variations, heat transfer coefficients and Nusselt
number distributions across the flow domain were analyzed for the selected set of
boundary conditions. The local features in the flow field showed that the largest total
pressure drops occur in and around the pin wakes in the streamwise direction. The
73
last row of pins, where the elliptical and airfoil-shaped pins stayed relatively closer
to the leading edge of the downstream lands and islands, the flow field experienced
stronger and wider zones of flow separation resulting in substantial loss generation
for this size of pins. The local features showed that the losses keep adding up through
the pin array. Despite its larger wetted area, the airfoil-shaped pin was found to have
less aerodynamic penalty mostly due to the separation reduction in its wake region,
while its thermal performance was at similar levels to those of similar size. On the
other hand, the smaller size pins produced less amount of loss as expected; however,
this is accompanied with a reduction in thermal performance in return.
In the second part, the investigation has focused on the effects of pin-fin arrays on
the flow structure at the slot exit region and the consequences of this structure on
the downstream region. The flow conditions generated by different pin arrays at the
end of the internal section of the trailing edge were applied at the slot exit as the
inlet boundary conditions to the external flow domain of the computations. Minor
differences were located in the velocity contours at the breakout region. The stream-
wise vorticity contours spanning the downstream region of the trailing edge slot exit
revealed that the influence of different coolant inlet conditions starts to diminish ap-
proximately around 15 mm downstream of the slot exit, and beyond that point the
flow structure for all pin cases behaves similarly. The comparison of pressure coef-
ficients showed that all pin shapes have also very similar pressure loss trends. For
the analysis of the thermal performance, heat transfer coefficient, Nusselt number,
and film-cooling effectiveness are examined. The small-sized pins demonstrate the
highest film-cooling effectiveness at the breakout surface due to lower temperatures
at the slot exit. According to the varying aerodynamic and thermal performances of
the pins across the internal and external sections, it can be inferred that there is no one
correct answer to the selection of the optimal cooling configuration, and the decision
should be made according to the needs of the design. On the other hand, performing
an optimization on the size of the airfoil-shaped pins would be a promising approach
as well.
With the goal of choosing an optimal pin fin configuration that is aerothermodynam-
ically more advantageous for slot film cooling, this thesis provides a through investi-
gation that would be of interest to the turbine designers. The comparison presented
74
in this paper considers airfoil-shaped pins of NACA 0033 that have not been studied
before within an array of pins for similar purposes according to the literature search
performed during this thesis work. The circular pins have been commonly used in
current designs so far, and there have been studies performed using elliptical pins, al-
though limited in number. The airfoil shape is rather a more novel shape type that has
been studied scarcely. With recent advances in the additive manufacturing, however,
the implementation of any random shape into the flow path design seems feasible
nowadays. For this reason, future studies focusing on either trailing-edge design, or
on any component design in general, seem to have room to investigate the use of
novel shapes in turbomachinery design.
75
76
REFERENCES
[1] J. Ling and J. Eaton, “Improvements in turbulent scalar mixing modeling for
trailing edge slot film cooling geometries: A combined experimental and com-
putational approach.,” tech. rep., Sandia National Lab.(SNL-CA), Livermore,
CA (United States), 2015.
[2] J.-C. Han, S. Dutta, and S. Ekkad, Gas turbine heat transfer and cooling tech-
nology. CRC Press, 2012.
[3] J. Ling, S. D. Yapa, M. J. Benson, C. J. Elkins, and J. K. Eaton, “Three-

dimensional velocity and scalar field measurements of an airfoil trailing edge
with slot film cooling: The effect of an internal structure in the slot,” Journal of
Turbomachinery, vol. 135, no. 3, p. 031018, 2013.
[4] D. Metzger, W. Shepard, and S. Haley, “Row resolved heat transfer varia-
tions in pin-fin arrays including effects of non-uniform arrays and flow con-
vergence,” in ASME 1986 International Gas Turbine Conference and Exhibit,
pp. V004T09A015–V004T09A015, American Society of Mechanical Engi-
neers, 1986.
[5] M. Cakan and M. Taslim, “Experimental and numerical study of mass/heat

