Sie sind auf Seite 1von 10

Proceedings of ASME

Proceedings Turbo Expo 2006


of GT2006
May 8-11, 2006, Barcelona,
ASME Turbo Expo 2006: Power for Land, Sea and AirSpain
May 8-11, 2006, Barcelona, Spain

GT2006-91039
GT2006-91039

THE INFLUENCE OF HEAT TRANSFER EFFECTS ON TURBINE PERFORMANCE CHARACTERISTICS

A. G. Stamatis K. Mathioudakis

Laboratory of Thermodynamics & Laboratory of Thermal Turbomachines


Thermal Engines. National Technical University of Athens
University of Thessaly Iroon Plytechniou 9, Athens 15780, Greece
Pedion Areos, 38834 Volos, Greece E-mail: kmathiou@central.ntua.gr
E-mail: tastamat@mie.uth.gr

ABSTRACT due to thermal expansion (tip clearance effects being the


prominent among those, as it is clearly influenced by
A method allowing the evaluation of the effects related to
dimension changes caused by thermal dilatation).
heat transfer to the turbine blades on its performance
Effects of heat transfer on boundary layer growth were
characteristics is presented. The effects investigated are the
considered only for compressors [5]. The effects of heat
change of passage dimensions, resulting from heat transfer and
transfer onto the performance characteristic of a turbine have
the change in flow field, exhibited mainly as a different
been considered to be very small for a number of reasons. The
boundary layer development. Change of hot gas temperature
main reason is that the effect of thermal dilatation in a turbine
combined with cooling air temperature and possibly flow rate,
passage is in an opposite direction to the effect of boundary
result in a change of the temperature of the blade material,
layer growth. Smaller wall temperature means smaller passage
leading to dimension changes, because of the thermal expansion
dimensions. On the other hand, heat transfer into the wall leads
(dilatation). The changes in dimensions have a direct effect on
to smaller displacement thickness of the boundary layer, which
turbine performance. An immediate consequence is a
means less blockage due to viscous effects. The two
modification of the mass flow characteristic, due to a change of
mechanisms are thus in opposite direction, and consequently
the throat area. Heat transfer also influences the properties of the
the overall effect has been considered to be less pronounced.
gas flowing through the passage and in particular the
An additional reason is that large turning angle of the blades
characteristics of the boundary layers developing on the nozzle
means that small changes in deflection angles represent a
vanes and hub, tip endwals. Change of the thickness of this layer
smaller relative change (than they would, for example,
results in a change of blockage through the passage, a fact that
represent in compressors), while the number of turbine stages
influences directly the turbine flow function. The influence of
is usually small, so the cumulative effects on a turbine is not
both effects on turbine performance is studied. The study is
significant.
performance oriented, aiming to the derivation of simplified
The objective of the present work is to derive a method
models, which can be introduced in engine cycle decks
that allows inclusion of the physical effects related to turbine
blade heat transfer in engine performance cycle decks. A better
INTRODUCTION understanding of engine test data can thus be achieved, while
this type of modeling can be introduced already during the
The impact of heat transfer effects on the performance of early steps of preliminary performance studies and allow more
gas turbines has been extensively studied. Results of such accurate predictions.
studies were presented in a number of papers published during The particular aspects of heat transfer influence that
the late seventies and the eighties by MacCallum and will be investigated here are:
Pilidis[1,2,3,4]. Different effects were studied, namely the effect Thermal dilatation (taking into account cooling flows)
of heat soakage of the components, the effects of size changes Effect on boundary layer development

1 Copyright ©2006 by ASME


Incorporation of these effects in performance modeling total pressure becomes fixed, and the equation takes the
allows the deduction of: following form:
(a) (pressure ratio)-(mass flow rate) turbine performance ⎡ 2 γ +1 ⎤
1
2
m Tt 2γ ⎢⎛ 2 ⎞ γ −1 ⎛ 2 ⎞ γ −1 ⎥ (2)
maps, at certain conditions, from maps obtained in other q= = CD A ⎜⎜ ⎟⎟ − ⎜⎜ ⎟⎟
conditions. For example: when the map has been obtained in a Pt (γ − 1)R ⎢⎢⎝ γ + 1 ⎠ ⎝γ +1⎠ ⎥

⎣ ⎦
cold rig, the map in actual working conditions can be deduced,
This is why for pressure ratios above some critical value,
with or without blade cooling.
the reduced mass flow rate q takes a constant value.
(b) maps during transients, when gas and blade temperatures
Equations (1) and (2) show the way that mass flow
vary significantly, and are away from their equilibrium values,
function is related to geometry and viscous effects. For a
from maps known at steady state conditions and thermal
chocked turbine, equation (2) shows that flow function does
equilibrium.
not depend on pressure ratio and is linearly related to throat
Before we proceed to the detailed examination of the two
area A and discharge coefficient CD. Equation (1), on the
effects, it is useful to discuss briefly some principles that govern
other hand, shows that this statement can be extended to an
the behavior of the characteristic curves for (mass flow
unchocked nozzle, as long as the ratio of static to total pressure
function)-(pressure ratio) interdependence.
is the same. We will postulate that this ratio will remain the
same when the overall pressure ratio of the stage remains the
The physics for turbine map constitution same. The implication is that when the (pressure ratio)-(mass
flow) characteristic of a turbine is considered, the change in
The (pressure ratio)-(mass flow) characteristic of a turbine
reduced mass flow rate for a particular value of pressure ratio
is similar to the one of a nozzle. This is a consequence of the
will be linearly related to changes in area and discharge
fact that flow is accelerated through successive passages of the
coefficient. We will have thus a relationship of the form:
turbine blade rows. In particular, the form is determined by the
δq δA δC D
flow through the passages of one of the blade rows, the one that = + (3)
chokes first, as the mass flow rate increases. The physics of the q A C D
flow, leading to the derivation of typical characteristic curves The resulting change of a characteristic curve can
has been explained in [6]. qualitatively be shown in Figure 2.
An analysis of the way performance curves can be generated In the following sections we will investigate how changes
for a turbine stage has been presented in the early work of δΑ/Α in the area, and δCD/CD in the discharge coefficient, can
Kochendorfer and Nettles [7]. A feature of this formulation is be estimated.
that the mass flow through a passage is represented through an δq
equivalent nozzle and the mass flow is derived using an one q
dimensional formulation. It is calculated by considering that a
blade passage behaves like a one-dimensional nozzle, fig 1. δΑ
Α
δC
C

