Aerody Design of Blade
Aerody Design of Blade
Aerody Design of Blade
2013
Abstract
The purpose of the thesis was to design a new rotor blade for the KTH test turbine
according to present design guidelines for gas turbines manufactured at Siemens
Industrial Turbomachinery in Finspng. Stage one of a real gas turbine was used as a
reference for the aerodynamic design providing a starting point for the project. Using
similar gas conditions the new rotor blade was optimized with regard to metal angles
and pitch/chord ratio at reference scale. With a satisfying geometry the new blade
was scaled back to test turbine size. The blade design could be evaluated and
modified using several different in house codes: MAC1, used for meanline design,
Beta2 for through flow design, CATO for airfoil design and Multall for 3D design.
During the project certain reference specifications restricted the design and had to be
considered. ANSYS CFX was used to analyze the final geometry in great detail not
possible in any of the other software. The new blade was first analyzed at reference
scale and then once again evaluated in Beta2, Multall06 and ANSYS CFX at test
turbine scale. As a consequence of generally having low Reynolds number in model
tests the results are not entirely comparable with the real case. Effects of transition
using different transition models were assessed providing valuable information about
the expected differences.
Acknowledgements
This thesis would not have been possible without the help of several dedicated
employees at the department of aerodynamics at Siemens Industrial Turbomachinery
in Finspng. Especially our supervisor Lars Hedlund who has provided expert
guidance and shared much of his extensive knowledge of turbine design during the
work, Christoph Mau and Tobias Pihlstrand for their CFX support, Ken Flydalen for
providing much needed guidance with design software and answers to many complex
questions and Navid Mikaillian for sharing his knowledge of the test turbine.
We would also like to thank Mats Annerfeldt for the fantastic opportunity to write
our thesis at the department, Magnus Genrup for initializing the contact and
introducing us to exciting world of turbines, Jens Fridh and Johan Dahlqvist at KTH
for letting our work become a part of the project TurboAero.
ii
Abstract ........................................................................................................................ i
Acknowledgements .....................................................................................................ii
List of figures .............................................................................................................. v
List of tables ..............................................................................................................vii
Nomenclature...........................................................................................................viii
Preface ........................................................................................................................ xi
Background................................................................................................................. 1
1 Theory .................................................................................................................. 2
1.1
Basic principles ............................................................................................ 2
1.1.1 Introduction to the gas turbine...................................................................... 2
1.1.2 The ideal gas turbine cycle Brayton cycle................................................. 2
1.1.3 The Euler work equation .............................................................................. 6
1.1.4 Velocity triangles ......................................................................................... 7
1.1.5 Blade geometry............................................................................................. 8
1.1.6 Key parameters............................................................................................. 9
1.1.7 Degree of reaction ...................................................................................... 11
1.1.8 Velocity distribution and curvature............................................................ 12
1.1.9 Zweifel........................................................................................................ 13
1.2
Vortex theory.............................................................................................. 14
1.3
Secondary flows ......................................................................................... 16
1.4
Losses ......................................................................................................... 17
2 The design process ............................................................................................ 20
2.1
1D Mean line design................................................................................... 21
2.2
2D Through flow design............................................................................. 21
2.3
2D Airfoil design........................................................................................ 22
2.4
3 D Flow analysis CFD ........................................................................... 22
3 Software ............................................................................................................. 24
3.1
1D Meanline MAC1 ................................................................................ 24
3.2
2D Through flow Beta2........................................................................... 24
3.3
Airfoil design CATO............................................................................... 24
3.4
3D analysis ................................................................................................. 24
3.4.1 Multall06 .................................................................................................... 25
3.4.2 ANSYS CFX 14.5.0 ................................................................................... 26
4 Design Method................................................................................................... 27
4.1
Design constraints ...................................................................................... 27
4.2
Initial parameter study and scaling............................................................. 28
4.2.1 Choice of reference stage ........................................................................... 28
4.2.2 Channel modification ................................................................................. 29
4.2.3 Scaling Build 4b to reference scale ......................................................... 29
4.2.4 Meanline stage matching of B0F................................................................ 31
4.3
Initial blade design ..................................................................................... 32
4.3.1 Pitch to chord ratio ..................................................................................... 32
4.4
Final design ................................................................................................ 33
iii
4.4.1
4.4.2
4.4.3
4.4.4
Optimal incidence....................................................................................... 34
Airfoil design - Curvature .......................................................................... 38
Airfoil - CATO 2D Navier Stokes ............................................................. 41
The final design - B6 .................................................................................. 44
5 Results ................................................................................................................ 47
5.1
Boundary conditions................................................................................... 47
5.2
Results - Full scale ..................................................................................... 47
5.2.1 Radial distributions - B6F vs. The reference stage .................................... 48
5.2.2 Radial distributions - B6F code comparison .............................................. 48
5.2.3 CFX results B6F...................................................................................... 52
5.3
Results - Model scale ................................................................................. 55
5.3.1 Radial distributions B6M code comparison ............................................ 55
5.3.2 CFX results B6M..................................................................................... 59
5.3.3 B6M - Off-design ....................................................................................... 62
5.4
Transition modelling .................................................................................. 67
5.5
Result summary.......................................................................................... 68
6
Conclusions........................................................................................................ 70
Future work....................................................................................................... 71
Appendix A........................................................................................................ 72
Bibliography.............................................................................................................. 77
iv
List of figures
Figure 1-1 The Brayton cycle.......................................................................................3
Figure 1-2 Velocity triangles by Siemens definition....................................................8
Figure 1-3 Blade geometry definition ..........................................................................8
Figure 1-4 Schematic velocity distribution ................................................................12
Figure 1-5 Qualitative dependence of pitch to chord ratio.........................................14
Figure 1-6 Various forces affecting a fluid element (figure from [1]) .......................15
Figure 1-7 An example of vortices caused by secondary flow (figure from [9]).......17
Figure 1-8 Influence of incidence on velocity distribution ........................................19
Figure 2-1 The aerodynamic design process. .............................................................20
Figure 2-2 Through flow computational domain of the test turbine build 4b ............22
Figure 3-1 Multall pinched tip example .....................................................................25
Figure 3-2 Multall TE cusp ........................................................................................26
Figure 4-1 Design process..........................................................................................27
Figure 4-2 Channel modification ...............................................................................29
Figure 4-3 Channel and blade fitting..........................................................................31
Figure 4-4 B6M tip recess ..........................................................................................34
Figure 4-5 B6F - Relative flow angle at blade leading edge......................................35
Figure 4-6 B6F - Relative flow angle at blade trailing edge ......................................35
Figure 4-7 B6F - Approximation of relative flow angle at leading edge ...................36
Figure 4-8 B6F - Approximation of relative flow angle at blade trailing edge..........37
Figure 4-9 B6F - Incidence at blade leading edge......................................................38
Figure 4-10 Comparison at mid section with B6F and the reference blade ...............39
Figure 4-11 Curvature distribution at tip section B6..................................................39
Figure 4-12 Curvature distribution at mid section B6................................................40
Figure 4-13 Curvature distribution at hub section B6 ................................................40
Figure 4-14 Isentropic Laval number at tip section B6F............................................41
Figure 4-15 Isentropic Laval number at mid section B6F..........................................41
Figure 4-16 Isentropic Laval number at hub section B6F..........................................42
Figure 4-17 Isentropic Laval number at tip section B6M ..........................................42
Figure 4-18 Isentropic Laval number at mid section B6M ........................................43
Figure 4-19 Isentropic Laval number at hub section B6M ........................................43
Figure 4-20 Metal angle .............................................................................................45
Figure 4-21 B6M Meridional coordinates..................................................................46
Figure 4-22 B6M ........................................................................................................46
Figure 5-1 B6F - Vane leading edge ..........................................................................49
Figure 5-2 B6F - Vane trailing edge ..........................................................................50
Figure 5-3 B6F - Blade leading edge .........................................................................51
Figure 5-4 B6F - Blade trailing edge..........................................................................52
Figure 5-5 B6F Vane - Mach at 25%, 50% and 75% of span ....................................53
Figure 5-6 B6F Blade - Mach at 25%, 50% and 75% of span ...................................54
Figure 5-7 B6F Isovolume for Ma>1 ......................................................................54
Figure 5-8 B6F - Isentropic Mach number and blade loading ...................................55
Figure 5-9 B6M - Vane leading edge ........................................................................56
Figure 5-10 B6M - Vane trailing edge .......................................................................57
Figure 5-11 B6M - Blade leading edge ......................................................................58
vi
List of tables
Table 1 Preliminary test turbine constraints after upgrade.........................................28
Table 2 Geometrical constraints.................................................................................28
Table 3 Reference stage parameters ...........................................................................29
Table 4 Scaling to reference scale..............................................................................30
Table 5 Meanline stage matching of B0F (MAC1)....................................................32
Table 6 Input and output values associated with Prof. Mamaevs correlation...........37
Table 7 Diffusion coefficients for B6F ......................................................................44
Table 8 Geometrical parameters for the final design .................................................45
Table 9 Boundary conditions .....................................................................................47
Table 10 Transition model vs. Low-Re performance prediction ...............................68
Table 11 Total to total efficiency code comparison ...................................................69
Table 12 Stage parameters (Beta2) ............................................................................69
Table 13 Section parameters (data from CATO) .......................................................69
Table 14 CFX mesh statistics.....................................................................................72
vii
Nomenclature
c
cp
[m/s]
[kJ/kgK]
absolute velocity
specific heat capacity at constant pressure
cv
Ma
La,
n
N
p
Re
T
u
w
h
w
W
q
Q
s
d
T
v
P
[kJ/kgK]
[-]
[-]
[rpm]
[MW]
[bar]
[-]
[K]
[m/s]
[m/s]
[kJ/kg]
[kJ/kg]
[J]
[J/kg]
[J]
[kJ/kgK]
[-]
[Nm ]
[m^3/kg]
[bar]
[-]
heat
specific entropy
denotes an incremental change
torque
specific volume
pressure
thermodynamic efficiency
m, m&
Rs
[kg/s]
[kJ/kgK]
[-]
[J]
mass flow
specific gas constant
ratio of specific heats
total energy
hss
[s]
[-]
[-]
[m/s]
[rad/s]
[mm]
[-]
[kJ/kg]
time
Zweifel coefficient
pressure ratio
rotational speed
angular speed
radius
velocity ratio
static to static enthalpy drop
[-]
[]
[]
[]
[-]
th
E
t
ZW
PR
U
ss
diffusion coefficient
absolute flow angle
relative flow angle
radial clearance
efficiency
viii
h
p
TI
Cf
[-]
[-]
[-]
[-]
[Pas]
[-]
[kg/m^3]
[-]
[-]
[-]
[-]
[Pa]
[%]
[-]
is
0
1
2
u
a
r
*
isentropic
vane inlet
vane outlet
blade outlet
tangential
axial
radial
total state
denotes quantities per unit time
denotes total state
total to total
.