transfer on an airfoil trailing-edge slots and lands,” Journal of Turbomachinery,
vol. 129, no. 2, pp. 281–293, 2007.
[6] B. A. Brigham and G. J. VanFossen, “Length to diameter ratio and row number
effects in short pin fin heat transfer,” Journal of Engineering for Gas Turbines
and Power, vol. 106, no. 1, pp. 241–244, 1984.
[7] J. Armstrong and D. Winstanley, “A review of staggered array pin fin heat trans-
fer for turbine cooling applications,” Journal of Turbomachinery, vol. 110, no. 1,
pp. 94–103, 1988.
[8] S. Arora and W. Abdel-Messeh, “Characteristics of partial length circular
77
pin fins as heat transfer augmentors for airfoil internal cooling passages,” in
ASME 1989 International Gas Turbine and Aeroengine Congress and Exposi-
tion, pp. V004T08A008–V004T08A008, American Society of Mechanical En-
gineers, 1989.
[9] P. M. Ligrani and G. I. Mahmood, “Variable property nusselt numbers in a chan-

nel with pin fins,” Journal of Thermophysics and Heat Transfer, vol. 17, no. 1,
pp. 103–111, 2003.
[10] M. Chyu, “Heat transfer and pressure drop for short pin-fin arrays with pin-
endwall fillet,” in ASME 1989 International Gas Turbine and Aeroengine
Congress and Exposition, pp. V004T08A011–V004T08A011, American Soci-
ety of Mechanical Engineers, 1989.
[11] D. Metzger, C. Fan, and S. Haley, “Effects of pin shape and array orientation on
heat transfer and pressure loss in pin fin arrays,” Journal of Engineering for Gas
Turbines and Power, vol. 106, no. 1, pp. 252–257, 1984.
[12] M. Chyu, Y. Hsing, and V. Natarajan, “Convective heat transfer of cubic fin ar-
rays in a narrow channel,” Journal of Turbomachinery, vol. 120, no. 2, pp. 362–
367, 1998.
[13] Q. Li, Z. Chen, U. Flechtner, and H.-J. Warnecke, “Heat transfer and pressure
drop characteristics in rectangular channels with elliptic pin fins,” International
Journal of Heat and Fluid Flow, vol. 19, no. 3, pp. 245–250, 1998.
[14] O. Uzol and C. Camci, “Heat transfer, pressure loss and flow field measurements
downstream of staggered two-row circular and elliptical pin fin arrays,” Journal
of Heat Transfer, vol. 127, no. 5, pp. 458–471, 2005.
[15] Z. Chen, Q. Li, D. Meier, and H.-J. Warnecke, “Convective heat transfer and
pressure loss in rectangular ducts with drop-shaped pin fins,” Heat and Mass
Transfer, vol. 33, no. 3, pp. 219–224, 1997.
[16] J. Ling, C. J. Elkins, and J. K. Eaton, “Optimal turbulent schmidt number

for rans modeling of trailing edge slot film cooling,” in ASME Turbo Expo
2014: Turbine Technical Conference and Exposition, pp. V05BT13A015–
V05BT13A015, American Society of Mechanical Engineers, 2014.
78
[17] F. Ames and L. Dvorak, “Turbulent transport in pin fin arrays: experimental data
and predictions,” Journal of Turbomachinery, vol. 128, no. 1, pp. 71–81, 2006.
[18] P. Martini, A. Schulz, H.-J. Bauer, and C. Whitney, “Detached eddy simulation
of film cooling performance on the trailing edge cutback of gas turbine airfoils,”
Journal of Turbomachinery, vol. 128, no. 2, pp. 292–299, 2006.
[19] F. Wang, J. Zhang, and S. Wang, “Investigation on flow and heat transfer char-
acteristics in rectangular channel with drop-shaped pin fins,” Propulsion and
Power Research, vol. 1, no. 1, pp. 64–70, 2012.
[20] R. Fernandes, M. Ricklick, A. Prasad, and Y. Pai, “Benchmarking reynolds av-

eraged navier–stokes turbulence models in internal pin fin channels,” Journal of
Thermophysics and Heat Transfer, pp. 1–7, 2017.
[21] K. K. Ferster, K. L. Kirsch, and K. A. Thole, “Effects of geometry, spacing,

and number of pin fins in additively manufactured microchannel pin fin arrays,”
Journal of Turbomachinery, vol. 140, no. 1, p. 011007, 2018.
[22] K. L. Kirsch and K. A. Thole, “Pressure loss and heat transfer performance
for additively and conventionally manufactured pin fin arrays,” International
Journal of Heat and Mass Transfer, vol. 108, pp. 2502–2513, 2017.
[23] F. Cunha and M. K. Chyu, “Trailing-edge cooling for gas turbines,” Journal of
Propulsion and Power, vol. 22, no. 2, pp. 286–300, 2006.
[24] S. Kacker and J. Whitelaw, “An experimental investigation of the influence of

