a lumped thermodynamic model of gas turbine blade cooling ... · turbine cycle performance induced...
TRANSCRIPT
PROCEEDINGS OF ECOS 2016 - THE 29TH INTERNATIONAL CONFERENCE ON
EFFICIENCY, COST, OPTIMIZATION, SIMULATION AND ENVIRONMENTAL IMPACT OF ENERGY SYSTEMS
JUNE 19-23, 2016, PORTOROŽ, SLOVENIA
A lumped thermodynamic model of gas turbine blade cooling: prediction of first-stage blades temperature
and cooling flow rates
Roberta Mascia, Enrico Sciubbab
a Dept. of Mechanical and Aerospace Engineering, University of Rome “Sapienza”, Rome, Italy,
[email protected] (CA) b Dept. of Mechanical and Aerospace Engineering, University of Rome “Sapienza”, Rome, Italy,
Abstract
Turbine Inlet Temperatures of 1500-2000K have become a sort of standard for most modern advanced
applications. First-stage blades are obviously the most exposed components to such hot gases, and
thus they need proper cooling. In the preliminary design of the blades and their cooling system,
designers must rely on simple models that can be further refined at a later stage, in order to have an
approximate but valuable set of guidelines and to reach a feasible first-order configuration. In this
paper, a simple lumped thermodynamic model of blade cooling is proposed. It is based on
mass/energy balances and heat transfer correlations and it predicts a one-dimensional temperature
profile on the blade external surface along the chord for a given gas temperature profile, as well as
the required cooling air flow rates to prevent blade material from creep. The greatest advantage of
the model is that it can be easily adapted to any operating condition, process parameter and blade
geometry, which makes it well suited to the last technological trends, namely the investigation of new
cooling methods and alternative coolants instead of air. Therefore, the proposed model is expected to
be a useful tool in the field of innovative gas turbine cycle analysis, replacing more computationally
intensive and very time-consuming models.
Keywords
Gas Turbine, Blade Cooling, Heat Transfer, Thermodynamics.
1 Introduction
Both thermal efficiency and power output of a gas turbine are known to increase with Turbine Inlet
Temperature (TIT). Therefore, over the past decades, aircraft and power generation gas turbine
designers have focused on increasing TIT until reaching the current temperature levels of 1500-
2000K in most advanced applications. Such high operating temperatures by far exceed the maximum
allowable blade material temperatures, so that an efficient cooling system is essential in order to
maintain the material temperatures under a defined threshold value. Turbine cooling is obtained by
the extraction of a small portion of air from the high-pressure compressor stages, which bypasses the
combustor and is reintroduced directly in the turbine to cool its most sensitive components. While on
the one hand such relatively cold air used as cooling medium induces a great benefit to the
components operational life, on the other hand, its extraction represents a direct loss of engine thermal
efficiency and power output. Therefore, maximum cooling with minimum cooling air flow rate is the
design goal. The best cooling is one that provides an as uniform as possible material temperature in
the blade (so to reduce thermal stresses) and with a peak value lower than the maximum allowable
for the blade material to avoid creep. Many improvements in blade cooling system design have been
made thanks to the industrial relevance of the problem and to the constant development of
Computational Fluid Dynamics (CFD) and of metallurgical methods over the last years.
2
In the preliminary design of gas turbine first-stage blading (both statoric and rotoric) and of their
cooling system, sophisticated or computationally intensive models are not required; they would result
exceedingly resource-consuming, due to the excessive computational burden required to attain a
satisfactory accuracy on a very complex geometry and the need for a less computationally demanding
solution is felt at this project stage. It is hence important for designers to rely on simpler models that
can be further refined later in the engineering process, mostly based on bulk quantities and one-
dimensional simplified equations, in order to arrive at approximate yet meaningful estimates of
temperatures and flow requirements.
El-Masri proposed a now classical, simple lumped model for expansion in cooled turbines in 1986
[5]. His aim was to quantify the turbine-cooling losses for different types of cooling methods, as a
detailed prediction of them was feasible for specific designs and operating conditions only. Many
models of blade cooling were published since [1, 4, 11, 12, 13]. Some of them arose as natural
developments of the El-Masri model, like those proposed by De Paepe [4] and by Bolland and Stadaas
[1], others as different approaches to the same problem, but with the same goals. In particular,
important contributions were made by Jordal [13] and Horlock [11] in their studies on effects on gas
turbine cycle performance induced by turbine cooling.
