Correlative Analysis of Effusion

Cooling Systems
Lorenzo Arcangeli
Gas turbine cooling has steadily acquired major importance whenever engine perfor-
Bruno Facchini mances have to be improved. Among various cooling techniques, film cooling is probably
one of the most diffused systems for protecting metal surfaces against hot gases in turbine
Marco Surace stages and combustor liners. Most recent developments in hole manufacturing allow us to
perform a wide array of microholes, currently referred to as effusion cooling. Though

Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
e-mail: marco.surace@htc.de.unifi.it
some drawbacks of such a concept still need to be solved (manufacturing costs, holes
Lorenzo Tarchi blockage, and then system reliability), its potential is now worth investigating. This paper
presents the validation of a simplified numerical two-dimensional conjugate approach
Department of Energy Engineering “Sergio through a comparison with the experimental results of effectiveness for an effusion plate.
Stecco,” A preliminary test is performed with the steady-state technique, using thermochromic
University of Florence, liquid crystal wide-band formulations. Results are obtained in terms of local distributions
Via Santa Marta 3, of adiabatic effectiveness. Average values are compared with calculations to validate the
50139 Firenze, Italy numerical code. Then, the design of experiment approach is used to perform several
conjugate numerical tests (about 180), so as to derive the behavior of different effusion
plates in terms of overall effectiveness and mass flow rate. Data are analyzed in detail,
and a correlative approach for the overall effectiveness is proposed.
关DOI: 10.1115/1.2749298兴

Introduction ering wall temperature. From this point of view, effusion can be
seen as an approximation of transpiration cooling by porous wall
In modern aircraft engines, the turbine hot section is always
means, with a slight decrease in performances but without the
cooled by air coming from the compressor. Heat transfer engi-
same structural disadvantages.
neers have to design the cooling system in order to obtain proper
Studies on effusion cooling or on multirow hole injection have
metal temperature distributions on vanes, blades, end walls, and
been performed for years. Sasaki et al. 关4兴 presented film cooling
shrouds. To increase engine performance, it is necessary to reduce
adiabatic effectiveness results, using a scanning infrared camera,
cooling mass flow and to improve cooling effectiveness and effi-
for a staggered array of discrete holes. Mayle and Camarata 关5兴
ciency. This implies a very accurate choice among different cool-
proposed an adiabatic effectiveness correlation for multiholes with
ing techniques, suitable to the configuration selected for each lo-
both streamwise and spanwise angles, i.e., compound angles. Due
cation, depending on external constraints and the heat loads of
to the air injection and mixing with hot gases, the flow field near
each part to be cooled. This is one reason why there have been so
many efforts to develop different cooling techniques and schemes. the wall is usually very complex, and the heat transfer coefficient
Among these, the injection of cooling air in the main flow through changes compared to that of a flat plate. This means that not only
the walls is one of the most used techniques. The aim is to pro- the effectiveness has to be evaluated, but so does the heat transfer
duce a thin film of cooling air that isolates the walls from the hot coefficient in order to correctly predict the heat flux. Crawford et
gases. Because of the complex mixing of air/hot gas and the struc- al. 关6兴 experimentally determined the Stanton number for an effu-
tural and manufacturing requirements, many different injection sion cooling geometry.
hole shapes and arrays have been proposed and a great deal of As previously stated, when injection holes are distributed over
research work is still ongoing. the whole surface, the heat transfer between the wall and the air
The most diffused design criterion presents a limited number of passing into the holes is not to be neglected. On the contrary, the
injection rows, each one made by holes with a very complex wall temperature is quite affected by the heat transferred between
geometry, at least at the exit. Examples are holes with a diverging air and wall before injection into the main stream, so the overall
exit section, such as fan shaped holes 共Bell et al. 关1兴兲. The design effectiveness must be evaluated and compared with adiabatic ef-
of a properly diverging exit enables us to control the film spread fectiveness. A detailed analysis was carried out by Andrews et al.
and film coverage, i.e., the film adiabatic effectiveness, up to the 关7,8兴, as a function of hole density and length to diameter ratio.
next row or along the plate 共Sechieder et al. 关2兴兲. Many different experimental techniques have been used to mea-
Only in recent years has the improvement of drilling capability sure wall temperature in such configurations, and high resolution
allowed us to perform a large amount of extremely small cylin- temperature maps have been acquired. Martini et al. 关9兴 used an
drical holes, whose application is commonly referred to as effu- infrared camera, but liquid crystal thermography has also been
sion cooling. Even if this solution does not guarantee the excellent widely used, as in Mayhew et al. 关10兴.
wall protection achievable with film cooling, the most interesting From the design point of view, it is important to develop a
aspect is the significant effect of wall cooling due to the heat simulation tool in order to predict overall effectiveness whenever
removed by the passage of coolant inside the holes 共Gustafsson boundary conditions and geometry parameters change. This could
关3兴兲. In fact, a higher number of small holes, uniformly distributed permit us to properly design the hole array geometry depending
over the whole surface, permits a significant improvement in low- on the location and the hot gas thermal loads. Moreover, such
tools need to be fast and easy to manage because in the prelimi-
nary design stage, several different analyses need to be performed.
Contributed by the International Gas Turbine Institute of ASME for publication in It may also be used to perform design of experiment 共DOE兲 cal-
the JOURNAL OF TURBOMACHINERY. Manuscript received May 18, 2006; final manu-
script received December 4, 2006; published online January 28, 2008. Review con-
culations in order to indicate useful correlations or trends and the
ducted by Je-Chin Han. Paper presented at the ASME Turbo Expo 2006: Land, Sea relative importance of the parameters involved.
and Air 共GT2006兲, Barcelona, Spain, May 8–11, 2006, Paper No. GT2006-90405. The first aim of this paper is to verify and eventually tune a