0
tt
total to static
critical state
denotes vector
ts
cr
denotes wall
c
b
B
d2
r1
r2
t
1m
2m
1
2
i
AR
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
[mm]
[]
[]
[]
[]
[]
[]
[-]
throat width
real chord
axial chord
trailing edge diameter
leading edge radius
trailing edge radius
pitch
inlet metal angle
outlet metal angle
uncovered turning
leading edge wedge angle
trailing edge wedge angle
incidence
aspect ratio
ix
The figures below show the used definitions regarding blade geometry and flow
angles.
Preface
The thesis consists of two main parts over seven sections. The first part is pure
theory based and the second describes the actual work and result of the thesis. A
brief summary of the content follows:
xi
Background
An air driven test rig owned by Siemens Industrial Turbomachinery but situated at
KTH with typical steam turbine blading has been the subject of this thesis. The rig is
originally designed for operation with up to three turbine stages but only two stages
have ever been used. For the current blading three configurations are possible, build
4a with only the first stage, build 4ab with both stages in series and build 4b with
only the second stage. This thesis is a part of the project TurboAero, a collaboration
between Siemens, KTH and GKN, where build 4b is subject to a number of
upgrades. Afterwards it will be possible to operate at conditions typical for a first
stage of an industrial gas turbine excluding the temperature level.
In conjunction with the upgrades this thesis aims to design a new stage for the test
turbine that is more representative of current Siemens gas turbines. It will be based
on an existing stage from Siemens product portfolio in order to attain a design
complying with present design rules.
The current scope of the project only enables the rotor to be switched and the
existing vane will remain as a consequence of limited funding. This is one of the
major limitations of the new stage since it will restrict the design to some degree.
Furthermore the maximum temperature and rotational speed are constrained of
structural and safety reasons.
As this thesis will become available to the public domain some results and
references have been excluded maintaining sensitive material confidential. However
this will not in any way affect the methodology followed or the conclusions.
1 Theory
1.1 Basic principles
1.1.1
The main contributor to the popularity of the gas turbine is without question the
aviation industry. For many years the piston engine was the only possible choice of
air born propulsion. During the Second World War a lot of effort was put in to
developing faster aircraft which in the end led to the first commercial gas turbine
engines. They were lighter and a lot more powerful compared to a piston engine of
equivalent size. Nowadays using anything else than a gas turbine for aircraft
propulsion is almost unthinkable. During the second half of the 20th-century gas
turbines were beginning to be used more and more in stationary applications such as
mechanical drive and power generation. Gas turbines can be started and ramped very
fast which can be of prime importance in certain applications.
When coupled to a steam turbine and used in a combined plant the efficiency can be
on par with the best of piston engines.
1.1.2
In essence a gas turbine consists of three major parts: a compressor, a combustor and
a turbine. First air is sucked in to the compressor part of the gas turbine where there
is an increase in pressure. The flow of air is then mixed with fuel and ignited in the
combustor. Finally the hot gas stream is expanded in the turbine and then expelled
through an exhaust pipe. Even though the vast majority of gas turbines operate
accordingly to the aforementioned process it is absolutely not the only way to make a
gas turbine work. Several other ideas have emerged over the years including making
use of another working media with external combustion etc.
A gas turbine can be considered a cyclic device hence it is often analyzed using a
thermodynamic cycle. By making certain assumptions for example; steady state
operation, isentropic turbine and compressor, no pressure loss, constant mass flow,
perfect gas, no difference in fluid velocity across the components and so forth,
simple analytical expressions of efficiency and specific work can be derived.
It is common to make use of a closed cycle approximation when making the
calculations.
In reality this kind of operation is seldom used because of the difficulties associated
with heat transfer from the combustion to the working fluid. With the simplifications
mentioned above the cycle is said to be ideal and often called the Brayton cycle after
its inventor. In Figure 1-1 below the Brayton cycle is described using a temperature
entropy-diagram. Between point one and point two the compression process takes
place consuming work and raising the temperature of the working media. Energy is
added in the form of heat when moving from point two to three causing a large
increase in temperature. Work is extracted during the expansion from point three to
point four lowering the temperature. Heat is then removed lowering the temperature
back to the starting value in order to complete the cycle.
dE system
dt
=0
Rearranged:
E& in = E& out
Mass cannot be created nor destroyed, this simple fact will state that all the mass
entering a control volume must also come out during steady state operation:
m& in m& out =
dE
= 0 m& in = m& out
dt
Below the governing equations, the energy equation and the continuity equation for a
steady flow process will be used in the analysis of the Brayton cycle. By treating
each part of the cycle separately using a control volume approach on a per unit mass
basis:
Compressor:
win = h2 h1
(1)
wout = h3 h4
(2)
q in = h3 h2
(3)
Turbine:
Combustor:
th =
Wnet ,out
th =
wnet ,out
Qin
(4)
qin
(5)
(6)
th =
h3 h4 (h2 h1 ) h3 h4 h2 + h1
=
h3 h2
h3 h2
(7)
th =
T3 T4 T2 + T1
T T
= 1 4 1
T3 T2
T3 T2
(8)
(9)
Since both the compression and expansion process are considered to be isentropic the
following simplification can be done:
Tds = 0 dh = vdP
By using dh = c p dT and the ideal gas law Pv = Rs T :
dP c p dT
=
P
Rs T
(10)
cp
(11)
cv
Rs = c p c v
(12)
cp
Rs
c
1
=
= 1 v = 1
cp
cp
Rs
(10) rewritten:
1
1
dP
dT
=
P 1 T
(13)
dP
dT
1 P = 1 1 T
2
(14)
T
T2
T1 , T3 = 3 T4
T1
T4
(15)
P
T2 = 2
P1
P
T1 , T3 = 3
P4
T4
P2 P3
=
= PR
P1 P4
th = 1
PR
T4 T1
1
= 1
1
1
T4 PR T1
PR
(16)
From (16) it is obvious that the thermal efficiency of the Brayton cycle is only
affected by the pressure ratio for a given working fluid. An expression for the
specific work will yield a dependency of both pressure ratio and maximum
temperature. The temperature also influences the efficiency when component losses
are taken into account. However even so the efficiency is much more dependent on
the pressure ratio than maximum temperature. Whenever dealing with non ideal
processes the concept of isentropic efficiency is very commonly used. It is defined
for an adiabatic turbine as the actual work output divided by the ideal work output
which corresponds to the actual enthalpy drop divided by the isentropic enthalpy
drop. Two different isentropic efficiencies exist, total to total and total to static
defined below:
tt =
h0in h0out
h0in h0out ,is
(17)
ts =
h0in h0out
h0in hout ,is
(18)
The reason for defining two different isentropic efficiencies is that the outlet kinetic
energy sometimes must be regarded as a loss. For example in a turbine consisting of
many stages, the exit kinetic energy for all but the last stage can be used downstream.
However for the last stage this is not the case as the kinetic energy possessed by the
flow will be of no use outside of the turbine hence it must be considered a loss of
energy.
1.1.3
All turbomachinery is based on the basic principle derived by Euler in the 18th
century. From elementary mechanics the interaction between force and acceleration
is well known from Newtons second law of motion, which states that the net force
equals the time rate of change of momentum. There is a very similar equation
describing how torque and angular momentum is related. That is the net torque will
be equal to the time rate of change of angular momentum. In a generalized
turbomachine fluid enters at radius r1 with the tangential speed cu1 and leaves at
radius r2 with the tangential speed cu2. Under the premise of steady flow the torque in
the tangential direction can be written:
T = m& (r2 cu 2 r1cu1 )
(19)
By definition power is work per unit time that is force times speed which for a
rotating device equals torque T times angular velocity :
N = T = m& (r2 cu 2 r1cu1 ) = m& (r2cu 2 r1cu1 ) = m& (U 2 cu 2 U 1cu1 )
(20)
N
= (U 2 cu 2 U 1cu1 )
m&
(21)
1.1.4
Velocity triangles
When studying turbines so called velocity triangles are very often used in order to
get an understanding of how the flow interacts with the blades. The fluid enters the
stationary part of the stage, the stator, where it is usually accelerated to some degree
in the absolute frame of reference. It then enters the rotating part of the turbine, the
rotor. Here the flow normally experiences acceleration in the relative frame of
reference. For most gas turbine stages the fluid will accelerate to some extent in both
rotor and stator part but there are turbines which behave differently. For example
many steam turbines can be of so called impulse design, that is all the acceleration
occurs in the stator. In Figure 1-2 velocity triangles for a typical turbine stage have
been drawn. Adding the blade speed and relative velocity vectorially yields the
absolute velocities:
C = U +W
The angles corresponding to velocity vectors are measured from the tangential plane
which is common in Russian and German literature. In the figure above the axial
velocity is constant through the stage as seen from the velocity triangles having equal
height. This is a reasonable approximation since in reality the axial velocity generally
varies very little through the stage in comparison to the other velocity components.
The approximation greatly simplifies the mathematical derivations of expressions
relating the shape of the triangles to different key parameters defining the stage.