Figure 2: Representation of change in turbine characteristic


for changes in area A and discharge coefficient C D
Figure 1: A turbine blade passage and its equivalent nozzle
EFFECT OF THERMAL DILATATION ON TURBINE
PERFORMANCE
At some section of a nozzle, where static pressure is equal
to p, mass flow rate can be written as: Evaluation of dimension changes due to thermal
2
1
γ +1 ⎤ 2
dilatation requires the determination of the temperature of a
⎡ turbine nozzle passage walls. The approach chosen here is to
m Tt 2γ ⎢⎛ p ⎞ γ ⎛ p ⎞ γ ⎥ (1)
= CD A ⎜ ⎟ −⎜ ⎟ employ correlations available in the open literature for this
pt (γ − 1)R ⎢⎢⎜⎝ pt ⎟⎠ ⎜⎝ pt ⎟⎠ ⎥⎥ purpose.
⎣ ⎦
CD is the discharge coefficient and A is the area of the
nozzle at that section. CD is used to account for the fact that Proposed method for predicting dilatation effect.
flow is not uniform across the passage. It accounts therefore for
the thickness and the form of boundary layer profiles developing As a starting point, it is considered that a turbine map has
along the passage walls. been measured or calculated at certain datum blade and
This relation is usually written for the throat section. The endwall temperatures. For these temperatures turbine nozzle
advantage of writing it at that section is that, when sonic throat area has a certain value resulting from the span and
conditions are reached at the throat, then the ratio of static to height dimensions corresponding to these temperatures.
Thermal dilatation occurs when operating conditions of the
2 Copyright ©2006 by ASME
engine alter these temperatures and new dimensions modify the Therefore the map correction requires knowledge of the
throat area, as shown schematically in figure 3. The throat on the following groups of data:
blade-to-blade plane is shown in fig. 3.a and on the actual 3-D Passage Geometry : ( H, Sc, Sh)
passage in fig. 3.b. The view on the plane of the throat is shown Metal properties: (Kb,Kc,Kh)
in fig. 3.c.The throat area changes due to change both on the Temperatures: (Tmd,Tm )
blade-to-blade plane (S) and the passage height (H). In the previous relations the blade, hub and casing
material are treated as having a uniform temperature. This is
not physically precise, since the temperature distribution
during operation is non-uniform. The relations can be
sufficiently accurate, however, by taking an average
temperature for the calculations.

Calculation of metal temperatures


Metal temperatures have to be determined through heat
transfer modeling. In order to take into account for transient
conditions it is proposed to follow an approach similar to the
one used by Stamatis et al., [15], with improvements
concerning the gas temperature used as well as incorporating
cooling arrangements.
The approach is illustrated in the figure 4 and is based on
the following reasoning:

Tg gas Tw
Figure 3: Throat area and its variation due to dilatation
TBC

Calculation of Throat area change


Tm
The area change is caused by the dimension changes due to Tc
dilatation of the blade height and the hub and casing segments of coolant
the blade passage. The area is given by the relation: A= (Sh+Sc) Figure 4: Blade wall and gas temperature configuration.
H/2
Which gives for area change: Heat addition to metal is equal to the heat transfer from
ΔA = ΔS + ΔH gas to metal (taking into account possible film cooling
arrangement) minus the heat transferred from metal to coolant
convectively. (It is assumed a uniform metal temperature)
where, the percent variation of span (S) and height (H) of
Therefore metal temperatures are given through the
the throat are:
solution of the following equation written correspondingly for
blade, casing and disc.
ΔS = (δS c + δS h ) /( S c + S h )
ΔH = δH b / H dT m
c pm M = h g A g (T aw − T w ) + hc A c (T c − T m ) (4)
dt
The changes in linear dimensions are calculated from the T gin + T gout
T avg =
equations for thermal expansion of the dimensions of blade, 2
casing and hub, and are given from the following relations: h g = 0.0201 ⋅ C ⋅ Re 0.8 ⋅
k
L
δ S h = S h K hδ T m h Re =
wg L
δ S c = S c K cδ Tm c Acs μ
0. 669
δ H = H K bδ Tmb μ = 4 × 10 −7T avg
0.78
k = 3 × 10 −4T avg
Where hg: gas to blade heat transfer coefficient.
K is the metal expansion coefficient Tw: chord wise mean wall temperature
δTm=Tm-Tmd hc: blade to coolant heat transfer coefficient.
Tmd: the metal datum temperatures, at which turbine map Tc: bulk temperature of the coolant.
was predicted or measured L, C, Acs, Ac, Ag, M, cpm given data for the particular engine: L
Tm : current metal temperature appropriate characteristic length, C a constant, Acs cross
sectional flow area, Ag surface area of metal exposed to gas,