Journal of Turbomachinery Copyright © 2008 by ASME JANUARY 2008, Vol. 130 / 011016-1
 Fig. 1 Test rig
Fig. 2 Effusion plate

Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
proper simulation tool using experimental data. Once verified, a
DOE approach is used to deepen the understanding of the effusion equipped with thermocouples and pressure probes and able to per-
cooling system behavior, investigating even very small holes. All form heat transfer measurements with two separated flows 共see
results will then be exploited to select the most promising geom- Facchini and Surace 关12兴兲. Three separate plenum chambers are
etries to be experimentally studied in the following years. The present, one for the mainstream flow, one for the coolant, and the
foreseen application is on end wall cooling of highly thermally last at the exit. Between the first plenum chamber and the flat
loaded vanes, as in high pressure turbine stages. plate, there is a smooth converging duct and, afterward, a 250 mm
long entrance region with a fixed cross section.
The measurement of the adiabatic effectiveness over the cooled
Tools Validation
plate consists of a steady-state test realized with assigned flow
Experiments. The experimental survey was performed at the conditions 共see Table 1兲. Starting from the definition,
Dipartimento di Energetica of the University of Florence. The Taw − Tg
final aim of this activity is the measurement of the adiabatic ef- ␩ad = 共1兲
fectiveness over a flat plate of a specific cooling geometry. The Tc − Tg
test rig 共Fig. 1兲 consists of a suction-type circuit that allows com- where, according to the definition of Lakshminarayana 关13兴, the
plete control of two separate flows, the mainstream and the cool- adiabatic wall temperature Taw represents the surface temperature
ant, in terms of both temperature and mass flow rate: a more of a perfectly insulated wall and the adiabatic effectiveness is
detailed description can be found in Facchini et al. 关11兴. The evaluated measuring the mainstream temperature Tg and the cool-
mainstream air passes through a 9.0 kW electronically controlled ant temperature Tc with several thermocouples, while Taw is mea-
electric heater; then, the flow rate is measured by an orifice. The sured by means of a TLC sheet. Actually, this temperature is not
coolant flow temperature is controlled by mixing heated air and exactly the adiabatic one since there is some heat transfer through
cool air. Two rotary vane vacuum pumps, powered by two 7.5 kW the PVC plate; anyway, the maximum wall temperature difference
electric motors, blow air outside and provide the suction for a between the adiabatic one and the measured one was verified with
maximum mass flow rate of 0.10 kg/ s. The two flow rates are set an isothermal test, as always below 1.0 K. Thus, this difference
up by guiding the motor speed between 300 rpm and 1300 rpm has not been taken into account for the calculation of the ␩ad, but
and by throttling the remote controlled motorized valves; the air only in the uncertainty analysis.
temperature exiting the heater is controlled by means of a four The test was run after the steady condition was reached by all
wire Pt100 resistance temperature detector. Three pressure scan- the measured quantities: mass flows, pressures, and temperatures.
ners Scanivalve® DSA 3217 with temperature compensated pi- These measurements are performed all at once while recording
ezoresistive relative pressure sensors allow us to measure the total with the camcorder for 10 s: afterward, an average value is used
or static pressure in 48 different locations with an accuracy of for the calculations.
6.9 Pa. Several T-type thermocouples connected to a data In order to measure a row-by-row discharge coefficient and
acquisition/switch unit 共HP/Agilent®34970A兲 measure the main- blowing ratio, 11 static pressure taps are present on the sidewall.
stream temperature, the coolant temperature and other tempera- These measurements are made taken from a cold test.
tures inside the model. A digital three-charge-coupled-device The uncertainty analysis was performed following the standard
共3CCD兲 camcorder 共Canon®XM-1兲 records a sequence of color ANSI/ASME PTC 19.1 关14兴 based on the Kline and McClintock
bitmap images 共720⫻ 576 pixel, 25 frames/ s兲 from the thermo- method 关15兴. The temperature accuracy is 0.5 K, the differential
chromic liquid crystal 共TLC兲 painted surface on a PC 共IEEE-1394 pressure is 6.9 Pa, the mass flow rate is 2%, and the maximum
standard兲. The illuminating system 共Shott-Fostec®KL1500 LCD兲 error in measuring the effectiveness is 0.05.
uses an optical fiber ring light to ensure a uniform illumination on
the test surface, and it allows us to keep both color temperature Code Features. In this section the numerical tool used to cal-
and light power constant. In order to reduce any undesired Plexi- culate the performances of effusion cooling geometries will be
glas reflections, two polarized lens filters are fitted on both ring presented. The study aims at evaluating the external metal tem-
light and camcorder lenses. perature of an effusion cooled flat plate, so as to determine the
TLCs are the devices used to evaluate surface temperature and,
consequently, the adiabatic effectiveness. For our purpose, we
used the 30’C20W formulation of Hallcrest active from Table 1 Experimental parameters
30° C to 50° C. Crystals are thinned with water and sprayed with
Geometric parameters Flow parameters
an airbrush on the test surface after the application of a black
background paint. TLCs have been calibrated, replicating the D 共m兲 4 ⫻ 10−3 M 1 − M 11 0.11–0.32
same optic conditions of the real test 共more details in Facchini and x 共m兲 30⫻ 10−3 ṁg 共kg/s兲 65⫻ 10−3
Surace 关12兴兲. y 共m兲 30⫻ 10−3 ṁc 共kg/s兲 3 ⫻ 10−3
The tested geometry consists of a flat plate with 11 rows of w 共m兲 108⫻ 10−3 Tg 共K兲 50
staggered holes, equally spaced in the spanwise and streamwise H 共m兲 20⫻ 10−3 Tc 共K兲 25
direction: Fig. 2 shows a schematic drawing of such a plate, and s 共m兲 40⫻ 10−3 ␳g 共kg/ m3兲 0.717
the geometric and flow parameter values are reported in Table 1. ␣ 30° ␳c 共kg/ m3兲 0.791
The plate is made by PVC and fitted inside a Plexiglas test article