1.1.5
Blade geometry
In this section some of the common parameters used to describe blades geometrically
will be presented. In Figure 1-3 below an arbitrary blade with definitions can be
seen:
The difference between metal and flow angle at inlet is called incidence and at outlet
deviation:
i = 1m 1
= 2m 2
1.1.6
Key parameters
In order to define a gas turbine stage the following set of parameters are often used:
Stage loading is a quantity often used in gas turbine practice and defined as the total
enthalpy drop divided by the square of blade speed:
h0
U2
(22)
Velocity ratio can be seen as the inverse of stage loading and it is often used when
studying steam turbines. Defined as the blade speed divided by the square root of two
times the isentropic static to static enthalpy drop:
U
ss =
(23)
2hss
Flow coefficient is defined as the axial component of the absolute speed divided by
the blade speed:
ca
U
(24)
Flow capacity is defined as the mass flow times the square root of total temperature
divided by the total pressure:
Flow Capacity =
m& T0
p0
(25)
hrotor
hstage
(26)
h =
p rotor
p stage
(27)
For an isentropic process (26) and (27) are equal. This can be derived using Gibbs
equation (9) with Tds = 0 yielding dh = vdp. Using the result in (26) and assuming
constant specific volume (and hence constant density):
h =
(28)
Both definitions of stage reaction are used in the literature and differ typically a
couple of percent but they give the same type of information.
Two dimensionless quantities are very often used when studying the velocity near
the surface of a blade. The first one is the Laval number defined as the local velocity
divided by the so called Laval velocity:
c
ccr
(29)
Using the steady flow energy equation for an arbitrary adiabatic channel between
static and stagnation states yields:
c2
h0 = h +
(30)
2
For a perfect gas with constant c p equation (30) can be rewritten:
T0 = T +
c2
2c p
(31)
c = 2c p (T0 T ) =
2Rs
(T0 T )
1
(32)
(33)
This critical speed can also be expressed using (32) and equated to (33):
ccr =
2Rs
(T0 Tcr ) = Rs Tcr
1
(34)
Tcr
2
=
T0 1 +
Solving for Tcr and inserting into (34):
10
(35)
ccr =
2Rs T0
1+
(36)
The ratio of the local speed and the critical speed yields the Laval number, also
known as the characteristic Mach number. For a more thorough explanation see
Anderson [12].
The second dimensionless quantity often used is the Mach number defined as the
local velocity divided by the speed of sound:
Ma =
c
c
=
a
Rs T
(37)
T0 p0
=
T p
(38)
Making use of equation (31) and (37) the left hand side of equation (38) can be
expressed:
T0
1
= 1+
Ma 2
T
2
(39)
By inserting the right hand side of equation (39) into equation (38) an expression for
the so called isentropic Mach number can be derived:
1
p0
2
Mais = 1
p
1
(40)
Equation (40) expresses a Mach number often used when studying velocity
distributions along a blade surface. The isentropic Laval number is also often quoted
in the same context. It is defined neglecting the influence of friction or viscosity and
heat transfer present in the boundary layer i.e. at conditions present for an isentropic
process.
1.1.7
Degree of reaction
The degree of reaction describes how the expansion is divided by the stator and rotor
blade. A low degree of reaction corresponds to a larger part of the acceleration taking
place in the stator than in the rotor. Consequently a high degree of reaction results in
the reverse situation. Looking at a simplified stage as the one illustrated in Figure
1-2, mathematical expressions linking degree of reaction and angles can be derived.
A common choice is the 50 percent reaction design implying symmetrical velocity
11
triangles, as in Figure 1-2. The exact choice of reaction is not that critical when
considering the efficiency of the stage. Many different philosophies regarding the
choice exist and designs with degree of reaction far removed from 50 percent can
have high efficiency.
The reaction varies along the span because of varying static pressure and static
temperature. As outlined in section 1.2, whenever the flow has a whirl component
there must be a pressure gradient for the condition of radial equilibrium to be
fulfilled. Looking at a general stage this will mean that the static pressure increases
with radius. Corresponding to this change is the velocity distribution that will vary in
the opposite way. When designing a stage it is important to check how the degree of
reaction varies with radius in order to achieve good efficiency since a too low value
in the hub or too high value in the tip tend to be detrimental.
1.1.8
12
D=
max
1
2
(41)
Where max is the maximum isentropic Laval number on the suction side and 2 is
the average isentropic Laval number downstream of trailing edge. The design should
comply with D Dopt [10]. Depending on the value of 2 different values of Dopt
will be used.
The area enclosed by the lines in Figure 1-4 can be thought of as a measure of blade
loading. Generally the pressure distribution is almost the complete inverse of the
velocity distribution and this integrated along both surfaces will yield the blade
loading. Depending on the distribution of loading the blade can be said to be front,
mid or aft loaded. The choice of loading is often not explicitly given but depends
very much on the design philosophy followed.
In general it is suggested that the velocity should increase smoothly along the
surface right up to the point of diffusion giving a continuous acceleration. This
provides stability to the flow and avoids unnecessary separations.
When the flow becomes transonic or supersonic shocks can appear leading to
increased losses. Since there is no real advantage of having supersonic flow it is
often avoided if possible.
A common geometrical parameter used in blade profile design along with velocity
distribution is the curvature. It is calculated as a function of the first and second
derivative of the polynomials used for describing the profile mathematically, and it
should be continuous for the whole profile. The curvature should decrease
monotonically towards the trailing edge providing the possibility of having
controlled diffusion with minimized risk of separation.
1.1.9
Zweifel
A very common parameter used in turbine design is the Zweifel blade loading
coefficient. It is used in order to get a value of the pitch/axial chord ratio which
yields minimum losses. Once this coefficient is determined, one can calculate the
pitch and hence the blade number if the axial chord is known. If certain assumptions
are made for example constant density, constant axial velocity and so on, the
coefficient can according to Moustapha et al. [2] be expressed as in equation (42) and
equation (43). Note the nomenclature used below. The flow angles are defined using
axial definition, s and c a denote the pitch and axial chord respectively.
For the stator using axial definition:
s
ZW = 2 (tan ( 1 ) + tan ( 1 )) cos 2 ( 2 )
ca
13
(42)
s
ZW = 2 (tan ( 2 ) + tan ( 3 )) cos 2 ( 3 )
ca
(43)
A qualitative reasoning for finding the pitch/axial chord ratio that will yield
minimum losses can be as follows. Increasing the pitch/axial chord ratio for a given
radius and axial chord means fewer blades that are spaced further apart. A
consequence of this is less area being wetted by the fluid and hence lower frictional
loss. On the other hand for a given stage work output each blade will have to
contribute with a larger quota of work. This corresponds to a higher blade loading
which will have greater diffusion losses. A schematic figure of how the pitch/axial
chord ratio affects the frictional and diffusion losses can be seen in Figure 1-5. From
this it is understood that there is a certain pitch/axial chord ratio which will give
minimum losses.
A Zweifel coefficient of 0.75-0.85 used to be quoted as optimal according to
Moustapha et al [2]. However nowadays it is not uncommon with coefficients
beyond unity. The main contributors to permitting such high values are advanced
tools in blade design. Even though blade designers today make use of sophisticated
fluid dynamics software during development rather than certain values of the Zweifel
coefficient, it is still used to get a rough estimate of what to expect.
14
pressure will most certainly affect the velocity varying in a similar manner. The
reason for this will be explained below along with a design philosophy that can be
used to construct the velocity triangles at any arbitrary radius.
A fluid element will generally have velocity components in all three directions
radial, axial and tangential. Often the radial velocity is very small compared to the
other two especially stages of high hub to tip ratio. A force balance can be derived
taking in to account all the forces associated with a fluid element, see Figure 1-6 [1]
Figure 1-6 Various forces affecting a fluid element (figure from [1])
In the radial direction the fluid element experiences centripetal forces resulting from
the tangential velocity, the curvature of the streamline as well as the radial
component of force needed to accelerate along the streamline. The resulting pressure
force in the radial direction is derived from the figure above. By equating the
resulting inertial and pressure force using the nomenclature given by Figure 1-6 one
arrives at:
dc
1 dp c w2 c s2
=
+ sin s + s sin s
dr
r
rs
dt
(44)
Often the terms including rs and s can be neglected since the streamline radius is
large and the streamline angle is small due to c r being considerably smaller than the
axial and tangential velocity components resulting in the famous Euler n-equation:
1 dp c w2
=
dr
r
(45)
Basically the equation above states that whenever the fluid is being deflected and a
whirl component is induced a pressure gradient in the radial direction is needed to
balance the inertial forces. This equation explains why the static pressure increases
from root to tip in a turbine.
Equation (45) will now be used to analytically derive an expression linking the
tangential velocity to the radius which will define the velocity triangles at any
15
dh0 dh
dc
dc
=
+ ca a + c w w
dr dr
dr
dr
(46)
From the Gibbs equation the static enthalpy for an incremental change with radius
can be derived and substituted above. Neglecting second order terms and making use
of the Euler n-equation:
dh0
ds
dc
dc
c2
=T
+ ca a + c w w + w
dr
dr
dr
dr
r
(47)
(48)
16
Figure 1-7 An example of vortices caused by secondary flow (figure from [2])
The boundary layer flow at the leading edge of the blade will split and form what is
called a horseshoe vortex because of its shape. Downstream passage vortices start to
roll up near the blade surface. Generally secondary flows are mainly dependent of
aspect ratio and the flow turning. The main flow will be less affected by secondary
flows considering blades of high aspect ratio compared to blades of low aspect ratio.
This is a consequence of secondary flows being an end wall phenomenon influencing
a relatively large part of the passage in the latter case. More flow turning means
larger pressure gradient in the blade passage resulting in a greater potential for
secondary flow formation.
1.4 Losses
According to Moustapha et. al. [2] there exist many types of loss generating
mechanisms in a turbine which can be difficult to differentiate since they often
interact in a very complex manner. However three major contributions to the overall
loss can be identified:
17
In the rotor part of a stage an additional major loss is present, tip loss. It is the result
of fluid leaking past the rotor blade contributing with little or no work. The reason
for this is the tip clearance needed as a result of manufacturing tolerances and
operational constraints. The blade will elongate under the centrifugal and thermal
load present. Depending on the temperature and material used, blade and casing
might differ in thermal growth. Two main leakage paths exist depending on if the
blade is shrouded or not. When designed with a shroud, a roof is present at the tip
making it possible to lock the rotor blades together. This makes it impossible for
flow leaking from pressure to suction at the blade tip. Instead the main leakage path
is over the shroud from leading to trailing edge. Compared to an un-shrouded blade
the tip clearance loss can be considerably lower, however mass has been added at the
worst possible location since the stress from centrifugal load will be great at the
blade root. For an un-shrouded blade the tip leakage flow from pressure to suction
side will form a vortex that together with the secondary flow vortices can create
complex flow patterns. Locally at the tip the flow leaking from pressure to suction
side will cause underturning. The main flow will in this region experience negative
deviation. Considering shrouded blades there will be fewer tendencies for the
behavior described above.