3 Copyright ©2006 by ASME


Ac Area exposed to coolant, M mass of metal, cpm specific heat Geometry: (M:Metal Mass, Ag:Metal Area exposed to Gas,
of metal Ac:Metal area exposed to coolant)
Various temperatures appearing in equation (4) could be Flow: (Toe : Total gas temperature at NGV inlet, Me:Mach
expressed using more convenient parameters, ([17-21]), such as number at NGV inlet, Toc: Inlet coolant temperature, assumed
the following: equal to total temperature at compressor exit)
Convective cooling efficiency Heat transfer :(hg , hc:Calculated using data L, C, Acs, for both
nc =
T c − T oc (5) gas and coolant , nc , εf :Given for existing configuration or
T m − T oc calculated using correlations.)
The above formulation is general enough and describes
Horlock et al, [16], claim that in practice there is relatively also the following simpler configurations:
little variation in cooling efficiency 0.6 to 0.8, and that a given
Steady State: T m = ( crT r + ccT oc ) / cm
form of cooling geometry gives sensibly a constant value over
the range of mass flows considered.
The resulting Tm could also be the initial value for a transient.
Film cooling effectiveness
T r − T aw (6) No thermal barrier coating: BTBC = 0
εf =
Tr −Tc c m = h g A g (1 − n c ε f ) + hc A c (1 − n c )
Taw : the adiabatic wall temperature that represents the cr = h g A (1 − ε f )
surface temperature of a perfectly insulated surface. cc = (hg Ag ε f + hc Ac )(1 − nc )
Young and Wilcock, [21], treat film cooling effectiveness as
a known empirical parameter whose value (typically 0.2-0.4) No film cooling: εf =0
reflects the level of film cooling technology. c m = hg A g + hc Ac (1 − nc )
Thermal Barrier Coating (TBC) Biot number,[21] cr = hg A g
hg tTBC (7) c c = hc Ac (1 − n c )
BTBC =
kTBC No cooling: hc = 0
tTBC: Thickness, kTBC: Thermal condactivity cm = hg A g
Assuming that heat tranfer rate from the gas to TBC is equal to
heat conducted through TBC implies cr = hg Ag
hg tTBC T w − T m cc = 0
=
kTBC T aw − T w (8) Indicative Application - Sensitivity Analysis
A preliminary sensitivity analysis has been carried out in
Using equations (5) to (8), equation (4) can be re-written as order to estimate the effect of the dilatation on percent area
dT change (which reflects on swallowing capacity change as
c pm M m = −cmT m + crT r + ccT oc
dt (9) shown above), and consequently the percent change in the
where turbine mass flow rate parameter caused by this effect. Data
hg A g were assumed to represent a typical turbofan engine, [15].
cm = (1 − n ε ) + h A (1 − n )
c f c c c
(10)
The data are separated in two groups for the purpose of
1 + BTBC
the present study: First, a group of data was considered fixed
hg A g (11) for the calculations that follow. A second group contains data
cr = (1 − ε )
f that are assumed to vary in a certain range around an initial
1 + BTBC
estimate. The values of this second group are used to carry out
⎛ hg A g ⎞ (12) a parametric study. The purpose of this study is to establish the
cc = ⎜⎜ ε f + hc Ac ⎟⎟(1 − nc ) sensitivity of the proposed method, to the accuracy of
⎝ 1 + B TBC ⎠
knowledge of certain parameters. Parameters that are not
Recovery temperature is given from accurately known, but are estimated according to existing
γ −1 2 ⎤
knowledge are studied. We can thus assess the impact of the

⎢1 + r 2 M e ⎥ accuracy of certain parameters on the final map correction
Tr = Toe ⎢
γ −1 2 ⎥ produced by the method.
⎢ 1+ Me ⎥
⎣ 2 ⎦ (13) The data used for calculation at a point that will be used as
where the recovery factor r is a reference for comparison of the results of the parametric
r = PR 1 2 ( Laminar ) PR 1 3 ( Turbulent ) studies are given in Appendix-I.
The calculated percent changes for throat area and mass
[6], and may be approximated as
1 1
flow parameter are:
r = αPR 2
+ (1 − a )PR 3
(14) ΔΑ= -0.436 % Δqdilatation= - 0.436%
with α an assumed constant and Prandl number is
μ g c pg
PR = Parametric Study
kg
(15)
The results of the parametric study are illustrated in figure
Summarizing, it is obvious from Equations (9) - (15) that
5 for the varying parameters considered, which were chosen
metal temperatures can be calculated when the following groups
mainly according to the uncertainty in knowing or calculating
of data are given:
4 Copyright ©2006 by ASME
them accurately. From this figure the following conclusions can
be drawn: convective cooling efficiency (ηc) film cooling effectiveness (εf)
- The most influential parameter is the map datum temperature. 0.5 0.6 0.7 0 0.2 0.4 0.6
Exact knowledge of this parameter is crucial. 0 0

- Cooling has a considerable effect with respect to no cooling -0.2

% Area change
-0.2

% Area change
case(εf=0,hc=0,tTBC=0). -0.4 -0.4

- The uncertainty on knowledge of the rest of the parameters -0.6 -0.6

seems to have a minor impact (~0.2 %) for the broad range of -0.8 -0.8

variation considered. Therefore a negligible error (~0.1%) is -1 -1

expected when assuming fixed and their values are the ones at
the middle of the range considered. TBC thickness (tTBC ) Cooling Area (Ac)
0 0.001 0.002 1 1.5 2
0 0
THE EFFECT OF ALTERATION OF THE FLOW FIELD -0.2 -0.2