011016-2 / Vol. 130, JANUARY 2008 Transactions of the ASME

overall effectiveness and influence on this parameter of design
criteria 共hole spacing, angle, and diameter兲 and external boundary
conditions. Heat conduction inside the metal and then its tempera-
ture distribution are obtained by using a two-dimensional finite
element method 共2D FEM兲 model of the plate with a commercial
code 共ANSYS®9兲 and then by calculating metal conduction as far as
hot and cold side boundary conditions are known. Hot side bound-
ary conditions 共heat transfer coefficient and adiabatic wall tem-
perature兲 are evaluated using a typical fully turbulent flat plate Fig. 3 Adiabatic effectiveness map over the effusion plate
correlation 共specified in the next section兲; the analysis of the cool-
ant fluid network is performed using a one-dimensional 共1D兲
steady code, stator rotor blade cooling 共SRBC兲, developed in-house with the measured expansion ratio of each row, it is possible to

Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
during industrial research projects and successfully used in several evaluate the mass flow rate of each hole. The validity of this
works 共Carcasci et al. 关16兴, Carcasci and Facchini 关17兴, Facchini correlation has also been tested, comparing the total coolant mass
et al. 关11,18兴兲. The cooling system consists of a fluid network flow rate 共measured with the orifice meter兲 with the sum of the
connecting one basic component to another, each one representing flow of each hole,
a particular region of the cooling system. In this particular case, 11
each simple model represents a single film cooling row. The speci-
fication of geometric characteristics of single components, such as
ṁorifice = 兺C
i=1
D关f共␤兲兴ṁis关f共␤,p0c,T0c,i兲兴 共3兲
hole diameter, pitch, length, roughness, inclination angle, and so
on, can be custom selected. A coolant is considered a perfect gas This equation has been verified in various conditions, and the
subjected to wall friction and heat transfer, and a flow field is maximum error found is 2.4%.
solved in subsonic regime using correlations to determine heat As regards the heat transfer measurements, the adiabatic effi-
transfer coefficient 共HTC兲 and friction factor. The user can specify ciency has been measured in the flow conditions reported in Table
boundary conditions for the fluid network in terms of inlet and 1. Looking at Fig. 3, where a 2D map of the adiabatic efficiency is
outlet pressures or mass flow rate, depending on design specifica- shown, the following comments can be made.
tions. Therefore, SRBC, solving the fluid network, provides coolant • The adiabatic effectiveness before the first row is equal to
side thermal boundary conditions for thermal calculation, particu- zero. Theoretically, this result is obvious and is a necessary
larly heat sink effect of coolant holes and film cooling effective- validation of the measuring technique.
ness. The heat sink effect is evaluated by applying to the FEM • Because of the low blowing ratio, film cooling is far from
model, for each row, the heat removal given by the mean heat the penetration regime. Then, the effectiveness always de-
transfer coefficient and the adiabatic wall temperature inside the creases after each hole.
holes. Film cooling effectiveness is provided by SRBC using the • The superposition of the coolant, together with an increasing
correlation in L’Ecuyer and Soechting 关19兴, which was derived in blowing ratio from the first to the last row, results in a gen-
1985 after an extensive search in the literature performed to as- eral increase of the streamwise effectiveness mean value.
semble a comprehensive bibliography of film cooling 共384 publi- • The presence of a small region before the first hole with