One important parameter influencing the losses is the incidence which has great
consequences for the blade design. Both profile and secondary losses change with
incidence to some extent. The optimal value of incidence minimizing the losses at
design point is often called design incidence. This is not achieved at zero degree
incidence since the pressure field resulting from the blades being aerodynamically
loaded extends somewhat outwards from the leading edge. At off design conditions
the blades will experience an incidence angle somewhat different than at design
conditions. The velocity distribution will be greatly influenced by incidence causing
velocity peaks near the leading edge. Subsequent diffusion can lead to separation if
taken too far leading to an increase in losses. Usually negative incidence is less
detrimental for the aerodynamic performance compared to positive incidence. The
reason for this can be explained considering the velocity distribution for the two
cases, see Figure 1-8. At positive incidence the stagnation point will move some
distance along the pressure side. This will require the flow to turn considerably
before reaching the suction side resulting in a high velocity spike near the leading
edge. The flow will experience substantial diffusion which could lead to separation.
At negative incidence the flow will reach a high velocity on the pressure side instead.
The turning done by the flow and therefore the blade loading is less than for positive
incidence becoming less susceptible of separation and associated losses.
18
19
20
21
Figure 2-2 Through flow computational domain of the test turbine build 4b
22
accuracy of the solution. The purpose of the CFD analysis is to validate the
performance prediction from the through flow analysis or to enlighten unknown flow
behavior that has not been accounted for. Based on the CFD result the blade design
may have to be reconsidered, in which case the designer has to change the airfoil
design or even revisit the through flow or meanline design. It is clear that a good
initial design saves a lot of time and the more work and thought spent at the earlier
steps, the more likely it is that the designer does not have to redo the whole
procedure. Finally it should be said that the results from a CFD calculation are not
more accurate than the boundary conditions, which in many cases are not fully
known. Even apart from this there are bound to be approximations, both in the model
of the blade geometry and in the CFD code with regard to turbulence modeling,
numerical discretization etc. and the actual performance of the blade will differ from
the calculated result.
23
3 Software
Throughout the thesis several different codes have been used. A brief description of
each program is presented here.
3.4 3D analysis
For the 3D flow analysis two different programs have been used; Multall which is a
program based on a code developed by John Denton at Cambridge and the
commercial CFD software ANSYS CFX. The programs differ greatly with regards to
complexity and serve totally different purposes in the design process. Multall is a
relative simple program fine tuned for turbomachinery applications and relies more
on robustness and simplicity than the ability to capture every detail in the flow.
These properties make it easy to create a working model and allow Multall to be used
as a design tool at an early stage in the design process. CFX however, being a
commercial software, offers more with regards to complexity and flexibility. It is
more suited at a later stage in the design process since a good model is very time
consuming to create and to solve. Due to the difference in time cost CFX models
were only created for what is to be considered as the final design of the test turbine
stage. Multall, however, was present much earlier in the design process and used to
iterate between 2D and 3D calculations until a final design was achieved.
One of the main purposes of the CFX model was to investigate to what extent the
boundary layers are laminar at model scale, and no reliable method to model
24
Multall06
Multall06 is a steady-state code and only one cascade passage was modeled.
The turbulence model used in Multall is the Baldwin-Lomax model which is a zero
equation model used together with wall functions. The Baldwin Lomax model is well
suited for attached high speed flows and is known to be robust and reliable for
turbomachinery applications [5]. The boundary layers are assumed to be fully
turbulent since no verified transition criterion is implemented in the code. Tip
clearance is modeled with a pinched tip method where the blade progressively thins
out across a number of cells near the casing. This method enables simple and quick
meshing but may give unrealistic flow physics with high Mach numbers at the tip if
care is not taken. An example of how the pinched tip method was used can be seen in
Figure 3-1. The trailing edges are modeled with a so called cusp to force the flow
to separate at the trailing edge. Without cusp, the flow stays attached and is
accelerated around the curvature of the trailing edge, again causing unrealistic
physics and high Mach numbers. Figure 3-2 shows the cusped trailing edge of one of
the modeled blades.
25
3.4.2
The mesh for CFX was created by the meshing software TurboGrid which is
especially developed for meshing of turbomachinery parts. Since one of the purposes
of creating a CFX model to begin with was to investigate the effects of turbulent
transition, the mesh had to be fine enough to resolve the boundary layers. TurboGrid
is limited to hexahedral mesh (structured mesh) which makes it unavoidable to have
very large element aspect ratios and element expansion rates if sufficient y+ values
are to be achieved. This particularly becomes a problem at the tip clearance where
the velocity gradients are very large and the mesh quality is inevitable poor. To
overcome initial instabilities at the tip clearance region every case was run with a
local timescale during the initial iterations. Later on the timescale was increased
and switched to the same for the whole domain to ensure convergence. For the cases
at full scale this method was not enough and the mesh refinement at the tip clearance
had to be sacrificed to reach a solution. Further details are presented in Appendix A.
The SST k- model with standard values was chosen to model the turbulence. The
model is widely used for turbomachinery flow simulations and also compatible with
the - transition model. CFX allows for transient solution where the whole stage can
be modeled. However, such approach is very demanding computational wise and it
was sufficient to model just one passage and to assume steady state.
26
4 Design Method
The method of this thesis has in many ways followed the steps of the design process
described in section 2 though with some important deviations. As the task was to
develop a blade for an existing turbine stage and with the aim to mimic a full scale
turbine, many conditions were already decided in beforehand. With the geometry of
the KTH test turbine given, the most fundamental geometrical parameters of the gas
channel were fixed. Since the existing vane will remain the flow and geometrical
characteristics of the stage will be limited. For example, the trailing edge thickness of
the vane is not representative of a cooled gas turbine stage and these restrictions have
to be considered during the blade design.
It is important to point out that the presented design process in this report is only
intended so serve the purpose of describing the main methodology used to reach the
final design proposal. As the project went on new constraints were set and earlier
known specifications had to be reconsidered. Consequently the actual design process
was all but straightforward with several blade versions which were updated and
optimized as new information become known. After every major change in blade
geometry all the design criteria had to be reconsidered and controlled once again, e.g.
optimal pitch to chord, number of blades etc. It serves no purpose to, in detail,
describe all the minor adjustments made during the work but instead the main design
steps are described together with the major changes that occurred during the design
which ultimately resulted into the final design. A condensed version of the actual
procedure is seen in Figure 4-1. The following sections will describe the different
design phases in further detail.
27
Parameter
Pressure ratio [-]
Inlet temperature [K]
Outlet static temperature
[K]
Mass flow [kg/s]
Shaft speed [rpm]
Maximum value
2
333
283
4.7
13000
Limiting factor
Outlet temperature
Safety regulation
Dew point at outlet
pressure
Compressor rating
Structural
Earlier calculations done at Siemens suggest that neither the mass flow nor the shaft
speed will exceed their maximum allowed values for any relevant test case for the
test turbine [7]. The pressure ratio however is greatly limited by the inlet temperature
constraint. Since the outlet temperature is to be held above the dew point the inlet
temperature will increase if the pressure ratio is to be increased. A higher pressure
ratio is desirable if the stage is to be redesigned to better represent a gas turbine stage
and because of this the test turbine casing may have to be insulated to allow a further
increase in inlet temperature.
Pure geometrical constraints are the channel height, the axial distance between the
vane and blade to allow for probe traversing and the axial width of the blade disc, see
Table 2. During the design of the new blade further consideration had to be made
regarding the minimal length of the chord of the new blade. This since the limitations
of the manufacturing were not fully known at the time of the design.
Parameter
Tip clearance [mm]
Max axial blade width
[mm]
Axial gap between vane
TE and blade LE at mid
span [mm]
Value
0.3
~ 25
10
28
Reference stage
2.155
1.385
0.412
0.412
0.818
[-]
[-]
p,mid [-]
[-]
Ma2rel [-]
4.2.2
Channel modification
The first change to the current model of the test turbine build (build 4b) was to set
the blade shroud hade angle to zero as seen in Figure 4-2. This in accordance with
how the gas channel commonly is designed for the first rotor blade due to tip
clearance limitations with respect to axial displacement during transients.
Channel comparison
4b channel
New channel
Compared to the original 4b, the change to the new channel decreased the outlet area
which, for given inlet conditions, increase the acceleration over the rotor blade and
therefore causes an increase in reaction. Since the existing build has a low degree of
reaction typical for steam turbines, this is a change in the desired direction.
4.2.3
With a suitable turbine set as reference the next step was to evaluate the most
appropriate way to perform the actual geometrical scaling. The radial coordinates
were scaled with a radial factor kr:
rhub ,ref
kr =
rhub , 4b
29
vaneoutlet
This will result in a very short axial chord when the blade is scaled back to test
turbine dimensions because of the aspect ratio difference. As a consequence of the
uncertainties regarding manufacturing of blades that small, it was desirable to
increase the axial width of the blade. With this in mind the axial chord of the new
design was set to the maximum allowed in the test turbine. This approach will result
in a blade with much larger axial chord and smaller aspect ratio than the reference
blade. Even though undesirable this deviation from the reference was said to be
necessary.
The new blade will be referred to as BxF/BxM where B - blade, x number
indicating the design version, M/F stand for model and full scale design respectively.
The first design version introduced here is B0F where no modification but pure
scaling was performed to the original blade geometry.