% Area change

% Are a change
The flow field though a passage with adiabatic walls differs
-0.4 -0.4

from that of a passage with heat transfer, when subjected to the


-0.6 -0.6

same inlet total pressure to outlet static pressure ratio . The


-0.8 -0.8

-1 -1
reason for this behavior is that fluid properties change with
temperature. The two properties influenced are density and Coolant-metal Heat Tran. coef. (hc)
Gas Metal Heat tranfer coef. (hg )
viscosity. 1000 2000 3000 4000 5000 6000 7000
The influence of a change in density is apparent even from a 0 0

one-dimensional consideration of the flow. The fact that the -0.2 -0.2

% Area change
% Area change
density is different means that for the same pressure ratio a -0.4 -0.4

different mass flow rate will pass through the nozzle. -0.6 -0.6

Change in viscosity influences the development of the -0.8 -0.8

boundary layer. At the same location the boundary layer can -1 -1

have a different thickness, while the velocity profile may also be


different. Mach Number (Me) Coolant temperature. (Toc)

Both effects may result in a change of the displacement


0.1 0.2 0.3 800 850 900
0 0
thickness, which is the quantity determining the blockage caused -0.2 -0.2
% Area change

% Area change
by the boundary layer development. -0.4 -0.4
Heat transfer effects on the development of boundary layers, -0.6 -0.6

from the point of view of the consequences on turbomachinery -0.8 -0.8

performance, have been examined in [8,9,10]. It must be -1 -1

commented here that boundary layer calculation in turbines has


received a lot of attention, because of the need to understand, Map Datum Temp. (Tmd )
No cooling no film

predict and improve blade cooling schemes. In all these studies


no convective Cooling
1200 1300 1400 1500

however, the effect on the displacement thickness of the


0.5
0

boundary layers is overlooked, because the main interest is in


0.3
% Area change

-0.1
% Area change

0.1
heat transfer effectiveness. -0.1
-0.2

The effects of heat transfer on the growth of boundary


-0.3
-0.3
layers, including the point of view of displacement thickness,
-0.4
-0.5
was investigated in the early work by Grant [8], who found that
-0.5

wall heating leads to boundary layer thickening. He also found Figure 5: Area change as a function of various flow parameters
that it favors separation but observed some interaction with the
magnitude of free-stream velocity that could not be explained.
The investigation focused on compressor boundary layers. questionable how it could be applied over a stationary nozzle,
The only work that reports an investigation of the effect of where no work transfer exists. Even though this procedure is
heat transfer on the blockage caused by the boundary layer mentioned, a different one is proposed further in the paper. It
growth, modified due to the presence of heat transfer, is the is a procedure that was developed in an earlier published paper
work of MaCallum [9]. In this work he investigated heat transfer (Hurris and Luxton [10]) and is based on boundary layer
effects on all components, devoting more attention to the effects calculations, in the presence of heat transfer. Hurris and
on compressors.He mentions the effect on the turbine mass flow, Luxton report that when a boundary layer develops, the
by quoting two approaches. One approach he mentions takes displacement thickness is reduced when the surface is cooled.
into account the presence of heat transfer in a one dimensional They present the results of calculations supporting this
manner and provides a formula for estimating the discharge assessment and they suggest that one dimensional channel
coefficient of a nozzle in the presence of heat transfer. Friction flow could be used to calculate this effect. (They give an
effects are ignored in this approach and the change calculated is example of calculating this effect. (They give an example of
due to the fact that heat transfer changes the density of the fluid. T
The formula he presents, includes a parameter that non- calculations with Sw = w − 1 = −0.46 , stage pressure ratio
Tr
dimensionalizes heat transfer over work exchange, and it is thus