In this paper, a simple lumped thermodynamic model is adopted for the cooling of the first stage
(stator and rotor) blading. The aim is to predict the one-dimensional surface temperature along the
chord (streamwise) of a gas turbine first stage vane and blade, for a given gas temperature profile
along the stator and rotor channel mid-streamline. Then, once the blade metal temperatures are
known, a proper Ni-based superalloy is chosen in order to prevent blade material from creep at that
stress level. Finally, the required cooling air flow rates needed to guarantee such material
temperatures are estimated both for the vane and for the blade.
2 Turbine blade cooling: state-of-the-art The most common blade cooling methods currently implemented in the turbine industry belong to
two main categories: internal and external cooling.
Internal cooling is realized by making the coolant flow through properly shaped channels in the
interior of the blade, removing by internal convection a portion of the heat transferred to the blade by
the hot external mainstream, thus reducing the blade metal temperature. To make cooling systems
more efficient and reduce compressed air extraction, different internal cooling techniques have been
proposed [6, 7, 8, 9, 10]. Most popular are: jet impingement, rib turbulated cooling and pin-fin
cooling. Their names originate from the different features, commonly known as turbulence promoters,
adopted to enhance the heat transfer coefficient by augmentation of the flow turbulence and/or surface
area.
External cooling consists in ejecting the coolant that passes in the blade internal cooling channels
through small holes drilled into the airfoil external shell. The resulting ejected air jets are promptly
squeezed towards the wall by the main flow, and create a thin, cooler, and insulating film along the
external surface of the turbine blade, whence the name “film cooling”. This film provides blade
protection from the hot combustion gases, thus keeping the blade temperature lower than that of an
internally cooled blade, which helps increasing blade operational life. On the other hand, the non-
neglegible interference between the injected coolant and the main flow makes the positive effect of
the film cooling technique global rather than local and thus more complicated to predict.
The latest major improvement to turbine blade material technology was the development of Thermal
Barrier Coatings (TBC), which increased turbine blade temperature capability thanks to their very
low thermal conductivity λ [0.5-2 W/(mK)], improving blade lifetime as a consequence, and they
improved corrosion and oxidation resistance as well, both of great concern as temperatures increases.
A detail representation of how the different cooling methods are currently combined in stator vanes
and rotor blades, to adequately cool them, is given in Fig. 1.
3
3 The proposed lumped thermodynamic model The proposed model is totally based on prime principles
(mass and energy balances) and heat transfer correlations.
The main difference with respect to the other models that can
be found in the literature (see section 1.2) is in the treatment
of the blade energy balance. In particular, the heat transfer
process between the blade and the two flows (hot gas and
coolant) is performed by a local thermal balance on an
infinitesimal control volume, shown in Fig 2. This
elementary volume embeds the blade, spanning its entire
height l and extending on both the pressure and suction sides
for a length t/2 equal to half of the blade spacing. Its
meridional (streamwise) length is equal to dx, where x is a
curvilinear adimensional coordinate measured along the
blade chord starting from the leading edge (x=0) and ending
in the trailing edge (x=1).
A distinction is made between the use of internal convection cooling alone and combined with film
cooling, so that two different procedures are applied to both the stationary vane and the rotating blade.
The aim is to estimate the blade surface temperature along the chord and the coolant mass flow rates.