Since the aspect ratio of the reference blade could not be kept in the B0F design, the
airfoil geometry had also to be scaled in order to conserve the relative profile
geometry. The profile factor was defined as:
k profile =
k r Bhub , 4b
Bhub ,ref
The relative conicity of the reference blade was to be maintained which was achieved
by calculating the axial chord at the tip section by the factor kc:
kc =
Btip ,ref
Bhub ,ref
In Table 4 a geometrical summary of the first design B0F can be seen. Figure 4-3
shows the channel and blade fitting to the reference scale. As a first assumption the
axial distance between the vane and the blade at mid span is kept the same as in the
reference. The decrease in hub radius at the trailing edge for the reference casing
shown in Figure 4-3 is neglected in further calculations and the hub radius is said to
be constant.
rhub
ref blade 1
rhub ,ref
B0F
k r rhub , 4b
rtip
rtip ,ref
k r rtip , 4b
Bhub
Bhub, ref
k r Bhub , 4b
Btip
Btip ,ref
k c Bhub , BxF
LE diameter (mid)
d 1,ref
k profile d 1,ref
TE diameter (mid)
d 2,ref
k profile d 2,ref
30
4.2.4
The next step was to determine the pitch to chord ratio which in turn will decide the
number of blades. Many correlations exist for calculation of optimal pitch to chord
ratio, e.g. the Zweifel coefficient discussed in section 1.1.9, and SIT has their own
versions used in this thesis. To provide the relevant parameters for the correlations a
meanline stage matching of B0F was performed. The calculation was done with the
same number of blades and at the same rotational speed as the reference stage. The
axial chord at mid span was at this stage approximated as a mean of the hub and tip
values and the real chord was calculated by conserving the B/b-ratio of the reference.
Special concerns had to be considered regarding the cooling flows of the reference
stage. Since it is too complicated to recreate such flows in the test turbine stage B0F
was modeled without any cooling injections. In order to preserve the characteristics
of the reference turbine the inlet temperature and the blade throat area had to be
adjusted properly due to lack of cooling. It was decided that an appropriate approach
was to adjust the inlet temperature for B0F so that the same relative total temperature
at the blade inlet was achieved. Ideally it was desirable to achieve matching degree
of pressure based reaction and at the same time have matching blade outlet Mach
number, Ma2,rel, to get a good agreement of stage characteristics. This however was
not possible within the limits of the design constraints and it was decided to prioritize
the matching of the outlet Mach number. To compensate for the smaller mass flow
through the B0F blade passage the throat area had to be decreased if the same blade
outlet Mach number was to be achieved.
This was done by manual modification of the throat area in the meanline code input
file until the same relative Mach number at the blade outlet was reached. The result
from the meanline stage matching is presented in Table 5. Although without cooling
B0F experiences very similar flow conditions compared to the reference stage.
31
[-]
[-]
p,mid [-]
[-]
1 []
1 []
2 []
2 []
Ma1abs [-]
Ma1rel [-]
Ma2rel [-]
La2rel [-]
Re1abs [-]
Re2rel [-]
Reference stage
2.155
1.385
0.412
0.412
14.52
63.00
71.23
19.91
0.787
0.206
0.818
0.836
29.75e5
15.90e5
B0F
2.154
1.363
0.404
0.431
14.72
55.17
71.15
20.60
0.793
0.237
0.817
0.835
29.14e5
21.20e5
In order to determine the optimal pitch to chord ratio, t 0 nm , with regard to flow
angles, Laval number and trailing edge thickness Prof. Mamaevs guidelines [4] were
used. The correlation is a function of three coefficients, t 0 nmO is the optimal pitch
chord for a reference exit Laval number, K KP is a correction for a reference trailing
edge diameter and t 0 nm is a correction for exit velocity:
t 0 nm = f (t 0 nmO , K KP , t 0 nm )
The first coefficient on the right hand side is dependent on the inlet and outlet flow
angle, the second one on trailing edge diameter and the third one on exit Laval
number. The coefficients are determined by reading values of graphs using input
flow parameters from previous MAC1 results together with the trailing edge
diameter defined above in Table 4. As soon as the pitch chord ratio is determined the
number of blades can be found for a given chord and radius. The only geometric
parameter known at this stage is the axial chord at the hub. In order to retain the
characteristics of the reference it was determined that B0F should have the same
conicity factor k c . With kc the axial chord at tip and mid can be calculated. The
value at mid section is simply an average of the hub and the tip section. The ratio of
real chord to axial chord at mid has been conserved just like the conicity factor:
32
b
b
kr =
=
B mid ,ref B mid , B 0
Both k c and k r have been calculated from a CAD-model of the reference stage. By
using the ratio above the chord at mid section for B0F can be determined and it was
used together with the axial chord at mid in the MAC1 calculations. The
corresponding radius at mid is the average value of hub and tip section from input to
the 1D calculation.
With known values of radius, chord and pitch chord ratio the number of blades can
be calculated:
Number of blades =
2Rmid
45
t
b bmid
opt
The data used as input to calculate the pitch chord ratio was generated with the same
number of blades as the reference since the program required this parameter in order
to run. After optimizing, the number of blades is now 45 and new data will be
generated to check the values of different flow parameters corresponding to 45
blades. Fortunately the number of blades does not influence the flow angles or the
exit relative Laval number to a very large extent. Subsequent calculations with
updated input values yield the same result and the blade number have been fixed
throughout the rest of the design process.
In the used airfoil profiling software CATO the real chord cannot be set. The reason
for this is the way the geometrical parameters are defined in the software, the real
chord is calculated as an output geometrical parameter among others. Hence the pitch
to chord ratio of the final design will differ to some degree compared to the optimal
value.
33
Tip recess
4b channel
New channel
Vane
B6M Blade
The decrease of the casing radius at the blade naturally decreases the blade throat
area, accelerating the flow more than intended. An increase in acceleration in the
rotor increases the degree of reaction and the outlet Mach number. To compensate
for this effect the throat area had to be increased until the target Mach number once
again could be achieved with the new casing design. This was done by rotation of the
blade, causing an increase in blade outlet metal angle. The blade geometry with
adjustments to the new channel is from now on referred to as B6F/B6M for full and
model scale respectively. Earlier versions have only acted as necessary design steps
and it serves no purpose to show results from any of these. The design methodology
regarding inlet metal angle and curvature for the B6 geometry is presented in the
following sections 4.4.1 and 4.4.2.
4.4.1
Optimal incidence
34
Below some of the results from Beta2 and Multall (3D-code) can be seen for B6F in
Figure 4-5 and Figure 4-6. The values of relative flow angles at hub, mid and tip
were used as input to Prof. Mamaevs correlations.
Fullscale Blade LE - Relative flow angle
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
Multall06
Beta2
0.1
0
40
45
50
55
60
65
70
75
80
1 []
0.7
0.6
0.5
0.4
0.3
Multall06
Beta2
0.2
0.1
0
15
16
17
18
19
20
21
22
2 []
35
23
24
In order to obtain the inlet relative flow angles at hub and tip from the Multall06
results in Figure 4-5 a line is inserted from an extrapolation between two points at
approximately 10 and 90 percent of normalized radius. The reason for this is the
behavior at the extremes. At the region towards the blade hub and tip many complex
flow patterns affect the angles considerably as can be seen in Figure 4-5. It would
have been unreasonable to design exactly to these since they do not represent the
flow in general at hub and tip. This can be seen when comparing to the Beta2 results.
The extrapolated line is seen in Figure 4-7. In the same figure the angle at mid can be
found simply by reading of the x-axis corresponding to a normalized radius of 50
percent.
Full scale blade LE - Relative flow angle
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
Multall06
Extrapolation
0.1
0
40
45
50
55
60
65
70
75
80
1 []
By looking at the Multall06 results in Figure 4-8 the outlet relative flow angle varies
considerably over the blade span. This is an effect of secondary flow vortices. The
amplitude of the variation though is relatively small, just a few degrees, excluding
the values near the hub. In the graph used for determining the blade inlet metal angle
the influence of 2 is very small especially in the region being considered. Along
with the fact that the precision in reading of a graph is not greater than a couple of
degrees, 2 was kept constant at 20 as an approximation. A vertical line is
superimposed in Figure 4-8 showing that this value is not that far off at 10, 50 and 90
percent of normalized radius.
36
0.7
0.6
0.5
0.4
0.3
Multall06
20 degrees
0.2
0.1
0
15
16
17
18
19
20
21
22
23
24
2 []
Figure 4-8 B6F - Approximation of relative flow angle at blade trailing edge
Considering the Beta2 results in Figure 4-7 and Figure 4-8 the flow angles have been
obtained just by reading of the x-axis at 0, 50 and 100 percent of normalized radius.
In the table below, the angles used as input to Prof. Mamaevs correlation from the
figures above and the resulting metal angles for B6F can be seen based on both Beta2
and Multall06 results.
1 [] Beta2
2 [] Beta2
1m [] Beta2
1 [] Multall06
2 [] Multall06
1m [] Multall06
Hub
45
22
51
47
20
56
B6F
Mid
53
20
59
56
20
62
Tip
76
21
77
61
20
66
Table 6 Input and output values associated with Prof. Mamaevs correlation
Figure 4-9 below shows a plot of the resulting incidence after optimizing accordingly
to the Multall06 results in Table 6.
37
0.7
0.6
0.5
0.4
0.3
0.2
Multall06
0.1
0
-40
-30
-20
-10
10
20
30
40
Incidence []
4.4.2
The sections at hub, mid and tip defining the blade should comply with the rule of
monotonically decreasing suction side curvature according to Siemens guidelines. In
the airfoil software CATO the influence of different geometrical parameters on
curvature can be evaluated. The curvature was first studied in the later parts of the
design process. The reason for this is that the basis for the whole design, the
reference stage complies with the constraint described above and the shape has not
been greatly modified, see Figure 4-10 below. This should imply that the design is
probably already reasonable in terms of curvature distribution. As this turned out to
be the case only very small adjustments were made in order to satisfy the criterion of
monotonically decreasing curvature on the suction side. Since the pressure side is
less sensitive to disturbances only the suction side has been adjusted.
The curvature has been plotted as a function of a normalized axial coordinate for
hub, mid and tip section belonging to the final design version, B6F. Here just
referred to as B6 since the curvature is the same both at full scale, B6F, and model
scale, B6M.