cations from 1964 to 1983兲 and to select the primary sources of adiabatic effectiveness slightly greater than zero shows the
data to be used in the development of a reliable flat plate adiabatic effect of nonperfect adiabaticity claimed in the previous sec-
effectiveness correlation. It calculates the adiabatic film cooling tion. This effect is taken into account in the experimental
effectiveness for each row 共once the blowing ratio, velocity ratio, uncertainty evaluation.
and row geometry are known兲, and then the principle proposed by
Sellers 关20兴 is used to superimpose the effects of each row. The In order to perform a validation of the SRBC prediction proce-
achieved film cooling effectiveness is then used together with the dure, it is necessary to extract an adiabatic effectiveness profile
gas temperature and outlet coolant temperature to evaluate the from the experimental results. In particular, the target of this com-
external adiabatic wall temperature. Since the actual component parison is
mass flow rate and pressure drop is wall temperature dependent,
an iterative procedure is required. Convergence is achieved when • to verify L’Ecuyer Soechting correlation 共single row adia-
differences in pressures and mass flow remain unchanged or be- batic effectiveness prediction兲 for effusion cooling geometry
low an error set by the user 共0.01% in this case兲. The SRBC code • to verify Sellers’ superposition criterion for multirow geom-
can also be used without the thermal FEM model to perform fixed etries
metal temperature or adiabatic calculations on fluid networks. In • to verify the hypothesis of an undefined superposition length
this paper, this feature has been used to validate adiabatic film
cooling prediction and superposition criteria via experimental In order to do that, experimental data have been used to properly
data. set up the SRBC model 共discharge coefficient and pressure ratio for
each row兲, and a comparison has been made in terms of spanwise
Experimental Results and Comparison. First, a series of cold averaged adiabatic effectiveness. Results are shown in Fig. 4.
tests was performed in order to measure the CD 共i.e., the mass The agreement between experimental and numerical results
flow rate兲 of each effusion hole, varying the coolant mass flow seems to be quite good, especially for x / D ⬎ 52, where both peak
rate 共mc = 1 – 61 g / s兲 and keeping the mainstream flow at 65 g / s values and decrease after each row are the same. For x / D ⬍ 52,
constant. Due to the relevant pressure losses in the mainstream the L’Ecuyer Soechting correlation slightly underpredicts the ef-
channel, the expansion ratio of each single row grows signifi- fectiveness between two adjacent rows but correctly predicts the
cantly from the first 共␤ = 1.0027兲 to the last 共␤ = 1.0064兲. So, this peak value and the general trend.
variation results in an increasing coolant flow in the x direction.
Experimental data have been correlated in order to define a CD Design of Experiment
typical of the tested geometry,
General Assumptions. The comparison presented in the previ-
CD = 0.75共1 − e−共␤−1兲/0.0072兲 共2兲 ous section was necessary to ensure the validity of the simplified
1D analysis for the adiabatic effectiveness evaluation by means of
which, in test conditions, was found to vary between 0.23 and a correlative approach. As previously described, conjugate simu-
0.44 from the first to the last row. Using this correlation together lations take advantage of the solution of 2D conduction equations

Journal of Turbomachinery JANUARY 2008, Vol. 130 / 011016-3

 Table 3 Design parameters

Variable inputs

Mais 0.2–0.95
D 共m兲 共0.07– 0.4兲 ⫻ 10−3
␣ 30–90°
x/D 10–54
y/D 4–18

Derived parameters

␤ 1.05–1.82

共 兲

Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
␳␯
Rex = x 共1.0– 6.0兲 ⫻ 106
␮ g