3.1 Internal Convection Cooling
The first goal of the model is to calculate the metal temperature profile along the chord on the blade
external surface Tbe(x), which are the maximum temperatures experienced by the material. A local
energy balance is applied to the elemental control volume by equalling the infinitesimal heat dqgas(x)
lost by the hot gas during its expansion along dx to the heat transferred to the blade by external
convection dqconv(x):
𝑑𝑞𝑔𝑎𝑠(𝑥) = 𝑐𝑝 [𝑇𝑔(𝑥) − (𝑇𝑔(𝑥) +𝑑𝑇𝑔(𝑥)
𝑑𝑥𝑑𝑥)] = −𝑐𝑝
𝑑𝑇𝑔(𝑥)
𝑑𝑥𝑑𝑥
𝑑𝑞𝑐𝑜𝑛𝑣(𝑥) = ℎ𝑒𝑥𝑡(𝑥) (𝑇𝑔(𝑥) − 𝑇𝑏𝑒(𝑥)) 𝑑𝐴𝑏 = ℎ𝑒𝑥𝑡 (𝑇𝑔(𝑥) − 𝑇𝑏𝑒(𝑥)) 2𝑙𝑑𝑥
𝑑𝑞𝑔𝑎𝑠(𝑥) = 𝑑𝑞𝑐𝑜𝑛𝑣(𝑥)
Combining the above expressions with equations 3.1 and 3.2:
−𝑐𝑝
𝑑𝑇𝑔(𝑥)
𝑑𝑥𝑑𝑥 = ℎ𝑒𝑥𝑡(𝑥) (𝑇𝑔(𝑥) − 𝑇𝑏𝑒(𝑥)) 2𝑙𝑑𝑥
(3.1)
(3.2)
(3.3)
Figure 2. The elemental control
volume [14]
Figure 1. Schematic view of a modern gas turbine vane (a) and
blade (b) with common cooling techniques [10]
(a) (b)
4
From eq. 3.3 we obtain:
𝑇𝑏𝑒(𝑥) = 𝑇𝑔(𝑥) +𝑐𝑝
2 ∗ 𝑙 ∗ ℎ𝑒𝑥𝑡(𝑥)∗
𝑑𝑇𝑔(𝑥)
𝑑𝑥
In the assumption of knowing the gas temperature profile along the channel Tg(x) and the values of
the external heat transfer coefficient hext(x), and assuming a constant cp, Tbe(x) can be computed from
eq. 3.4. First-stage turbine rotating blades are the limiting components of the gas turbine, since they
are the most thermally and mechanically loaded airfoils. In particular, for such airfoils long-term
creep resistance is the most relevant life-limiting factor. Therefore, a creep test on the rotor blade
must be performed in order to choose the proper superalloy able to withstand such hot gas
temperatures, without failure risks, for a prescribed number of operating hours. The Larson-Miller
Parameter (LMP) relation is used for this purpose:
LMP = T (C + log t)
where C is a material specific constant often approximated as 20, t is the stress-rupture time in hours
and T is the temperature in Kelvin.
The heat transferred to the blade surface is computed from eq. 3.2. The same heat is transferred within
the blade by thermal conduction, and Tbi(x) can be computed from eq. 3.6:
𝑑𝑞𝑏𝑙𝑎𝑑𝑒(𝑥) =𝜆
𝑠(𝑥)(𝑇𝑏𝑒(𝑥) − 𝑇𝑏𝑖(𝑥))2𝑙𝑑𝑥 = 𝑑𝑞𝑐𝑜𝑛𝑣(𝑥)
In the assumption that the heat flux dqblade(x) is totally absorbed by the coolant (no heat transfer to
the disk and to the casing), the following relation holds:
𝑑𝑞𝑖𝑛𝑡(𝑥) = ℎ𝑖𝑛𝑡(𝑥) 𝑟 (𝑇𝑏𝑖(𝑥) − 𝑇𝑐(𝑥))𝑑𝐴𝑐 = 𝑑𝑞𝑏𝑙𝑎𝑑𝑒(𝑥) (3.7)
Where r is a geometry factor expressly introduced to represent
the different heat transfer enhancements that are currently
adopted. In particular, according to their effectiveness, whose
schematic representation is given Fig. 3, they can be expressed
by means of factor r as:
- r=1 Smooth surface
- r=1.2 Jet impingement
- r=1.4 Turbulence promoters
- r=1.6 Pin fins
In the calculations presented here below, only the smooth-
channel configuration (r=1) will be analyzed, so that the worst
heat transfer conditions are tested. Finally, the heat transferred per unit of length q(x) can be analytically computed, since it turns out to
be equal to any of these different expressions:
𝑞(𝑥) = −𝑐𝑝
𝑑𝑇𝑔(𝑥)
𝑑𝑥= ℎ𝑒𝑥𝑡(𝑥) (𝑇𝑔(𝑥) − 𝑇𝑏𝑒(𝑥)) 2𝑙 =
𝜆
𝑠(𝑥)(𝑇𝑏𝑒(𝑥) − 𝑇𝑏𝑖(𝑥))2𝑙
= ℎ𝑖𝑛𝑡(𝑥) 𝑟 (𝑇𝑏𝑖(𝑥) − 𝑇𝑐(𝑥))2𝑙 [𝑊
𝑚]
Approximating the cooling channels to cylinders, the heat q(x) is then integrated for each portion of
the blade chord related to a certain cooling cylinder. Thus, for a number j of cylinders a correspondent
number of Δqj is obtained, whose sum is the whole heat transferred to the blade from the hot gases.