38
B6F
Reference blade
Figure 4-10 Comparison at mid section with B6F and the reference blade
Curvature B6 tip
5
Curvature
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-1
B6 tip SS
B6 tip PS
-2
-3
Norm. x
39
Curvature B6 mid
5
Curvature
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-1
B6 mid SS
B6 mid PS
-2
-3
Norm. x
Curvature B6 hub
5
Curvature
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
-1
B6 hub SS
B6 hub PS
-2
-3
Norm. x
40
4.4.3
The last step considering the airfoil was to study the velocity distribution at hub, mid
and tip section. Since the starting point of the design was a reference stage used in a
real gas turbine along with its flow conditions the velocity distribution should be
relatively similar. Below the isentropic Laval number as a function of a surface
coordinate can be seen for B6F and B6M at hub, mid and tip section for the blade.
is
1.1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.9
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
41
is
1.1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.9
is
1.1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
42
is
1.1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.9
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
From the figures above a small velocity spike can be seen just to the left of the
stagnation point. This is mainly a consequence of the transition between the circular
leading edge and the subsequent curvature of the suction side not being entirely
continuous. The reason for this is due to a shortcoming of the profile generator used
at time of the design of the reference stage. It has been kept to be as similar to the
43
reference stage as possible. The spike could possibly be removed in the new profile
software CATO.
The diffusion coefficient has been calculated from the velocity distributions above
for hub, mid and tip section belonging to B6F. Mamaevs criterion, D Dopt , is
fulfilled at every section, see Table 7 below:
D
0.157
0.159
0.152
Tip
Mid
Hub
Dopt
0.21
0.22
0.23
4.4.4
The specification of the hot geometry of B6, both in full and model scale are
presented here and can be seen summarized in Table 8. Especially noteworthy is the
aspect ratio and the number of blades of the proposed design which are lower than
typical gas turbine values. The reason for is, as mentioned earlier, a deliberate choice
because of possible manufacturing limitations. This low aspect ratio together with
the relative low number of blades is a direct consequence of the decision to extend
the axial chord.
Another important result is the relative trailing edge thickness, d2/b, which is much
smaller for 4b compared to B6. The difference originates from the fact that the
reference blade has a trailing edge slot, enabling cooling air to be ejected, which
limits the minimum trailing edge diameter. Since the airfoil profile of the reference is
directly scaled to B6 the relative trailing edge thickness will follow. This is a
deliberate decision and the relative trailing edge thickness of 4b is not characteristic
of a cooled gas turbine blade.
44
Blade parameter
rhub [mm]
rtip [mm]
bhub [mm]
bmid [mm]
btip [mm]
Bhub [mm]
Bmid [mm]
Btip [mm]
Blade heightentry [mm]
tmid [mm]
ARmid,axial [-]
ARmid,real [-]
(rtip/rhub)entry [-]
d2 [mm]
d2/bmid [%]
# of blades
(t/b)mid [-]
tip clearance [mm]
B6F
430.44
493.97
75.65
76.46
77.46
61.13
57.72
54.48
63.53
64.54
1.10
0.83
1.15
2.80
3.66
45
0.84
0.73
B6M
177.50
203.70
31.20
31.53
31.94
25.21
23.8
22.47
26.20
26.61
1.10
0.83
1.15
1.15
3.65
45
0.84
0.30
4b
177.50
204.80 (entry)
26.11
26.14
26.14
24.59
24.59
24.59
28.29
20.65
1.15
1.08
1.16
0.27
1.03
58
0.79
0.20
0.7
0.6
0.5
0.4
0.3
0.2
B6M
0.1
0
55
56
57
58
59
60
61
62
1m []
45
63
64
65
66
67
Figure 4-20 shows the blade inlet metal angle for B6M. The metal angle for B6 is
designed with regards to optimal incidence at full scale as described in section 4.4.1.
Since B6 is designed without any cooling flows the metal angle is almost linearly
distributed. The corner coordinates for B6M is shown in Figure 4-21.
A 3D view of the new turbine design, generated in ANSYS CFX, is shown in Figure
4-22.
46
5 Results
5.1 Boundary conditions
Since the total to static pressure ratio of the reference stage was to be achieved, the
total pressure was set at the inlet and the static pressure at the outlet. The goal was to
match the radial distribution of total pressure to measured data from the test turbine
as inlet condition. However, because of stability issues in CFX, with backflow at the
inlet, the total pressure was approximated as constant across the span. This was
applied as a boundary condition for all the programs so that the results could be
compared. The effects on the result caused by this approximation are believed to be
minimal.
The total inlet temperature was modeled as constant with radius for both model and
full scale. At model scale this is in good agreement with the actual profile provided
by the test turbine compressor however the set value of 345 K is slightly above the
target value in Table 1. This was not considered a problem since the test turbine
casing probably will be insulated to allow for a higher inlet temperature.
In Multall there is no option but to model the fluid as a perfect gas while CFX offers
the ability to have fluid properties which depends on pressure and temperature. Even
though it would have been desirable to define a real gas for the full and model scale
cases in CFX the fluid was modeled as a perfect gas due to time shortage. Experience
from other simulations at Siemens suggests that the effects of the perfect gas
approximation should be acceptable, especially at model scale where the test turbine
operates at very moderate temperatures. The heat capacity for each case was
provided by results from the through flow code (Beta2) and the rotational speed was
adjusted to give the same stage loading as the reference stage in Beta2. Steady state
and just a single passage were assumed in all of the calculations. The boundary
conditions used in both the 3D analysis and the through flow analysis for B6F/B6M
are summarized in Table 9.
Full scale
21.879
1540
10.229
2.139
6
9140
Model scale
2.167
345
1.013
2.139
6
10270
47
5.2.1
A comparison between B6F and the reference stage was done both in Beta2 and
Multall06. Because of confidential reasons, no values or figures are presented of this
study. In general the result was in good agreement between the turbines and the key
characteristics are maintained. Relatively large local deviations did occur though
because of the lack of cooling in B6F.
5.2.2
Figure 5-1 to Figure 5-4 show the radial distribution of the total pressure, static
pressure, total temperature, Mach number and tangential flow angle for the vane and
blade leading and trialing edge respectively as predicted by Beta2, Multall06 and
CFX. As can be seen the predictions differ between the codes. The general trend is
that CFX predicts lower static pressure and higher Mach number than Multall06 and
Beta2 except at the blade trailing edge where the opposite is true. Therefore the
predicted degree of reaction is lower in CFX compared to Beta2 and Multall06. Both
Multall06 and CFX are able to capture the effects of secondary flows to some degree
which are seen by strong radial variation in pressure and velocity at the blade trailing
edge. CFX predicts a smoother flow near the blade tip compared to Multall06,
explained by the more accurate tip modeling in CFX. Regarding the tangential flow
angle at the blade leading edge CFX and Multall06 predict very similar values which
confirm the decision to design the blade inlet metal angel according to Multall06
results. Overall the main predicted behavior is similar between the codes even
comparing Beta2 and CFX. However, there are differences especially regarding
Mach number and pressure. These deviations may be explained by the difference in
how the flow physics are captured, e.g. 2D vs. 3D, mesh density, etc., and, for
Multall06 and CFX, how the three-dimensional flow is averaged across the span. It is
also important to note that the positions of the planes in Multall06 and CFX from
which the data is extracted differs somewhat between the programs which also may
be a factor affecting the result.
48
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
B6F Multall06
B6F CFX Transition
B6F Beta2
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0.2
0.1
21.5
21.6
21.7
21.8
21.9
22
22.1
22.2
22.3
22.4
22.5
21.5
21.55
21.6
21.65
0.9
0.9
0.8
0.8
0.7
0.7
21.75
21.8
21.85
21.9
21.95
22
0.6
0.5
0.4
0.6
0.5
0.4
B6F Multall06
B6F CFX Transition
B6F Beta2
0.3
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0.2
0.1
0
0
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
0.08
1545
0.09
0.1
T [K]
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0
85
86
87
88
89
90
0.11
0.12
Ma [-]
21.7
P [bar]
P [bar]
91
92
93
94
95
0 []
49
0.13
0.14
0.15
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0
17
18
19
20
21
22
23
24
25
14
14.1
14.2
14.3
14.4
14.5
14.6
P* [bar]
14.9
15
15.1
15.2
15.3
15.4
15.5
0.9
0.9
0.8
0.8
0.7
0.7
14.8
P [bar]
0.6
0.5
0.4
0.3
0.6
0.5
0.4
B6F Multall06
B6F CFX Transition
B6F Beta2
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.2
0.1
0.1
1535
1536
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
0.5
0.55
0.6
0.65
T* [K]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0
10
11
12
13
14
15
0.7
0.75
Ma [-]
14.7
16
17
18
19
20
1 []
50
0.8
0.85
0.9
0.95
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
14.5
14.6
14.7
14.8
14.9
15
15.1
15.2
15.3
15.4
15.5
15.6
15.7
15.8
15.9
16
14
P* [bar]
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
1400
1420
1440
1460
1480
1500
1520
1540
0.05
0.1
0.15
T* [K]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0
40
50
60
70
0.2
0.25
Ma [-]
15
P [bar]
80
90
100
1 []
51
0.3
0.35
0.4
0.45
0.5
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
0
12
12.5
13
13.5
14
14.5
15
15.5
16
16.5
17
8.5
9.5
10
P* [bar]
11
11.5
12
12.5
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
0.2
B6F Multall06
B6F CFX Transition
B6F Beta2
0.1
B6F Multall06
B6F CFX Transition
B6F Beta2
0.2
0.1
1405
1410
1415
1420
1425
1430
1435
1440
1445
1450
0.6
0.65
0.7
T* [K]
0.75
0.8
0.85
0.9
0.95
Ma [-]
13
10.5
P [bar]
0.6
0.5
0.4
0.3
0.2
B6F Multall06
B6F CFX Transition
B6F Beta2
0.1
0
0
10
15
20
25
30
35
40
45
50
2 []
5.2.3
Figure 5-5 and Figure 5-6 show Mach contours as predicted by CFX at a radius of
25%, 50% and 75% of the span for the vane and blade respectively. Higher Mach
numbers are reached at the blade but the values are below 0.95 even at the blade
throat. The vane is designed for an inlet swirl angle which can be seen by the
location of the stagnation point in Figure 5-5. This is also the cause of the Mach
number peak near the leading edge in Figure 5-8. As can be seen from the contour
plots and from Figure 5-7 the flow is subsonic in the whole domain except at a
smaller region at the tip. This is to be expected since the pressure difference between
the pressure and suction side forces the flow across the tip, causing acceleration to
supersonic flow. The isentropic Mach number and blade loading for the vane and the
blade at 25%, 50% and 75% of the span is seen in Figure 5-8. The isentropic Mach
52
number is plotted against normalized surface position, with starting point at the
trailing edge and moving upstream the suction side with increasing s. The blade
loading is plotted against the axial fraction. A comparison with the isentropic Laval
number distributions shown in Figure 4-14 to Figure 4-19 indicates on the same trend
regarding the isentropic Mach number. The peak in Mach number close to the blade
stagnation point is explained, as mentioned in section 4.4.3, by the transition between
the cylindrical leading edge and the suction side curvature. A similar peak occurs for
the vane near the stagnation point, a consequence of the original 4b design. The 4b
vane was designed for 20 interstage swirl in contrast to 0 when used in a first stage
as in this case.