5 Copyright ©2006 by ASME


1.1, M 2 = 0 ,55 , which they find to produces a 1% mass flow An(T0 Tw )
1−
increase). McCallum produces some figures for calculations he 7 ,08 A 1 A′ An 2
CD = 1 − + ⋅ ⋅
has performed and mentions that changes in the HP turbine Re D D 4 D 2 − T0 Tw
swallowing capacity influence the surge line. At his conclusions, We observe that this formula can be written as follows
however, reports that the estimations were preliminary and that An(T0 Tw )
further work would be required in this direction. 1−
Examination of similar solutions in compressible boundary C D = 1 − (1 − C D ,ad ) An 2
2 − T0 Tw
layers [12], lead to the conclusion that displacement thickness is
or equivalently
reduced with wall temperature and may even become negative
An(T0 Tw )
for very small temperatures. This effect was attributed by the 1−
1 − CD An 2
authors to the increase in density near the wall, which results in a =
layer mass flow rate, compared to the one when the boundary 1 − CD ,ad 2 − T0 Tw
layer gas is hot. The parameter considered to express the effect This expression establishes a relationship for evaluating
of heat transfer is Sw, defined as the effect of heat transfer on the discharge coefficient. It is
T noted that the factor expressing the change of the discharge
Sw = w −1 (T in K ) (16) coefficient from the adiabatic conditions is a function of the
T0
ratio of the wall and free stream temperatures. This ratio is
Tw is the wall temperature and T0 is the temperature of the conveniently represented by the parameter Sw introduced
gas. above.
This parameter is actually used in all studies of this type, to A graphical representation, for different values of the
express the magnitude of the effect of different wall temperature adiabatic discharge coefficient is given in Figure 6 . It is
on boundary layer development. indicated here too that lowering the wall temperature increases
A quantity directly related to blockage caused by the the discharge coefficient. What is of more interest is that the
development of boundary layers along the walls is the discharge interdependence of discharge coefficient and Sw is of a rather
coefficient. The discharge coefficient can be considered to be a simple form. For temperatures that do not depart much from
parameter more suitable to characterize the effect of viscous the free-stream temperature (for example in the range of –
action in the passage. The reason is that it embeds all viscous 0.5<Sw <0.5), the relationship is close to linear.
effects, including blade surface and endwall boundary layers as
well as corner flows, secondary flows. From the relation for CD, the quantity δCD/CD can be
The parameters determining the discharge coefficient of a derived. If we take as a reference condition the condition with
nozzle have been discussed by several authors. The interest for SW=0, then the values of Figure 6 can be used to derive
such discharge coefficient stems from applications in other δCD/CD. Such values are plotted in Figure 7. For plotting this
areas. One such area is rocket propulsion, where converges- figure, we have chosen a range of variation of SW that would
divergent nozzles are employed. Here there is an interest in heat represent expected variation according to temperatures
transfer effects, since large temperature differences exist encountered in gas turbines today, extended to cover possible
between the hot gases and the nozzle walls. Another area is the advances in the field. Also, discharge coefficients are chosen
study of nozzles used in mass flow measurements. Finally, with values that would be closer to what is expected in a
nozzle flows and related effects have also been studied in jet turbine cascade (a typical value of 0.98 was used by
engines, as the exhaust nozzle is an important component of a jet Kochendorfer F.D., Nettles [7])
engine.
A simplified analysis for determination of the discharge 1

coefficients of circular axisymmetric nozzles with laminar


Cd

0.98 Cd,adiab
boundary layer is presented in [11]. Although the analysis may 0.99
be considered rather simplistic, it presents very clearly the 0.96 0.98
concept and the way of deriving the discharge coefficient. 0.96
Moreover, it provides analysis for assessing the effect of heat 0.94
0.94
transfer to the discharge coefficient. Of course, in this case too 0.92
0.92

simplifying assumptions are used, so the results obtained are


only of qualitative nature for our purposes, indicating the 0.9
parameters upon which the influence is expected to depend.
The relationship derived for the discharge coefficient for no 0.88
-1 -0.5 0 0.5 1
heat transfer is: Sw=(Tw/T0-1)

Figure 6: Heat transfer effect on the discharge coefficient,


7 ,08 A 1 A′ calculated by the method of [11].
( C D )ad = 1 − + ⋅
Re D D 4 D
For given wall temperature Tw, and freestream temperature
T0, the author proposes the following formula:

6 Copyright ©2006 by ASME


1.4 actual cooled turbine conditions. Predictions may also be run
δCD/CD (%)

Cd,adiab
at actual conditions. It is therefore more suitable that the
1.2
0.99 changes are expressed in function of a parameter that produces
1 0.98 a picture of deviations from reference conditions. A
0.96 formulation that would be suitable for this purpose is the
0.8
following:
0.6 ⎛ Tw ⎞
⎜⎜ ⎟⎟
0.4 ⎝ T0 ⎠ ACTUAL (17)
S rw = −1
⎛ Tw ⎞
0.2 ⎜⎜ ⎟⎟
⎝ T0 ⎠ REFRENCE
0
-0.5 -0.4 -0.3 -0.2 -0.1 0 T0: the free stream gas temperature, ACTUAL: the current
Sw=(Tw/T0-1) operating condition, REFERENCE: the condition at which the
Figure 7: Fractional change of discharge coefficient, calculated map was estimated, Tw: the temperature of the wall.
from the values of Figure 6. Tw is an average temperature of the walls of the blade
passage passage. Average temperatures have been calculated
In view of the above analyses, a procedure is proposed for for blade surfaces and hub, tip endwalls, by the method
accounting for the effects of heat transfer a flow through turbine previously presented. A suitable average for the overall
passages. passage is considered to be a weighted average, according to
the surface represented by each temperature and is given by
the following relation:
A method to be used for turbine blade passages HTBW + S hTHW + S C TCW
Tw =
In view of the results and the mechanisms presented above, H + S h + SC
a procedure that is considered to give the possibility for accurate where: H is the blade height, Sh span at hub, Sc span at
reconstitution of the effect of heat transfer is now proposed. casing, see Figure 3. TBW is the blade wall temperature, THW
Since simplified boundary layer or axisymmetric flows hub wall temperature, TCW casing wall temperature. Since the
resemble but are not well representative of the flow in a turbine blade passages we consider are of small hub-tip ratio, we can
passage, in order to derive information about how the flow is consider that Sh=Sc=S , which transforms the previous
altered by the presence of cold walls, a more realistic model has equation to
to be used. Advantage can be taken from present day computing ( 2 H / S )TBW + THW + TCW
capabilities of flows in turbines. Computational tools using Tw =
solution to the Navier Stokes equations, including heat transfer 2H / S + 2
effects can be used. By using such a model and performing rans This parameter obtains zero value when the actual
at different wall temperatures, a correlation between mass flow condition considered is the one for which the map has been
rate and wall temperature can be derived. evaluated. On the other hand, it is a unique function of the
The quantity that is most suitable for representing the effect blade/gas temperature ratio, and is thus suitable for correlating
on the performance characteristics is the discharge coefficient. It the effect, according to our previous comments.
incorporates in a global manner the change for the mass flow in
the passage, while it is directly linked to the mass flow capacity. Sample calculation example
The parameter that is considered to be appropriate for
representing the effect of different wall temperature, is the wall Sample calculations have been performed, to demonstrate
difference parameter Sw defined previously, eq.(16). As shown the feasibility of the proposed approach. The geometry of a
above, this parameter was shown to have a good correlative two dimension turbine cascade has been used to calculate the
property for simple flows and has been used in the past by flow field and evaluate the effects of different wall
researchers that studied the effect of heat transfer on boundary temperature. A cascade with geometry available in the open
layers. We must add here, that this parameter has also been literature has been chosen. It is the cascade described in [13].
found to be suitable for representing some heat transfer effects in The rans were performed the N-S solver Athena, with the
compressors. McCallum and Pilidis [3], have used this same method described in [14]. The boundary conditions imposed
parameter for incorporating the effect of boundary layer are a given total pressure at the inlet, a given static pressure at
development change due to heat transfer, on compressor the outlet. First, a calculation is done by imposing on the wall
performance. They expressed this change as a shift of the a temperature equal to the one of the free stream. Then two
constant speed lines, which represents the change is mass flow different wall temperatures, lower than the free stream are
capacity of the compressor. imposed and new calculations are performed. A free stream
A comment that has to be mentioned here, is that the way temperature of 1300 K was used, and the lower temperatures
that Sw is defined, has the disadvantage that it gets zero value were 1150 and 1000 K . The results are shown in Figure 8. It is
not for the case of no heat transfer. As we discussed earlier, in observed that the mass flow exhibits the same trend as the one
the adiabatic case the wall temperature is lower and thus Sw will demonstrated for nozzles in the previous sections.
be less than zero. On the other hand, a turbine map is not always Additionally, the figure shows that
obtained on a turbine with adiabatic blades. For example, on- Δm/m=-0.014Sw
engine measurements for estimating the map, produce a map in