∆𝑞𝑗 = ∫ 𝑞(𝑥)𝑑𝑥𝑥𝑗
0
In the assumption that each of these Δqj is absorbed only by the coolant that flows into the
correspondent cylinder, the following relation can be written for the coolant in each cylinder:
(3.4)
(3.6)
(3.5)
(3.9)
(3.8)
Figure 3. Schematic view of the
enhance systems effectiveness [4]
5
𝛥𝑞𝑗 = 𝑚𝑐𝑐𝑝𝑐𝛥𝑇𝑐𝑗= 𝑚𝑐𝑐𝑝𝑐 (𝑇𝑐𝑜𝑗
− 𝑇𝑐𝑖𝑗)
Expressing the coolant mass flow rate as a fraction i of the gas mass flow rate:
𝑖 =𝑚𝑐
𝑚𝑔
→ 𝑚𝑐 = 𝑖 𝑚𝑔
Equation 3.10 can be rewritten as:
𝛥𝑞𝑗 = 𝑚𝑐𝑐𝑝𝑐𝛥𝑇𝑐𝑗= 𝑖 𝑚𝑔𝑐𝑝𝑐𝛥𝑇𝑐𝑗
The coolant temperature Tcj progressively increases along the blade height, from Tcij to Tcoj, as the
coolant heats up while flowing through the channels. A growing trend with blade height should hence
be given to Tcj to simulate such phenomenon. However, since in this work a two-dimensional analysis
of an airfoil section is performed, no 3-D radial phenomena can be included. Therefore, the Logarithm
Mean Temperature Difference (LMTDj) that drives the heat transfer between the blade and the coolant
in each cylinder is defined as:
𝐿𝑀𝑇𝐷𝑗 =∆𝑇𝑐𝑜𝑗
− ∆𝑇𝑐𝑖𝑗
𝑙𝑛 (∆𝑇𝑐𝑜𝑗
∆𝑇𝑐𝑖𝑗
)
=(𝑇𝑏𝑖𝑗
− 𝑇𝑐𝑜𝑗) − (𝑇𝑏𝑖𝑗
− 𝑇𝑐𝑖𝑗)
𝑙𝑛 ((𝑇𝑏𝑖𝑗
− 𝑇𝑐𝑜𝑗)
(𝑇𝑏𝑖𝑗− 𝑇𝑐𝑖𝑗
))
Knowing the internal heat transfer coefficient hint and the finite heat transfer area Ac of each cylinder,
relation 3.7 can be rewritten for the j-th cylinder as:
𝛥𝑞𝑗 = ℎ𝑖𝑛𝑡 𝑔 𝐿𝑀𝑇𝐷𝑗 𝐴𝑐
From which the LMTDj can be computed:
𝐿𝑀𝑇𝐷𝑗 =∆𝑞𝑗
ℎ𝑖𝑛𝑡 𝑔 𝐴𝑐
For a given coolant inlet temperature in the cooling system, the coolant temperature difference
between inlet and outlet in each cylinder ΔTcj, as well as Tcij and Tcoj, for different values of i, can be
estimated from the following relation, derived from eq. 3.12:
𝛥𝑇𝑐𝑗=
𝛥𝑞𝑗
𝑚𝑐 𝑐𝑝𝑐
=𝛥𝑞𝑗
𝑖 𝑚𝑔 𝑐𝑝𝑐
The internal blade temperatures Tbij on the cylinders surfaces can now be computed for each cooling
ratio. The following formula derived from the LMTD definition (3.13) is used:
𝑇𝑏𝑖𝑗=
𝑇𝑐𝑖𝑗− ( 𝑇𝑐𝑜𝑗
𝑒
(𝑇𝑐𝑜𝑗−𝑇𝑐𝑖𝑗
)
𝐿𝑀𝑇𝐷𝑗 )
1 − 𝑒
(𝑇𝑐𝑜𝑗−𝑇𝑐𝑖𝑗
)
𝐿𝑀𝑇𝐷𝑗
By joining all the Tbij temperatures, different profiles for the blade internal surface temperature Tbint(x)
can be obtained for each cooling mass flow rate. However, the more realistic temperature profile for
Tbi(x) is that derived from eq. 3.6, since it is related to the blade external surface temperature Tbe(x)
and to the thermal conductivity λ of the blade material. Therefore, in order to assess the coolant mass
flow rate required to guarantee such blade internal temperature Tbi(x), a comparison between it and
the various temperature profiles Tbint(x) is made. To this extent, one must consider that the selected
coolant mass flow rate will be the one for which the two profiles most closely match. As criterion of
choice the standard deviation is used, calculated as follows for each value of i:
𝜎𝑖 = √∫ |(𝑇𝑏𝑖𝑛(𝑥) − 𝑇𝑏𝑖(𝑥))2| 𝑑𝑥1
0
(3.14)
(3.15)
(3.16)
(3.17)
(3.18)
(3.11)
(3.12)
(3.13)
(3.10)
6
The value of i for which the standard deviation is lowest will be the needed cooling mass ratio, from
which then the absolute value of the required coolant mass flow rate can be assessed.