slot-lip-thickness on the impervious-wall effectiveness of the uniform-density,
two-dimensional wall jet,” International Journal of Heat and Mass Transfer,
vol. 12, no. 9, pp. 1196–1201, 1969.
[25] S. Sivasegaram and J. Whitelaw, “Film cooling slots: the importance of lip
thickness and injection angle,” Journal of Mechanical Engineering Science,
vol. 11, no. 1, pp. 22–27, 1969.
[26] T. Horbach, A. Schulz, and H.-J. Bauer, “Trailing edge film cooling of gas tur-
bine airfoils—effects of ejection lip geometry on film cooling effectiveness and
heat transfer,” Heat Transfer Research, vol. 41, no. 8, 2010.
79
[27] T. Horbach, A. Schulz, and H.-J. Bauer, “Trailing edge film cooling of gas tur-
bine airfoils—external cooling performance of various internal pin fin configu-
rations,” Journal of Turbomachinery, vol. 133, no. 4, p. 041006, 2011.
[28] M. Taslim, S. Spring, and B. Mehlman, “Experimental investigation of film

cooling effectiveness for slots of various exit geometries,” Journal of Thermo-
physics and Heat Transfer, vol. 6, no. 2, pp. 302–307, 1992.
[29] N. Fiala, I. Jaswal, and F. Ames, “Letterbox trailing edge heat transfer: Ef-
fects of blowing rate, reynolds number, and external turbulence on heat transfer
and film cooling effectiveness,” Journal of Turbomachinery, vol. 132, no. 1,
p. 011017, 2010.
[30] Z. Yang and H. Hu, “Study of trailing-edge cooling using pressure sensitive
paint technique,” Journal of Propulsion and Power, vol. 27, no. 3, pp. 700–709,
2011.
[31] O. Uzol, C. Camci, and B. Glezer, “Aerodynamic loss characteristics of a tur-

bine blade with trailing edge coolant ejection: Part 1—effect of cut-back length,
spanwise rib spacing, free-stream reynolds number, and chordwise rib length on
discharge coefficients,” Journal of Turbomachinery, vol. 123, no. 2, pp. 238–
248, 2001.
[32] O. Uzol and C. Camci, “Aerodynamic loss characteristics of a turbine blade with
trailing edge coolant ejection: part 2—external aerodynamics, total pressure
losses, and predictions,” Journal of Turbomachinery, vol. 123, no. 2, pp. 249–
257, 2001.
[33] D. S. Holloway, J. H. Leylek, and F. A. Buck, “Pressure-side bleed film cool-

ing: Part i—steady framework for experimental and computational results,” in
ASME Turbo Expo 2002: Power for Land, Sea, and Air, pp. 835–843, American
Society of Mechanical Engineers, 2002.
[34] D. S. Holloway, J. H. Leylek, and F. A. Buck, “Pressure-side bleed film cooling:

Part ii—unsteady framework for experimental and computational results,” in
ASME Turbo Expo 2002: Power for Land, Sea, and Air, pp. 845–853, American
Society of Mechanical Engineers, 2002.
80
[35] T. Barth and D. Jespersen, “The design and application of upwind schemes on
unstructured meshes,” in 27th Aerospace Sciences Meeting, p. 366, 1989.
[36] F. R. Menter, “Two-equation eddy-viscosity turbulence models for engineering

applications,” AIAA Journal, vol. 32, no. 8, pp. 1598–1605, 1994.
[37] C. Ansys, “Release 11.0,” ANSYS CFX-Solver Theory Guide, ANSYS, 2006.
[38] J. C. Telisinghe, Film cooling of turbine blade trailing edges. PhD thesis, Uni-
versity of Oxford, 2013.
[39] L. Hylton, M. Mihelc, E. Turner, D. Nealy, and R. York, “Analytical and exper-
imental evaluation of the heat transfer distribution over the surfaces of turbine
vanes,” NASA Report, 1983.
[40] M. J. Benson, C. J. Elkins, S. D. Yapa, J. B. Ling, and J. K. Eaton, “Effects of

varying reynolds number, blowing ratio, and internal geometry on trailing edge
cutback film cooling,” Experiments in Fluids, vol. 52, no. 6, pp. 1415–1430,
2012.
81

527710

Uploaded by

Copyright:

Available Formats

527710

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

527710

Uploaded by

Copyright:

Available Formats

AEROTHERMODYNAMICS OF TURBINE BLADE TRAILING EDGE

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

AEROTHERMODYNAMICS OF TURBINE BLADE TRAILING EDGE

submitted by TUĞBA TUNÇEL in partial fulfillment of the requirements for the

Prof. Dr. Halil Kalıpçılar

Prof. Dr. Ozan Tekinalp

Asst. Prof. Dr. Harika Senem Kahveci

Examining Committee Members:

Prof. Dr. Mehmet Cevdet Çelenligil

Asst. Prof. Dr. Harika Senem Kahveci

Assoc. Prof. Dr. Oğuz Uzol

Assoc. Prof. Dr. Metin Yavuz

Asst. Prof. Dr. Sıtkı Uslu

Name, Last Name: TUĞBA TUNÇEL

AEROTHERMODYNAMICS OF TURBINE BLADE TRAILING EDGE

SEPTEMBER 2018, 81 pages

Keywords: Turbine Blade Cooling,Trailing Edge,Pin Fin Cooling, Film Cooling,

TÜRBİN KANADI FİRAR KENARI SOĞUTMASININ

Eylül 2018 , 81 sayfa

Gaz türbinlerinin termal verimliliğinin, çalışma akışkanının türbin giriş sıcaklığına

I am so grateful to my friends Metehan Yayla, Ezgi Orbay Akcengiz, Ayşe Usta

Finally, I would like to express my deepest gratitude to my mother, my father and my

LIST OF TABLES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv

LIST OF ABBREVIATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . xix

1.3 Thesis Outline . . . . . . . . . . . . . . . . . . . . . . . . . 5

2.1 Internal Cooling . . . . . . . . . . . . . . . . . . . . . . . . 7

2.2 External Cooling . . . . . . . . . . . . . . . . . . . . . . . . 10

3.1 Mathematical Modeling . . . . . . . . . . . . . . . . . . . . 15

3.2 Numerical Modeling . . . . . . . . . . . . . . . . . . . . . . 16

3.3.1 Internal Cooling . . . . . . . . . . . . . . . . . . . 21

3.3.1.1 Boundary Conditions . . . . . . . . . 21

3.3.1.2 Geometry and Meshing . . . . . . . . 22

3.3.1.3 Solution Convergence . . . . . . . . . 25

3.3.2 External Cooling . . . . . . . . . . . . . . . . . . 26

3.3.2.1 Boundary Conditions . . . . . . . . . 26

3.3.2.2 Geometry and Meshing . . . . . . . . 26

3.3.2.3 Solution Convergence . . . . . . . . . 29

3.4 Comparison Cases . . . . . . . . . . . . . . . . . . . . . . . 30

3.4.1 Internal Cooling . . . . . . . . . . . . . . . . . . . 30

3.4.1.1 Boundary Conditions . . . . . . . . . 30

3.4.1.2 Geometry and Meshing . . . . . . . . 31

3.4.1.3 Solution Convergence . . . . . . . . . 35

3.4.2 External Cooling . . . . . . . . . . . . . . . . . . 35

3.4.2.1 Boundary Conditions . . . . . . . . . 35

3.4.2.2 Geometry and Meshing . . . . . . . . 37

3.4.2.3 Solution Convergence . . . . . . . . . 38

4 INTERNAL COOLING ANALYSIS & RESULTS . . . . . . . . . . 41

4.2 Aerodynamic Results From Comparison Cases . . . . . . . . 43

4.3 Thermal Results From Comparison Cases . . . . . . . . . . 50

5 EXTERNAL COOLING ANALYSIS & RESULTS . . . . . . . . . . 55

5.2 Aerodynamic Results From Comparison Cases . . . . . . . . 56

5.3 Thermal Results From Comparison Cases . . . . . . . . . . 65

Table 3.1 Experimental Conditions From Ref [1] . . . . . . . . . . . . . . . . 22

Table 3.2 External Flow Experimental Conditions From Ref [1] . . . . . . . . 26

Table 3.3 Internal Flow Comparison Case Boundary Conditions . . . . . . . 31

Table 3.4 Internal Comparison Case Model Dimensions . . . . . . . . . . . . 32

Figure 1.2 Cooling Methods On A Turbine Blade [2] . . . . . . . . . . . . . . 3

Figure 3.1 Schematic Of Experimental Model [3] . . . . . . . . . . . . . . . . 20

Figure 3.2 Schematic Of Validation Model a) Top View , b) Isometric View . . 21

Figure 3.3 Computational Domain Used For Internal Validation Case . . . . . 23