7 Copyright ©2006 by ASME


We note that Pilidis and McCallum [4] have mentioned a 0.4

value of –0.07 for the same proportionality factor, for the case of 0.35
compressors 1 (for which a larger sensitivity is expected as
mentioned previously). 0.3

At this point it is useful to compare the trend produced by 0.25

Δm %
the Navier Stokes solver predictions to the trends of the method 0.2
referenced previously. The comparison is shown in Figure 9. It
is remarkable that the trends established by calculation on a two
0.15

dimensional turbine cascade are very close to the ones of the 0.1

analytical predictions, for values of CD that could be considered 0.05


realistic according to Kochendorfer F.D., Nettles [7]. 0
The results shown above point to the opinion that the -0.25 -0.2 -0.15 -0.1 -0.05 0
proposed method is meaningful and can be applied. It should be Sw
remembered however, that the test cases considered are rather Figure 8: Variation of mass flow rate through a cascade, for
simplified. Consequently, the numerical values produced are given inlet total to outlet static pressure ratio and different wall
probably not representative of actual 3-D passages. A following temperatures.
step, in order to examine the behavior for realistic situations,
would be to use a realistic geometry and flow conditions, and

δC D/C D (% )
0.5
perform calculation of 3-D flow field in the passage. The mass
flow calculated will give the possibility to establish relationships 0.45
as the above ones, but with much more confidence to the result 0.4 .96

obtained. 0.35 calculated

0.3
Proposed procedure for incorporating the heat transfer effect on 0.25
boundary layers 0.2
.98
On the basis of these observations, a procedure is 0.15
proposed below for estimating the effect of boundary layer 0.1
changes due to heat transfer on the turbine characteristics. The 0.05
CD,ad =.99
procedure has two aspects: (a) determination of the law of
0
dependence a particular engine, (b) use of this law in engine
-0.25 -0.2 -0.15 -0.1 -0.05 0
model.
Sw=(Tw/T0-1)
The procedure to be followed for estimating the effect of the
change in the flow field can be summarized as follows: Figure 9: Comparison of trends established by calculations
A Navier Stokes code is used to predict the flow in the and viscous flow solutions, to predictions according to the
passage, for walls at some reference temperature. A temperature method of [11].
distribution is imposed on the surface of the blade, lower than
the one used in the first run, and a new prediction is done. The
lower temperature can be defined by subtracting a specific the blade temperature is estimated, from the procedure
number of degrees from the initial values (e.g.100 K). Inlet total presented in the previous section. The turbine performance
pressure and outlet static pressure are kept the same as in step 1. map is assumed to be known at a reference condition, for
Step two is repeated for two further reductions of blade which both gas temperature and wall temperature are known.
surface temperature For any other operating condition, for which the gas
The mass flow rate is evaluated for each run and is plotted temperature is known, the following steps are used to produce
versus the parameter Srw. The relationship between mass flow a correction for boundary layer induced heat transfer effects:
rate and this parameter is thus established. It is expected that the The blade wall temperature is determined by the
percentage change of mass flow will be a linear function of Srw: procedure proposed.
Δq=CSW Srw (18) The parameter Srw ,is calculated from formula (17), and
thus the component of Δq caused by flow field changes from
equation (18).
Implementation in a Model
The deviation of flow capacity is added to any other
When implementing in an engine performance prediction corrections that have been evaluated at the same time (as for
program, the effect of this change can be estimated as soon as example the correction for dilatation)

APPLICATION
1
The quantity they used for correction was δN/N, where N is the rotational
The two methods proposed in the present paper can be
speed. For high speed compressors, the change in mass flow capacity of the
blade passages can be expressed as a shift in the constant speed lines of the incorporated in an engine model, to take into account these
performance map, and thus the representation chosen by the authors. It should effects on engine performance. The inclusion of the two effects
also be noted that for establishing the value of the factor, the authors report to in an engine model is possible by using appropriate routines.
have run compressor map prediction codes. In this respect, the way we propose
in our approach is similar to the way they followed.