3.2 Film cooling
A different method is adopted to assess the coolant mass flow rates when film cooling is also applied.
The internal cooling channels are always approximated by cylinders, with the addition of the cooling
holes. Firstly, q(x) is computed from eq. 3.8, then, applying the same procedure as internal cooling,
the Δqj are computed from eq. 3.9, while the ΔTcj, Tcij and Tcoj are calculated from eq. 3.16 for each i-
th cooling mass ratio. A film cooling efficiency is introduced to compute the film temperature Tf(x),
defined as:
𝜀𝑓 =(𝑇𝑔(𝑥𝑓) − 𝑇𝑓)
(𝑇𝑔(𝑥𝑓) − 𝑇𝑐𝑜𝑗)
Once a value for the efficiency is assigned, such formula is applied to each cooling hole, in the
assumption that the film temperature remains constant between one hole and the following one.
Although this is certainly a strong assumption, it can be considered acceptable within the present
context. Indeed, in a physical sense this is the only reasonable assumption. The temperature Tgas is
considered as the gas temperature at each hole position xf along the chord. The value for Tcoj, which
is the temperature at which cooling air is discharged into the main flow, is assigned according to the
cylinder to which the hole refers.
𝑇𝑓(𝑥𝑓) = 𝑇𝑔(𝑥𝑓) − 𝜀𝑓 (𝑇𝑔(𝑥𝑓) − 𝑇𝑐𝑜𝑗)
A film temperature profile Tf(x) for each value of i is then obtained by fitting such Tf values. The heat
transfer is now between the cooling layer and the blade surface, so a new heat transfer equation can
be written:
𝑞(𝑥) = ℎ𝑓(𝑥) (𝑇𝑓(𝑥) − 𝑇𝑏𝑒(𝑥)) 2𝑙
Where hf is the heat transfer coefficient under film cooling conditions, which is known to be higher
than hext. Such increase in the heat transfer coefficient is accounted for by introducing an enhancement
factor u, taken equal to 1.5, so that:
ℎ𝑓(𝑥) = 𝑢 ℎ𝑒𝑥𝑡(𝑥) = 1.5 ℎ𝑒𝑥𝑡(𝑥)
Eq. 3.21 is then used to calculate the blade external surface temperature Tbe(x) for each cooling
percentage i, with the resulting following equation:
𝑇𝑏𝑒(𝑥) = 𝑇𝑓(𝑥) −𝑞(𝑥)
1.5 ℎ𝑒𝑥𝑡(𝑥)2𝑙
The first value of i (coolant mass flow rate) that provides a temperature profile Tbe(x) that has a
maximum temperature below the maximum allowable by the material is hence chosen.
4 Cooling analysis of a representative gas turbine first stage The proposed model was implemented in MATHEMATICA© in order to carry out the calculations.
Two separate codes were written to account for the different set of equations needed by internal
convection and film cooling analysis. In particular, for each cooling method, two additional sub-codes
were written to apply the model to both the turbine first-stage stationary vanes and rotating blades.
The first input for the MATHEMATICA© code is the hot gas temperature profile along the mid-
streamline of both stator and rotor channels. Since neither reliable nor realistic gas temperature
profiles have been found in the literature, and industrial data are unavailable as well, it was decided
to obtain the needed profiles from CFD simulations. Therefore, numerical 2-D simulations of the flow
on the midspan section of a gas turbine first stage vane (internally cooled by convection) and of a
(3.23)
(3.19)
(3.20)
(3.21)
(3.22)
7
rotor blade were performed. The midspan section was chosen for the sake of generality, so that the
calculations could then be equally adapted to any airfoil section (e.g. hub and tip).
The simulations were carried out with the commercial solver ANSYS Fluent 15.0. It is important to
remark in advance that all the successive results strongly depend on the particular choice of the gas
turbine, whose first-stage configuration parameters are listed in Table 1, while the geometry
parameters for the NGV and rotor blade are listed in Table 2 and their midspan sections are shown in
Fig.4. The assigned inlet boundary conditions values are all listed in Table 3. The static temperature
maps in the stator and rotor channel obtained from the simulations are plotted in Fig. 5.