Rec = 共 兲
␳␯
␮ c
D 共1.0– 6.0兲 ⫻ 104

ṁ
G= 共kg/ m2 s兲 1.0–26.0
tw
Fig. 4 Spanwise average adiabatic effectiveness comparison

through the metal. Validation was not retained worthwhile for the
FEM tool, as it is well known that a 2D conduction in a thin metal effects along the hole兲, coolant pressure ratio, and blowing ratio
plate can be easily solved with high accuracy. are set once Mais is known. It implies that where the Mach num-
Effusion cooling configurations were applied on a flat plate, ber is high, the hot gas mass velocity is high, the static pressure is
representative of a HP stage platform, which conventionally pre- low, and the coolant pressure ratio and, with it, the coolant mass
sents few film cooling rows of bigger diameter holes, while the velocity are high. It derives that it is not possible to infer the
proposed solution features a large number of uniformly distributed blowing ratio trend by a qualitative approach, and the quantitative
microholes 共Fig. 5兲. In fact, one of the aims of the broad investi- analysis will show that the blowing ratio variation range is
gation carried out is to show the possible increase in performance limited.
by adopting effusion instead of film cooling. Hole discharge coefficients CD are inferred from data presented
Table 2 shows the main input parameters fixed for all simula- by Gritsch et al. 关21,22兴, thus taking into account the dependence
tions. All other inputs were selected in a wide range, so as to on the main flow Mach number, the coolant pressure ratio, and the
allow the comprehension of general validity aspects. Starting with injection angle. These data were not taken from the experimental
them, several significant design parameters can be derived, as re- results presented in the previous section as the main parameters
ported in Table 3. range investigated afterward is much wider, requiring reference to
This passage requires a detailed explanation of the adopted as- more detailed cases.
sumptions and derived steps. Then, the geometric features have to be decided in terms of
Referring to a subsonic HP stator vane, a hot gas Mach number hole diameter and angle, streamwise and spanwise pitches, in or-
usually varies from very low values at the beginning of the stage der to achieve the desired result. The dimensionless pitches x / D
共Mais ⬇ 0.2兲, up to approaching sonic conditions 共Mais ⬇ 0.95兲. and y / D assume particular importance: together with the CD, they
Assuming an isentropic flow for the hot gas and fixing the total determine the coolant mass flow per unit platform area G 共and
temperature and pressure for it and the coolant, the pressure in the then the overall coolant mass flow兲 independently of the diameter
main channel, coolant mass velocity 共i.e., ␳vc, aside from heating size. In detail, it can be easily demonstrated that
ṁ ␲ 共␳v兲cCD
G= = 共4兲
tw 4 x/Dy/D
Each simulation was performed at a fixed Mais: On the one hand,
such simplification causes some distance from the real flow con-
ditions, as the actual gas accelerates across the stator. On the other
hand, it allows us to investigate the performances of different
effusion geometries independently of the vane profile and then the
pressure distribution along the channel.
The hot side heat transfer coefficient was calculated as a func-
tion of position and gas Reynolds number Rex, with the flat plate
correlation from Lakshminarayana 关13兴, and is reported below
Nux = 0.0288 Re0.8 1/3
x Prg 共5兲
derived for an uncooled plate. It was found that hg varied from
3000 W / m2 K to 7000 W / m2 K with Mais from 0.2 to 0.95, re-
spectively. For each calculation 共i.e., imposed Mais兲, the heat
Fig. 5 Actual and proposed configurations transfer coefficient decreased by about 10% along the plate.
The experimental comparison effected in the previous section
Table 2 Fixed flow and geometric conditions was helpful in indicating that the complete effectiveness superpo-
sition guarantees a correct simulation of the 1D behavior. This
T0g / T0c p0g / p0c t 共mm兲 s 共mm兲 assumption was maintained for all performed runs.
2.24 1.02 0.05 1.5⫻ 10−3 Numerical Procedure. To obtain the desired results, an itera-
tive calculation procedure was followed for each geometry and

011016-4 / Vol. 130, JANUARY 2008 Transactions of the ASME

 Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
Fig. 7 Overall effectiveness of performed tests as a function
of coolant mass flow rate, with reference to the length to diam-
eter ratio