Figure 5-5 B6F Vane - Mach at 25%, 50% and 75% of span
53
Figure 5-6 B6F Blade - Mach at 25%, 50% and 75% of span
54
25% span
50% span
75% span
0.9
2000000
0.7
0.6
Mais [-]
25% span
50% span
75% span
2100000
0.8
0.5
0.4
0.3
1900000
1800000
1700000
1600000
1500000
0.2
1400000
0.1
0
1300000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
s/stot [-]
0.4
0.5
0.6
0.7
0.8
0.9
1600000
25% span
50% span
75% span
0.9
0.8
25% span
50% span
75% span
1500000
1400000
0.7
Mais [-]
0.6
0.5
0.4
0.3
1300000
1200000
1100000
1000000
0.2
900000
0.1
0
800000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
s/stot [-]
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
In Figure 5-9 to Figure 5-12 are radial distributions for B6M presented. The code
comparison regarding the radial distributions shows the same trends as in the full
scale case. Again, CFX predicts lower static pressures and higher Mach numbers
except at the blade trailing edge compared to Beta2 and Multall. The agreement
between Multall and CFX regarding the tangential flow angle at the blade leading
edge is conserved to the model scale. This indicates that the metal angle is
appropriately designed also at model scale.
55
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0.1
0
2.16
2.165
2.17
2.175
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0
2.14
2.18
2.145
2.15
P* [bar]
0.9
0.9
0.8
0.8
0.7
0.7
2.16
0.6
0.5
0.4
0.6
0.5
0.4
0.3
0.3
0.2
0.2
B6M Multall06
B6M CFX Transition
B6M Beta2
0.1
0
340
341
342
343
344
345
346
347
348
349
B6M Multall06
B6M CFX Transition
B6M Beta2
0.1
0
0.06
350
0.07
0.08
0.09
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0
85
86
87
88
89
90
0.1
0.11
Ma [-]
T* [K]
2.155
P [bar]
91
92
93
94
95
0 []
56
0.12
0.13
0.14
0.15
0.16
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
2.05
2.07
2.09
2.11
2.13
2.15
2.17
2.19
1.3
1.35
1.4
P* [bar]
1.5
1.55
1.6
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
340
341
342
343
344
345
346
347
348
349
350
0.6
0.65
0.7
T* [K]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0
10
11
12
13
14
15
0.75
Ma [-]
1.45
P [bar]
16
17
18
19
20
1 []
57
0.8
0.85
0.9
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0
1.4
1.45
1.5
1.55
1.6
1.3
1.35
1.4
1.45
P* [bar]
1.55
1.6
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M beta2
0.2
0.1
0.2
B6M Multall06
B6M CFX Transition
B6M Beta2
0.1
305
310
315
320
325
330
335
0.05
0.1
0.15
T* [K]
0.9
0.8
0.7
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0
0
10
20
30
40
0.2
0.25
Ma [-]
1.5
P [bar]
50
60
70
80
90
1 []
58
0.3
0.35
0.4
0.45
0.5
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
0.2
0.1
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
B6M Multall06
B6M CFX Transition
B6M Beta2
0.1
0
1.2
1.25
1.3
1.35
1.4
1.45
1.5
1.55
1.6
1.65
1.7
0.8
0.85
0.9
0.95
P* [bar]
1.05
1.1
1.15
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.5
0.4
0.3
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0.2
B6M Multall06
B6M CFX Transition
B6M Beta2
0.1
305
310
315
320
325
330
0.6
0.65
0.7
T* [K]
0.75
0.8
0.85
Ma [-]
1.2
P [bar]
0.6
0.5
0.4
0.3
B6M Multall06
B6M CFX Transition
B6M Beta2
0.2
0.1
0
0
10
15
20
25
30
35
40
2 []
5.3.2
Figure 5-13 and Figure 5-14 show Mach number contour plots for the vane and the
blade respectively. Compared to the full scale case in Figure 5-5 and Figure 5-6 the
Mach number is in general lower at model scale. As can been seen in Figure 5-15 the
tip leakage flow is supersonic to a lesser degree compared to the full scale case. This
could be explained by the lower Mach number in general but also by the fact that the
mesh was far more refined in the model scale cases (see section 3.4.2 and Appendix
A). The more refined mesh should allow for better representation of the actual flow
physics. Figure 5-16 shows the isentropic Mach number and blade loading compared
to the full scale case. There are very small deviations, indicating that the design
philosophy used at full scale is well translated to model scale.
59
0.9
Figure 5-13 B6M Vane - Mach at 25%, 50% and 75% of span
60
Figure 5-14 B6M Blade - Mach at 25%, 50% and 75% of span
61
0.9
220000
0.8
210000
25% span
50% span
75% span
200000
0.7
190000
Pressure [Pa]
Mais [-]
0.6
0.5
0.4
180000
170000
160000
0.3
150000
0.2
140000
25% span
50% span
75% span
0.1
130000
120000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
s/stot [-]
0.4
0.5
0.6
0.7
0.8
0.9
1
0.9
25% span
50% span
75% span
150000
0.8
140000
Pressure [Pa]
0.7
Mais [-]
0.6
0.5
0.4
130000
120000
110000
0.3
100000
25% span
50% span
75% span
0.2
0.1
90000
80000
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
s/stot [-]
5.3.3
B6M - Off-design
Design conditions are only achieved at a certain pressure ratio and rotational speed
but the test turbine will run far from design during experiments. It is therefore
interesting to explore how the turbine will behave at off design. The results from
sweeps done in Beta2 with pressure ratios and values of stage loading corresponding
to probable operating points can be seen below. Figure 5-17 to Figure 5-19 are
plotted against total to static pressure ratio for 80%, 100% and 120% N / T * ,
where 100% is given by the value at design load (=2.139, =1.37).
In Figure 5-17 the point of optimal efficiency moves to higher pressure ratios when
increasing the speed parameter. At some point the curves flatten and become less
sensitive, a consequence of turbine blades being relatively unaffected of arising
incidence losses. This is a typical behavior encountered when studying turbines
operating at off design, see Moustapha [2]. Figure 5-18 shows similar trends.
The flow capacity is not very susceptible to changes in the speed parameter, pressure
ratio being the dominant driving force as seen in Figure 5-19. When the curve starts
to become entirely horizontal the turbine has choked meaning the mass flow
becoming independent of pressure ratio as soon as sonic conditions are reached in the
blade throat.
Figure 5-20 and Figure 5-21 show approximately the same trends described above,
though Figure 5-20 clearly indicates that the maximum efficiency is at a lower load
than design load. This characteristic is deduced to the lack of cooling air which
moves the maximum efficiency to a lower loading. Optimally the stage should be
designed to have its maximum efficiency at the design load, this was however not
possible with the vane geometry fixed.
62
The trend of an increase in degree of reaction with pressure ratio for a fix value of
stage loading is shown in Figure 5-22. As the back pressure is fixed, increasing the
pressure ratio will lead to a greater static pressure drop across the rotor compared to
the stator and the degree of reaction will rise. It is seen that for low pressure ratios
and high stage loading there is a sudden increase in reaction. For these cases the
positive incidence is very large (>35) and it seems that the Beta2 code has problem
to achieve physical results. Further studies in Beta2 in this region indicate numerical
instability and the reliability cannot be verified.
Torque is increasing considerably with pressure ratio, stage loading having a
significant, but still smaller, impact as seen in Figure 5-23. The pressure ratio
dependency could be explained by the increase in mass flow together with the
increased isentropic enthalpy drop over the stage. This is just a simplified way of
explaining the trend in the figure to some extent.
B6M - Off design efficiency characteristics
0.95
0.9
* [-]
0.85
0.8
100% n/T*
80% n/T*
120% n/T*
0.75
0.7
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.1
[-]
Figure 5-17 B6M - Total to total efficiency against pressure ratio (Beta2)
63
2.2
0.85
[-]
0.8
0.75
0.7
100% n/T*
80% n/T*
120% n/T*
0.65
0.6
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
2.1
2.2
[-]
Figure 5-18 B6M - Total to static efficiency against pressure ratio (Beta2)
25
20
100% n/T*
80% n/T*
120% n/T*
15
10
1
1.1
1.2
1.3
1.4
1.5
1.6
1.7
1.8
1.9
[-]
64
2.1
2.2
* [-]
0.9
0.85
Pi 1.15
Pi 1.53
Pi 2.1
0.8
0
0.5
1.5
2.5
3.5
Figure 5-20 B6M - Total to total efficiency against stage loading (Beta2)
3.5
2.5
1.5
Pi 1.15
Pi 1.34
Pi 1.53
Pi 1.72
Pi 1.91
Pi 2.1
0.5
0
0.5
1.5
2.5
65
3.5
p [-]
0.3
0.25
0.2
0.15
Pi 1.15
Pi 1.34
0.1
Pi 1.53
0.05
Pi 1.72
Pi 1.91
Pi 2.1
0
0.5
1.5
2.5
3.5
250
Torque [Nm]
200
150
100
Pi 1.15
Pi 1.34
Pi 1.53
Pi 1.72
50
Pi 1.91
Pi 2.1
0
0.5
1.5
2.5
66
3.5
Cf
w
2
0.5 C
w =
u
y
The freestream values for the density and velocity was extracted at a distance of 0.75
- 1.5 mm from the wall surface outside the boundary layer and the wall shear stress is
directly given by CFX at the surface. The result from the CFX cases with and
without the transition model can be seen in Figure 5-24. In the full scale case
transition occurs on the suction side at a distance about 45% from the leading edge
for the vane whilst transition occurs almost immediately at blade leading edge
according to the transition model. At model scale the transition model predicts
completely laminar boundary layer on the suction side of the vane and transition at a
distance about 55% from the leading edge at the suction side of the blade. The big
difference between full and model scale is to be expected and is explained by the
greater Reynolds number in the full scale case, which is about an order of magnitude
greater, see Table 12. Table 10 shows the computed efficiencies for each case. In
accordance with the skin friction coefficient calculation the predicted gain is greater
at model scale.