8 Copyright ©2006 by ASME


In order to study the impact of the heat transfer effects included, to facilitate checking predictions of variables inside
modeled in the present effort on overall engine performance, test the procedure itself.
cases were performed using the turbofan engine model This last approach seems to the authors to be the most
incorporating or not these effects. realistic for assessing the effectiveness of the methods
Two test cases are presented. The first one represents a proposed here.
typical idle to takeoff acceleration (Figure 10.1) and the second 1.5

one corresponds to a fuel step of 50% increase from idle (Figure


10.3). It can be seen for the rest of the figures (Fig 10.2, Fig 10.4 1.2

correspondingly) that there is a considerable deviation on HPC

FUEL FLOW (kg/s)


surge margin when the effects are included. It must be noticed
0.9

though that these results are based on the assumed data


incorporated in the used engine model.
0.6

It should be noted here, that the two effects are in opposed 0.3
direction. Lower passage wall temperature:
-decreases the passage dimensions and thus mass flow capacity 0
-increases discharge coefficient and thus mass flow capacity 0 1 2 3 4 5

25

Possible ways for validation of the proposed procedures


20
A first comment, validation concerning intermediate results

HPC SURGE MARGIN


of the methods is probably non-meaningful, since the proposed
models are for estimating global effects. For example, the value 15

of blade surface temperature is representative of blade WITHOUT HPT MAP CORRECTION

temperature in a global manner and it would be difficult to 10


WITH HPT MAP CORRECTION-only BL effect

compare it to specific test data. If test data were collected, then WITH HPT MAP CORRECTION-both effects

an appropriate average value would have to be derived for


comparison with the present predictions. It is thus thought, that 5

any validation would have to be done on overall performance


0 1 2 3 4 5
0.5
parameters. Possibilities for validation of the two effects exist,
but they would require rather sophisticated experiments.
Examples of such experiments are mentioned below.
The effect on the development of the boundary layers would
FUEL FLOW (kg/s)

0.4

be more difficult to isolate. An experiment that could be


envisaged for this case, would be to run tests for given free
stream gas temperature and varying blade temperature, via 0.3

controlled amounts and temperature of cooling air. If the


dimensions of the cascade are precisely monitored (for example
by optical measurements), then extraction of the effect of 0.2

dimension changes would leave the effect of heat transfer on 0 1 2 3

boundary layers alone. In any case, if materials with very small


20

expansion coefficients are chosen and temperature differences


are not too large, such effects could be observed directly.
In the opinion of the present authors, such experiments are
HPC SURGE MARGIN

15

not worth to be performed, because they would be very costly WITHOUT HPT MAP CORRECTION

and the information they provide would not be of great value. WITH HPT MAP CORRECTION-only BL effect

The effects are expected to be of very small magnitude and one 10


WITH HPT MAP CORRECTION-both effects

additional factor to be considered is that measurement accuracy


would have to be very high, in order to assess them precisely.
For this reason the approach that seems to be more realistic,
while at the same time it provides a direct assessment of the 5
0 1 2 3

effectiveness of the present method would be to collect data TIME (s)