Tw − Tg
␩ov = 共6兲
Fig. 6 Iterative procedure Tcin − Tg
where Tw indicates the average wall temperature on the hot sur-
face.
flow condition. Once the 1D fluid network is generated with SRBC Figure 7 shows the overall effectiveness as a function of cool-
and the plate geometry is created and meshed in ANSYS®, the ant mass flow rate for all the performed tests. Data are dispersed
following calculation steps are followed, with reference to Fig. 6 around the dotted line, which represents a parabolic fit. An in-
crease of both effectiveness value and its dispersion as the mass
• The fluid network is initialized with a first attempt metal flow rate increases is evident. Looking through the plotted data,
temperature distribution 共constant兲, and coolant boundary the dispersion can be attributed to the length to diameter ratio,
conditions are imposed in terms of pressure, temperature, which acts as a fundamental parameter to distinguish among per-
and geometry. formances of different configurations. As a general trend, the
• SRBC provides a distribution of adiabatic effectiveness 共and highest effectiveness values are associated with the highest L / D
then adiabatic wall temperature兲; it depends on the assumed ratios, which depend on both hole diameter D and angle ␣, as the
metal temperature as the coolant exit conditions are affected plate thickness s is fixed.
by this value and consequently affect ␩ad. The same considerations cannot be expressed, examining Fig.
• The adiabatic effectiveness distribution is given as a bound- 8, which shows the adiabatic effectiveness as a function of coolant
ary condition to the FEM calculation, together with hot side mass flow rate, with reference to the length to diameter ratio. The
data; the solution, attained in terms of 2D wall temperature significant dispersion of data prevents us from deriving a direct
distribution, is used to reinitialize the SRBC run. relation between the plotted quantities or the influence of the hole
• The convergence is achieved when the maximum error on length. Indeed, it underlines the importance of the heat transfer
the hot surface metal temperature is below the desired value inside the hole and points out that a configuration showing a better
of 0.1 K. adiabatic behavior can have a worse overall performance, thus
indicating that ␩ad cannot be used as an independent assessment
Tests. To ensure a proper evaluation of the influence of the
parameter.
different input parameters on the overall effectiveness, reducing
As previously stated, the blowing ratio M does not vary with an
CPU time as much as possible, a DOE approach has been used to
easily predictable trend; hence, it is not an input but rather an
determine the test matrix for calculations. For each input variable
output of a complex combination of quantities. In particular, while
reported in Table 3, three values have been chosen, particularly
the lower, the average, and the higher value of the range. Since for
each calculation the CPU time is about 25 min, a full factorial
DOE sequence has not been considered suitable, and a reduced
factorial approach has been adopted. The set of runs has been
reduced from 243 for a three level full factorial DOE to 180 of the
reduced factorial approach, reducing the calculation time by about
2 / 3, without affecting the reliability of results 共Montgomery
关23兴兲. MODE-FRONTIER® code capabilities have been used to deter-
mine the DOE sequence to develop an automatic procedure able
to prepare input files for each design and to post process the
results. It was then possible to obtain, for each set of input vari-
ables, the coolant mass flow rate, the temperature distribution
along the plate, and, thus, the overall effectiveness on the external
surface.

Results and Discussion

Fig. 8 Adiabatic effectiveness of performed tests as a function
Achieved results will now be presented in terms of adiabatic of coolant mass flow rate, with reference to the length to diam-
共Eq. 共1兲兲 and overall effectiveness, which is defined as follows: eter ratio

Journal of Turbomachinery JANUARY 2008, Vol. 130 / 011016-5

 Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
Fig. 9 Blowing ratio variation for all performed runs Fig. 11 Correlation of obtained results

deserves a detailed interpretation, as it takes into account several

the hot gas mass velocity ␳vg is uniquely identified by the selected important effects: on the one hand, the pressure ratio, the coolant
Mais and static pressure, the coolant mass velocity ␳vc depends heating before injection, and the dimensionless streamwise and
both on that pressure and on the heating effect across the plate. It spanwise pitches, by means of G, as indicated by Eq. 共4兲; on the
is, in turn, affected by the wetted surface of the holes and by
diameters, pitches, and angles. Looking at Fig. 9, we see how such other hand, the heat load on the hot side with h̄g. Thus, a lower ␺
geometric features influence the blowing ratio, which can change with the same ␩ov 共i.e., a lower metal temperature兲 implies a lower
from 0.6 to 1.3 for the same coolant mass flow rate. coolant mass flow rate or a higher thermal load, thus better
The best configurations, taken from the set previously shown, performances.
have been tested again, maintaining all dimensionless quantity Figure 11 shows the correlation of present data in terms of the
and flow conditions and changing only the hole diameter. As to be overall effectiveness as a function of ␺ and L / D,
expected from exposed comments, best configurations present the ␩ov = a2共R␺兲2 + a1R␺ + a0 共8兲
lowest hole size, i.e., D = 0.07 mm. New simulations were per-
formed at D = 0.4 mm, thus adopting a value typical of with
conventional—though small—film cooling holes. The results of R = b2共L/D兲2 + b1共L/D兲 + b0 共9兲
this comparison are shown in Fig. 10, which highlights the good
potential of microhole solutions, which appear able to guarantee Equation 共8兲 fits all the test results with a correlation coefficient
an improvement in cooling performances, other conditions being of 90%, determined through a regression analysis 共Table 4兲. It was
equal. A similar DOE approach including large holes was not actually run with many cross products of terms, but only the sub-
carried out as was thought to be useless. As a consequence, ob- set reported here was found to be significant. This correlation
tained results probably do not represent the best configurations for format now identifies the important parameter relationships that
D = 0.4 mm, but previous considerations keep their validity. affect effusion array performances. No extrapolation of the pre-
sented equation is feasible; rather, its use may be significant only
Correlation. A statistical approach was followed to correlate within the selected range of input parameters 共see Table 3兲. Sum-
all the results from performed tests. It was found that the main marizing, it should be R␺ ⬍ 3.0, 30° ⬍ ␣ ⬍ 90° and 10⬍ L / D
output, ␩ov, was essentially dependent on the length to diameter ⬍ 43.
ratio and the dimensionless quantity ␺, defined as It is interesting to show that the hole angle, despite being varied
in a wide range, was not found to act directly on overall perfor-
Gc p
␺= 共7兲 mances, but only to affect the hole length, which is in turn a
h̄g correlation input.
Despite the unavoidable uncertainty of the expression, it is very
where, in particular, h̄g is calculated, averaging the local heat interesting to underline the novelty of this correlation, which helps
transfer coefficients on the hot side of the plate. This quantity to increase the comprehension of effusion cooling related phe-
nomena and whose alternative forms have not been published in
the available literature to the best of the authors’ knowledge. This
tool can be extremely helpful in the first design stage of cooling
systems for turbine blade and vane hottest regions.