Configuration parameters
Power output, P [MW] 4
Diameter, D [m] 0.5
Rotational speed, ω [s-1] 1200
Velocity, U [m/s] 300
Blades, z 44
Gas mass flow rate, mg [kg/s] 45
Euler work, weul [KJ/kg] 90
Degree of reaction, Rρ 0.5
ψ = V2t/U 1
Geometry parameters NGV Rotor
Blade pitch, t [m] 0.045 0.033
Blade height, l [m] 0.04 0.04
Blade chord, c [m] 0.04 0.03
Blade maximum spanwise
thickness, smax [m] 0.008 0.008
Gas Turbine Inlet
Boundary Conditions
Turbine Inlet Temperature [K] 1672
Gauge pressure [bar] 18
Mass flow-rate [kg/s] 0.65
Flow velocity [m/s] 136
Table 2. Selected gas turbine first stage
NGV and rotor blade geometry parameters
Table 1. Selected gas turbine first
stage configuration parameters
Table 3. Inlet boundary conditions
assigned for the simulations
Figure 4. NGV and rotor blade midspan section
Figure 5. Contours of Static Temperature in the stator and rotor channels and representation
of the channel mid-streamlines (blue lines) on which temperature values are extrapolated
8
The temperature values on the stator and rotor channel mid-streamline section corresponding to the
real gas expansion (blue lines in Fig. 5) are extrapolated and imported in the MATHEMATICA© code
for the cooling analysis. The obtained expansion gas temperature profiles Tg(x) are plotted in Fig. 6
as a function of a non-dimensional chord length, going from 0 to 1, on the x-axis.
4.1 Internal Convection Cooling Results
Blade Temperature Calculation
According to the set of equations presented in section 3.1, knowing the gas temperature profiles Tg(x),
a trend for the external convective heat transfer coefficient hext(x) is needed to compute the blade
external surface temperature Tbe(x) from eq. 3.4. Many typical trends can be found in the literature,
most of them of experimental origin. The following trends for the NGV and the rotor blade, given by
Han [6], were selected for our calculations. The obtained metal temperature profiles on the NGV and
rotor external surface Tbe(x) are plotted in Fig. 8, while their values are listed in table 4.
x 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
NGV Tbe(x) [K] 1229 1199 1123 1130 1075 1010 994 960 990 1092 1215
Blade Tbe(x) [K] 1233 1183 1192 1104 1021 1192 1052 1043 1061 1210 1222
Figure 6. Gas temperature profiles along stator (a) and rotor (b) channel mid-streamline
0.0 0.2 0.4 0.6 0.8 1.0
1640
1650
1660
1670
Tg(x)
0.2 0.4 0.6 0.8 1.0
1610
1620
1630
1640
Tg(x)
(a) (b)
Figure 7. External convective heat transfer coefficient trend along the NGV (a) and blade (b) chord
0.0 0.2 0.4 0.6 0.8 1.0
500
600
700
800
900
1000
1100
hext
0.0 0.2 0.4 0.6 0.8 1.0
400
600
800
1000
hext
(a) (b)
Table 4. External surface temperature values along the non-dimensional chord x
T [K] T [K]
x x
x x
h [W/m2K]
h [W/m2K]
9
Thermal gradients within the NGV and blade
material along the chord (red line in Fig. 9),
were computed by eq. 3.6, with a material
thermal conductivity equal to 22 W/(mK)
(common value for Ni-based superalloys). As a
consequence, knowing Tbe(x), the temperature
profiles Tbi(x) along the chord on NGV and
blade internal surfaces (Fig.10) can be
computed by subtracting such thermal gradients
from Tbe(x).
Creep test on rotor blade
Considering the most advanced technology for the selected turbine, a third-generation Single Crystal
(SC) superalloy, which is currently used for gas turbine blades, is chosen for the test: CMSX-10.