from operating turbines, in a variety of operating conditions. The Figure 10: Transient variation of various performance
check to be performed would then be if the scattering of data parameters, using different modeling approximations.
reduces, when corrections using the present methods are applied.
A typical set of test data that would be helpful for such a CONCLUSIONS
validation could include turbine performance maps that have Methods for the evaluation of heat transfer effects in
been obtained in a 'cold' test rig, and data from an engine in performance models of a jet engine have been developed and
steady state operation, sufficient to determine the on-engine presented. The methods allow the evaluation of the effects of
turbine performance map. In both cases, apart from the thermal dilatation and altered boundary layer development.
performance data, data for material temperatures should also be Evaluation of both effects relies on the calculation of wall
surface temperatures of a turbine passage. The approach
9 Copyright ©2006 by ASME
developed is based on a one dimensional consideration and of [13] Consigny H., Richards B.E., 1982, Short Duration
semi-empirical relations, taking into account the cooling scheme Measurements of Heat-Transfer Rate to a Gas Turbine Rotor
employed. It can be used for steady sate and transient engine Blade.ASME transactions, Journal of Engineering for Power,
operation. vol 104, July 1982, pp 542-551
Applications to typical turbine geometries were presented, [14] G. Simandirakis and K.D. Papailiou " Heat transfer
in order to produce an indication of the order of magnitude of analysis of turbine cascades through a Navier-Stokes solver",
the changes caused by the two effects. For the particular case of Proceedings of Eurotherm Seminar 46, University of Pisa,
the dilatation effect, a parametric study was carried out, to show Italy, 3-4th July, 1995
how sensitive the calculation is to the different parameters used [15] Stamatis A., Mathioudakis K., Ruiz J., Curnock B., 2001,
for the calculation. At this point it should be mentioned that the ‘Real time engine model implementation for adaptive control
results produced are considered only indicative. In order to & performance monitoring of large civil turbofans’, to be
produce values that would be realistic for current engines, the presented in ASME TURBOEXPO 2001
calculation should be repeated using data from such engines. [16] Horlock J.H.,Watson D.T., Jones T.V., 2000, “Limitations
Implementation of the method was discussed and sample on gas turbine performance imposed by large Turbine cooling
results obtained by incorporating the method in an engine model flows”. ASME paper 2000-GT-635.
were presented. It was shown that heat transfer effects as [17] MacArthur C., 1999, “Advanced aero-Engine Turbine
modeled here, may have a non-negligible effect on engine Technologies and their Application to Industrial Gas Turbine”,
transient performance. ISABE 99-7191.
[18] Mayle R.E., Camarata F., 1975, “Multihole Cooling Film
ACKNOWLEDGEMENTS effectiveness and Heat Transfer” Journal of Heat Transfer, pp.
The authors would like to express their thanks to SNECMA for 534-538 Nov. 1975.
financing the work that led to this paper and for allowing its [19] Suo M., 1985, “Turbine Cooling”, Aerothermodynamics
publication. Thanks are expresses to X. Ruiz, for many useful of Aircraft Engine Components, AIAA EDUCATION
comments and discussions. SERIES, Ed. G. C. OATES ISBN 0-919928-97-3.
[20] Tagashiva T., Vichida T., 1999, “Consideration on Gas
REFERENCES Turbine performance improvement by an advanced cooling
system”, ISABE 99-7178.
[1] MacCallum N.R.L., 1979, Thermal influences in gas turbine
[21] Young J.B., Wilcock R.C.,2001 “Modelling the air-cooled
transients- Effects of changes in compressor characteristics.
Gas Turbine –Part 2-Coolant flows and losses” ASME paper
ASME paper, 79-GT-143
2001-GT-0392.
[2] Pilidis P., MacCallum N.R.L., 1984, A study of the
prediction of tip and seal clearances and their effects in gas
APPENDIX – I
turbine transients. ASME paper 84-GT-245
The data below correspond not to a single blade passage
[3] MacCallum N.R.L., Pilidis P.,1985, The prediction of surge
but represent all the passages circumferentially.
margins during gas turbine transients. ASME paper 85-GT-208. BLADE CASING HUB
[4] Pilidis P., MacCallum N.R.L., 1986, The effect of heat Cross Sectional Area Acs (m2) 0.1375 0.1375 0.1375

transfer on gas turbine transients. ASME paper 86-GT- Constant in Heat tranfer coeff. (eq. 3.71) C (-) 1.26 4.13 4.13
Characteristic length for Reynolds L (m) 0.036 0.72 0.72
[5] MacCallum N.R.L., A.D. Grant, 1977,The effect of boundary Area exposed to heat tranfer (gas) Ag (m2) 1 0.4 0.4
layer changes due to transient heat transfer on the performance Metal Thermal expansion coefficient K (K-1) 1.30E-05 1.30E-05 1.30E-05
Datum Temperature Tmd (K) 1490 1490 1490
of an axial flow compressor. SAE paper 770284. Thermal conductivity of TBC kTBC (J.m-1.K-1) 10 - -
[6] Lakshminarayana B., 1996, Fluid dynamics and heat transfer TBC thickness tTBC (m) 0.001 - -
Area exposed to heat tranfer (coolant) Ac (m2) 2 0 0
of turbomachinery. John Wiley, ISBN 0-471-85546-4 Coolant metal heat transfer coefficient hc (J.m-2.K-1) 6000 0 0
[7] Kochendorfer F.D., Nettles J.C. , 1948, An analytical method Convective cooling efficiency
Film cooling effectiveness
nc
εf
(-)
(-)
0.6
0.4
0
0
0
0
pf estimating turbine performance. NACA Report 930, Lewis Mass flow rate (kg/sec) 100
Flight Propulsion Laboratory. Inlet Mach number - 0.2
[8] Grant A.D., 1973, The Effect of Heat Transfer on Boundary Inlet cooling temperature Toc (K) 890
Gas Temperature Toe (K) 1490
Layer Stability in Axial Flow Compressor. Institution of Mech.
The analysis was based on the steady state model from the
Engrs, Conference Publication 3, 1973, C109/73, pp 252-258.
proposed method and below are the calculation results.
[9] MacCallum N.R.L., 1973, Effect of “Bulk” Heat Transfers in
Gas metal heat transfer coefficient is calculated for
aircraft Gas Turbines on Compressor Surge Margins. Institution
blading casing and hub.
of mech. Engrs, Conference publication 3, 1973, C36-73, pp 94-
hgb=2371.48 (J.m-2.K-1)
100.
hgc=4269.66 (J.m-2.K-1)
[10] Harris H.D., Luxton R.E., 1966 Calculation of Heat
hgh=4269.66 (J.m-2.K-1)
Transfer Coefficients in Cooled Turbine Cascades, The
Biot number for the TBC BTBC=0.237.
Aeronautical quarterly, Aug. 1996, pp 253-268.
B

A recovery factor r=0.836 is calculated assuming a=1 and


[11] Simmons F.S., 1955, analytic Determination of the
PR=0.7. Recovery temperature Tr=1483.37 K
Discharge Coefficients of Flow Nozzles, NACA TN 3447.
Metal temperatures.
[12] Cohen C.B. Reshotko E., 1954, Similar Solutions for the
Tmb=1161.3 K
Compressible Laminar Boundary Layer with Heat Transfer and
Tmc=1483.37 K
Pressure Gradient NACA Report 1293.
Tmh=1483.37 K

10 Copyright ©2006 by ASME

Das könnte Ihnen auch gefallen