A comment about the increase in skin friction for the model scale case close to the
blade trailing edge has to be made. This is a local effect caused by a secondary flow
vortex which interrupts the flow and increases the wall shear strain at the line from
which the data is extracted. The vortex is still present in the Low-Re case but passes
the blade at a greater radius, avoiding the mid span line.
67
0.01
0.01
0.009
Transition model
Low-Re
0.008
0.007
0.007
0.006
0.006
Cf [-]
Cf [-]
0.009
Transition model
Low-Re
0.008
0.005
0.005
0.004
0.004
0.003
0.003
0.002
0.002
0.001
0.001
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
0.2
0.3
0.4
s/stot [-]
0.6
0.7
0.8
0.9
0.01
0.01
0.009
0.009
0.008
0.008
0.007
0.007
0.006
0.006
Cf [-]
Cf [-]
0.5
s/stot [-]
0.005
0.005
0.004
0.004
0.003
0.003
0.002
0.002
Transition model
Low-Re
0.001
Transition model
Low-Re
0.001
0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
0.1
s/stot [-]
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
s/stot [-]
Low-Re
0.8742
0.9213
B6F
Transition
0.8754
0.9232
Low-Re
0.8675
0.9073
B6M
Transition
0.8722
0.9190
68
Beta2 1d
0.906
0.904
B6M [-]
B6F * [-]
Multall
0.9141
0.9149
CFX Low-Re
0.9073
0.921349
CFX Tran
0.9190
0.923193
[-]
[-]
[-]
p,mid [-]
1 []
1 []
2 []
2 []
Ma1abs [-]
Ma1rel [-]
Ma2rel [-]
Re1abs [-]
Re2rel [-]
Ref. stage
2.139
1.365
0.407
0.405
14.45
63.18
73.10
19.91
0.794
0.236
0.812
30.3e5
16.0e5
B6F
2.138
1.364
0.428
0.398
14.66
54.79
71.02
20.53
0.797
0.251
0.811
29.3e5
20.7e5
B6M
2.138
1.369
0.425
0.386
14.69
52.91
72.75
20.63
0.781
0.252
0.78
6.9e5
4.9e5
4b
1.23
2.035
0.396
0.089
14.73
26.37
73.84
19.50
0.532
0.280
0.323
3.1e5
2.1e5
Parameter:
r [mm]
d1 [mm]
d2 [mm]
1m []
2m []
1 []
2 []
[]
t [mm]
b [mm]
B [mm]
t/b [-]
ARreal [-]
ARaxial [-]
d2/b [%]
# of blades
tip clearance [mm]
Hub
430.44
7.50
2.80
56.18
18.96
65.93
9.75
22.27
60.10
75.65
61.13
0.79
0.84
1.04
3.70
B6F
Mid
462.21
7.44
2.80
61.83
18.93
59.23
9.66
22.53
64.54
76.46
57.72
0.84
0.83
1.10
3.66
Tip
493.97
7.46
2.80
66.36
18.88
54.76
9.60
23.20
68.97
77.46
54.48
0.89
0.82
1.17
3.61
Hub
177.50
3.09
1.15
56.19
18.95
65.93
9.73
22.28
27.78
31.20
25.21
0.79
0.84
1.04
3.69
45
0.73
B6M
Mid
190.60
3.07
1.15
61.83
18.93
59.23
9.66
22.53
26.61
31.53
23.80
0.84
0.83
1.10
3.65
0.30
69
Tip
203.70
3.08
1.15
66.36
18.88
54.75
9.61
23.20
28.44
31.94
22.42
0.89
0.82
1.17
3.60
6 Conclusions
A new blade design for the KTH test turbine has been presented in this thesis. The
aerodynamic design of the blade is based on an industrial gas turbine, which acted as
a reference. The design was performed at the reference scale and with similar gas
conditions. A summary of the main conclusions:
Full geometrical similarity between the reference blade and the B6M design
was not possible because of manufacturing limitations and the chord had to
be expanded. The aspect ratio is 0.83 for B6M which is considerable lower
than for the reference. As a consequence the number of blades is much fewer
for B6 compared to the reference.
The metal angle was redesigned at full scale according to optimal incidence
by Mamaev which resulted in a different blade twist compared to the
reference blade, even though the meanline analysis indicated similar vane
outlet flow angle. However, Beta2 and Multall06 predicted a great difference
in blade inlet flow angel at relatively large regions near the hub and the tip
due to lack of cooling flows. This resulted in a more uniform and almost
linear twist compared to the reference blade. CFX predicted a similar result as
Multall06 regarding blade flow angles and no redesign was considered due to
the updated results.
The final design at full scale B6F accomplishes the set target to mimic a
reference stage although the test turbine vane geometry was maintained and
no cooling was used. Both through flow and 3D calculations support this
statement.
At model scale the predicted result was very much similar and the general
characteristics at full scale were well kept. However, since parameters as the
Reynolds numbers and Mach numbers are smaller for a given load at model
scale the predicted performance will differ in absolute values.
According to the transition study in CFX both the vane and the blade at
model scale experience laminar boundary layers to a much greater degree
than at full scale. This effect must be considered when evaluating results from
the test turbine.
70
7 Future work
The constraint of matching the axial chord at hub led to the reference blade profiles
had to be enlarged in order to maintain the shape resulting in B6M. This design is the
final result of this report complying with the requirement. During the work resulting
in the final design no consideration has been taken to structural limitations.
Regardless of design structural analysis must be performed considering the whole
blisk (blade row and disc machined from one solid piece).
The scaling is based on the 4b geometry at cold conditions but was assumed to be at
hot conditions, since the actual hot geometry of the blisk is not known. In reality
centrifugal and thermal load will expand the dimensions somewhat. Even though
conditions in the test turbine are far removed from what is normally expected in a gas
turbine, some elongation will still occur. Of course this has to be investigated further
before manufacturing can commence.
71
8 Appendix A
The mesh and y+ statistics for B6F and B6M are shown in Table 14. In the full scale
case the mesh had to be coarsened at the tip clearance region because of problem
with numerical stability, resulting in larger y+ values than desired. The effect of this
is clearly seen in Table 14 where the maximum and average y+ is considerable larger
at full scale than small scale for the blade domain. According to a study regarding the
effects of too large y+ presented in the CFX Solver Guide [6], y+ should be kept
between 0.001 and 8 to effectively predict the transition point with the Gamma Theta
transition model. This criterion could not be fulfilled at the tip in the full scale case
but for the rest of the domain (values in parenthesis) the criterion is met. Since the
flow at the tip clearance probably is fully turbulent anyway this compromise is not
believed to have affected the transition model study in any major way. Figure 8-1
shows two computational domains in CFX for illustrative purposes. Only one
passage is actually modeled and the result is mirrored to the other passages. The vane
and blade mesh in the B6M case is shown in Figure 8-2 and Figure 8-3.
# of
elements
y+ max
y+ min
y+ area
averaged
Vane
1943250
2.40
7.28e-3
0.98
Full scale
Blade
Outlet
1551068
121720
47.10
(7.60)
8.69e-3
3.66
(1.60)
Vane
1115400
Model scale
Blade
Outlet
2200581
153000
2.86
4.12
1.38e-2
0.84
1.42e-2
0.54
Figure 8-4 to Figure 8-7 shows contour plots of the total pressure at the vane and
blade trailing edge for full and model scale respectively. Both the full and model
scale result indicates that the vane is exposed to relatively small secondary flows
while the main flow at the blade trailing edge is highly affected by secondary flows.
In Figure 8-8 and Figure 8-9 is velocity vectors close to the surface and streamlines
at the hub and tip shown. The figures clearly show the extent of the secondary flows
in a low aspect ratio blade.
72
73
74
75
76
Bibliography
1) Saravanamuttoo, H.I.H., Rogers, G.F.C., Cohen, H., Straznicky, P.V Gas turbine
theory. 6th Edition. London: Pearson Education Limited, 2009.
2) Moustapha, H., Zelesky, M.F., Baines, N.C Japikse, D. Axial and radial turbines.
Vermont: Edwards Brothers Incorporated, 2003.
3) Dixon, S.L., Hall, C.A. Fluid Mechanics and Thermodynamics of
Turbomachinery. 6th Edition. Burlington: Elsevier Inc., 2010.
4) Mamaev, B.I., Klebanov, A.G., Loss model for vane/blade rows. Moscow:
Siemens Industrial Turbomachinery AB, 2007.
5) CFD-Online, 2012. Baldwin-Lomax model
http://www.cfd-online.com/Wiki/Baldwin-Lomax_model
(used 2013-03-03).
6) CFX-Solver Modeling Guide. 4.1.10. CFX Transition Model. ANSYS, Inc, 2011.
7) Hedlund, L. TurboAero Preliminary plan for Johan Dahlqvists visit to Finspong
in January 2013. Finspong: Siemens Industrial Turbomachinery, 2013
8) Cengel, Y.A., Boles, M.A., Thermodynamics: An Engineering Approach Sixth
Edition (SI Units). McGraw-Hill, 2007.
9) Genrup, M., Lecture notes from the courses MVKF05/MVK051 Lund: LTH
Department of Energy Sciences 2012.
10) Hedlund, L., Li, Y.S., Mamaev, B.I., On the aerodynamic design of turbines
Notes from meetings with Boris Mamaev. Finspong: Siemens Industrial
Turbomachinery, 2012.
11) Langtry, R.B., Menter, F.R., Transition Modeling for General CFD Applications
in Aeronautics. Otterfing: ANSYS CFX, 2005.
12) Anderson, J.D., Fundamentals of Aerodynamics. 5th Edition. New York:
McGraw-Hill., 2011.
77