Conclusions
The present study has investigated the cooling performances for
a series of effusion configurations of microholes applied to a flat

Table 4 Correlation coefficients

a0 a1 a2

9.134⫻ 10−2 4.060⫻ 10−1 −6.381⫻ 10−2

b0 b1 b2

Fig. 10 Comparison between the best configurations of micro- −9.0⫻ 10−2 5.8⫻ 10−2 −5.6⫻ 10−4
holes and conventional holes

011016-6 / Vol. 130, JANUARY 2008 Transactions of the ASME

plate. The analysis has initially focused on the validation of the is ⫽ isentropic
available numerical tools, then a detailed DOE approach followed ov ⫽ overall
to investigate the effects of the main design parameters. A big x ⫽ abscissa along the plate
influence on global performance has been shown by the hole w ⫽ wall
length to diameter ratio, as it was found that the overall effective-
ness always increases with L / D. Moreover, for a fixed geometry
共x / D , y / D , ␣兲, a smaller diameter always implies better perfor- References
mance in terms of ␩ov. Finally, all reported data have been ana- 关1兴 Bell, C. M., Hamakawa, H., and Ligrani, P. M., 2000, “Film Cooling From
lyzed, and the great number of performed tests has allowed the Shaped Holes,” ASME J. Heat Transfer, 122, pp. 224–232.
关2兴 Schnieder, M., Parneix, S., and von Wolfersdorf, J., 2003, “Effect of Shower-
formulation of a simple correlation, covering the entire range of head Injection on Superposition of Multirow Pressure Side Film Cooling With
parameters tested. Fan Shaped Holes,” ASME Paper No. GT2003-38693.
关3兴 Gustafsson, K. M. B., 2001. “Experimental Studies of Effusion Cooling,”