Rotor blades are subjected to both mechanical stresses induced by centrifugal force and thermal
stresses due to thermal gradients within blade material. Therefore, the maximum total stress can be
expressed by relation 4.1, and according to the selected turbine parameters (Table 1), the computed
maximum thermal gradient value within the blade material (ΔTmax=23 K) and CMSX-10 density
equal to 9.05 g/cm3, we obtain:
𝜎𝑡𝑜𝑡 = 𝜎𝑐 + 𝜎𝑡 =𝜌𝜔2
2(𝑟𝑒𝑥𝑡 − 𝑟) ∗ 10−6 + 𝐸 𝛼 ∆𝑇𝑚𝑎𝑥 = 130 𝑀𝑝𝑎
0.2 0.4 0.6 0.8 1.0
1000
1050
1100
1150
1200
1250 Tbe(x)
(a) (b)
0.0 0.2 0.4 0.6 0.8 1.0
1000
1100
1200
1300
Figure 8. NGV (a) and blade (b) external surface temperature profiles
Tbe(x)
Figure 9. Internal paths representing NGV (a)
and blade (b) chord
(a) (b)
0.2 0.4 0.6 0.8 1.0
900
1000
1100
1200 Tbi
(x)
0.2 0.4 0.6 0.8 1.0
900
1000
1100
1200
1300
Tbi
(x)
(a) (b)
Figure 10. NGV (a) and blade (b) internal surface temperature profiles
(4.1)
T [K] T [K]
x x
x x
T [K] T [K]
10
The resulting Stress/Density ratio is equal to 14.4
MPa/g/cm3, which, from the CMSX-10 Larson-Miller
curve, corresponds to a LMP value equal to 30.2 (Fig.
11). Applying the LMP relation (eq. 3.5) for a blade
lifetime expected to be equal to 24000 operating hours,
and assuming material constant C=20, the maximum
allowable temperature is given by:
𝑇𝑏𝑒𝑚𝑎𝑥=
𝐿𝑀𝑃 ∗ 103
(𝐶 + log 𝑡𝑟) = 1235 𝐾
Such value is higher than the maximum Tbe(x) values at
the leading edge of both the rotor blade (Tbemax=1233 K),
and the NGV (Tbemax=1229K). Therefore, such superalloy
is proper to be used as it prevents the blade from creep for
the estimated time.
Preliminary assessment of cooling air flow rates
The cooling system must be designed according to Tbi(x) trend. In particular, the cooling air has to be
injected in the cylinders that are in the coldest portion of the blade. A common practice is to split the
total amount of injected coolant into two or three portions, each of them following a specific winding
path through the internal channels. It was decided to split the coolant into two equal mass flows in
the NGV and into three mass flows in the blade, following the paths shown in Fig. 12. For a coolant
inlet temperature equal to 750 K, the coolant temperatures along each cylinder and thus the blade
internal temperature profiles Tbint(x), shown in Fig.13, can be computed as explained in section 3.1
for different cooling mass flow rates, i.e. for i going from 0.05 to 0.09.
Figure 11. Larson-Miller stress-rupture
curves for third generation SC Ni-based
superalloy CMSX-10 [3]
Figure 12. Coolant paths in NGV (a) ad rotor blade (b) cooling cylinders
mc1 mc2 mc1 mc2 mc3
(a) (b)
0.0 0.2 0.4 0.6 0.8 1.0
1000
1050
1100
1150
1200
1250
Tbint(x)
0.0 0.2 0.4 0.6 0.8 1.0
950
1000
1050
1100
1150
1200
1250
Tbint(x)
(a) (b)
Figure 13. Internal temperature profiles along NGV (a) and rotor blade (b) chord for each value of
i (0.06-0.09)
x x
T [K] T [K]
11
The cooling air flow rates were
assessed according to the
procedure presented in section
3.1. Plots of the resulting
standard deviation values for
each i are given in Fig. 14 for
both the NGV and rotor blade.
The values of i for which the
standard deviation is lower are
respectively 0.075 and 0.06.
4.2 Film Cooling Results
Blade Temperature Calculation
A film cooling efficiency equal to 0.4 was introduced to compute the film temperature values Tf for
each hole from eq. 3.20. A film temperature profile Tf (x) for each value of i going from 0.05 to 0.1
(Fig. 15) was hence obtained by joining consecutively such Tf values. The NGV and rotor blade
external surface temperature profiles Tbe(x) were computed for each value of i going from 0.05 to 0.1
(Fig. 16) according to the procedure presented in section 3.2.
Preliminary assessment of cooling air flow rates
The first value of i that provides a temperature profile Tbe(x) that has a maximum temperature below
the maximum allowable by the material (1235K) resulted to be equal to 0.06 for the NGV and to
0.055 for the rotor blade.