Downloaded from http://asmedigitalcollection.asme.org/turbomachinery/article-pdf/130/1/011016/6917911/011016_1.pdf by Indian Institute of Technology Bombay user on 23 June 2024
Acknowledgment Ph.D. thesis, Chalmers University of Technology; 共www.tfd.chalmers.se/~lada/
postscript_files/bernhard_phd.pdf兲.
The present work was supported by the European Commission 关4兴 Sasaki, M., Takahara, K., Kumagai, T., and Hamano, M., 1979, “Film Cooling
as part of the FP6 STREP AITEB-2 research program, which is Effectiveness for Injection From Multirow Holes,” ASME J. Eng. Power, 101,
pp. 101–108.
gratefully acknowledged together with consortium partners. 关5兴 Mayle, R. E., and Camarata, F. J., 1975, “Multihole Cooling Film Effective-
ness and Heat Transfer,” ASME J. Heat Transfer, 97, pp. 534–538.
Nomenclature 关6兴 Crawford, M. E., Kays, W. M., and Moffat, R. J., 1980, “Full-Coverage Film
Cooling, Part 1,” ASME J. Eng. Power, 102, pp. 1006–1012.
cp ⫽ specific heat 共J kg−1 K−1兲 关7兴 Andrews, G. E., Asere, A. A., Mkpadi, M. C., and Tirmahi, A., 1986, “Tran-
CD ⫽ effusion hole discharge coefficient spiration Cooling: Contribution of Film Cooling to the Overall Cooling Effec-
D ⫽ effusion hole diameter 共m兲 tiveness,” Int. J. Turbo Jet Engines, 3, pp. 245–256.
关8兴 Andrews, G. E., Asere, A. A., Gupta, M. L., and Mkpadi, M. C., 1990, “Ef-
G ⫽ coolant mass flow per unit platform area fusion Cooling: The Influence of Number of Hole,” Power Eng. J., 204, pp.
共kg m−2 s−1兲 175–182.
关9兴 Martini, M., Schulz, A., and Witting, S., 1995, “Full Coverage Film Cooling
h̄g ⫽ average heat transfer coefficient 共W m−2 K−1兲 Investigations: Adiabatic Wall Temperature and Flow Visualization,” Paper
L ⫽ hole length 共m兲 No. 95-WA/HT-4.
ṁ ⫽ coolant mass flow rate 共kg/s兲 关10兴 Mayhew, J. E., Baughn, J. W., and Byerley, A. R., 2002, “The Effect of
Freestream Turbulence on Film Cooling Adiabatic Effectiveness,” ASME Pa-
M ⫽ blowing ratio 共␳v兲c / 共␳v兲g per No. GT-2002-30172.
Ma ⫽ hot gas Mach number 关11兴 Facchini, B., Surace, M., and Tarchi, L., 2005, “Impingement Cooling for
Nu ⫽ Nusselt number Modern Combustors: Experimental Analysis and Preliminary Design,” ASME
Paper No. GT2005-68361.
Pr ⫽ Prandtl number 关12兴 Facchini, B., and Surace, M., 2006, “Impingement Cooling for Modern Com-
R ⫽ correlation parameter bustors: Experimental Analysis of Heat Transfer and Effectiveness,” Exp. Flu-
Re ⫽ Reynolds number ids, 40, pp. 601–611.
p ⫽ pressure 共Pa兲 关13兴 Lakshminarayana, B., 1996, Fluid Dynamics and Heat Transfer of Turboma-
chinery, Wiley, New York.
s ⫽ plate thickness 共m兲 关14兴 ASME, 1985, “Measurement Uncertainty,” Instrument and Apparatus, Perfor-
t ⫽ plate length 共m兲 mance Test Code, Vol. ANSI/ASME PTC 19.1-1985, ASME, New York.
T ⫽ temperature 共°C兲 关15兴 Kline, S. J., and McClintock, F. A., 1953, “Describing Uncertainties in Single
Sample Experiments,” Mech. Eng. 共Am. Soc. Mech. Eng.兲, 75, pp. 3–8.
v ⫽ velocity 共m/s兲 关16兴 Carcasci, C., Facchini, B., and Ferrara, G., 1995, “A Rotor Blade Cooling
x ⫽ streamwise pitch 共m兲 Design Method for Heavy Duty Gas Turbine Applications,” Paper No. 95-
y ⫽ spanwise pitch 共m兲 CTP-90.
w ⫽ plate width 共m兲 关17兴 Carcasci, C., and Facchini, B., 1996, “A Numerical Procedure to Design In-
ternal Cooling of Gas Turbine Stator Blades,” Rev. Gen. Therm., 35, pp.
Greek 257–268.
关18兴 Facchini, B., Surace, M., and Zecchi, S., 2003, “A New Concept of Impinge-
␣ ⫽ effusion hole angle 共°兲 ment Cooling for Gas Turbine Hot Parts and its Influence on Plant Perfor-
␤ ⫽ coolant pressure ratio 共p0 / p兲c mance,” ASME Paper No. GT-2003-38166.
␩ ⫽ effectiveness 关19兴 L’Ecuyer, M. R., and Soechting, F. O., 1985, “A Model for Correlating Flat
Plate Film Cooling Effectiveness for Rows of Round Holes,” AGARD Heat
␮ ⫽ viscosity 共kg m−1 s−1兲 Transfer and Cooling in Gas Turbines, Paper No. SEE N86-29823 21-07,
␳ ⫽ density 共kg/ m3兲 provided by the Smithsonian/NASA Astrophysics Data System 共http://
adsabs.harvard.edu/cgi-bin/nph-bib_query?bibcode
Subscripts ⫽1985htcg.agarQ....L&db_key⫽PHY兲.
0 ⫽ total quantity 关20兴 Sellers, J. P., 1963, “Gaseous Film Cooling With Multiple Injection Stations,”
AIAA J., 1, pp. 2154–2156.
1,…,11 ⫽ row number 关21兴 Gritsch, M., Schulz, A., and Wittig, S., 1998, “Method for Correlating Dis-
ad ⫽ adiabatic charge Coefficients of Film Cooling Holes,” AIAA J., 36, pp. 976–980.
aw ⫽ adiabatic wall 关22兴 Gritsch, M., Schulz, A., and Wittig, S., 1998, “Discharge Coefficient Measure-
ments of Film Cooling Holes With Expanded Exits,” ASME J. Turbomach.,
c ⫽ coolant 120, pp. 557–563.
g ⫽ hot gas 关23兴 Montgomery, D. C., 1991, Design and Analysis of Experiments, 4th ed., Wiley,
in ⫽ inlet New York.

Journal of Turbomachinery JANUARY 2008, Vol. 130 / 011016-7