Figure 14. Standard deviation values for each i in the NGV (a)
and rotor blade (b)
0 2 4 6 8
50
60
70
80
90
i = 0.075
0 2 4 6 8 10 12
60
80
100
120
i = 0.06
(a) (b)
0.0 0.2 0.4 0.6 0.8 1.0
1380
1400
1420
1440
1460
1480 Tf(x)
0.0 0.2 0.4 0.6 0.8 1.0
1340
1360
1380
1400
1420
1440
1460
1480
Tf(x)
(a) (b)
Figure 15. Film temperature profiles for each i (0.05-0.1) in the NGV (a) and rotor blade (b)
0.0 0.2 0.4 0.6 0.8 1.0
1000
1050
1100
1150
1200
1250
1300
Tbe(x)
0.2 0.4 0.6 0.8 1.0
1000
1050
1100
1150
1200
1250 Tbe(x)
Figure 16. NGV (a) and rotor blade (b) external surface temperature profiles for each i (0.05-0.1)
x x
x x
T [K] T [K]
T [K] T [K]
12
i 0.05 0.055 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 0.1
Tbemax
[K](x=0)
NGV 1253 1239 1227 1220 1214 1209 1205 1201 1197 1194 1190
Blade 1248 1228 1218 1207 1197 1288 1280 1274 1268 1263 1258
5 Conclusions All the coolant mass flow rates estimated by the model are in accordance with some well-known
considerations, which confirms the reliability of the model. When internal cooling alone is used, much
more cooling air is required, while in both cooling methods less coolant is needed to cool the rotor
blades with respect to the stator vanes. Therefore, despite of its limitations, the method outlined in
this thesis is a useful design tool when the goal is to obtain a sufficiently accurate first estimate of the
cooling requirements in a quick and reliable way, with which a preliminary cycle calculation aiming
at efficiency prediction can be done. Furthermore, the greatest advantage of the model is that it can
be easily adapted to any operating condition as well as different process parameters, cooling methods,
fluid properties and blade geometries. Such versatility is particularly well suited for the last
technological trends, namely the investigation of new cooling methods (like porous and transpiration
cooling) and alternative coolants instead of air.
6 References [1] Bolland O., Stadaas J.F., Comparative evaluation of combined cycle and gas turbine systems
with water injection, steam injection, and recuperation, ASME Journal of Engineering for Gas
Turbines and Power; 108, 151-159.
[2] Cerri G., Preliminary turbine cooling requirement, Energy (2008).
[3] Caron P., Khan T., Evolution of Ni-based Superalloys for Single Crystal Gas Turbine Blade
Applications, Aerospace science and technology 1999; 12, 1-9.
[4] De Paepe M., Dick E., Cycle improvements to steam injected gas turbines, Int. J. Energy Res.
2000; 24: 1081-1107.
[5] El-Masri M.A., On thermodynamics of gas turbine cycles: part 2 - A model for expansion in
cooled turbines, J. Eng. Gas Turbines and Power 1986; 108, 151-159.
[6] Han J-C., Fundamental gas turbine heat transfer, J Therm. Sci. Eng. Appl. 2013; 5(021007):1-
15.
[7] Han J.C., Dutta S., and Ekkad S.V., Gas Turbine Heat Transfer and Cooling Technology, Taylor
& Francis, Inc., New York, December 2000, ISBN: 1-56032-841-X, 646 pages.
[8] Han, J. C., Ekkad, S. V., Recent developments in turbine blade film cooling, International Journal
of Rotating Machinery 2001; 7(1), 21–40.
[9] Han, J. C., Park, J. S., and Lei C. K., Heat transfer enhancement in channels with turbulence
promoters, ASME Journal of Engineering for Gas Turbines and Power 107(1);628–635.
[10] Han J. C., Wright L. M., Enhanced internal cooling of turbine blades and vanes, The Gas
Turbine Handbook, U.S. Department of Energy, Office of Fossil Energy, National Energy
Technology Laboratory (NETL), 2006, pp. 321– 354.
[11] Horlock J.H., Watson D.T., Jones T. V., Limitations on gas turbine performance imposed by
large turbine cooling flows, ASME Journal of Engineering for Gas Turbine and Power 2001;
123, 487-494.
[12] Jonsson M., Bolland O., Bücher D., Rost M., Gas turbine cooling model for evaluation of novel
cycles, In: Proc. ECOS 2005, Trondheim, Norway, June 20-22; 2005.
Table 5. External surface temperature maximum values at the NGV and rotor blade leading
edge (x=0) for each value of i (0.05-0.1)
13
[13] Jordal K., Bolland O., Klang A., Aspects of cooled gas turbine modeling for the semi-closed
O2/CO2 cycle with CO2 capture, Journal of Engineering for Gas Turbines and Power 2004; 126,
507-515.
[14] Sciubba E., Air-cooled gas turbine cycles - Part 1: An analytical method for the preliminary
assessment of blade cooling flow rates, Energy (2015).