Proceedings of NCFMFP2006

33rd National and 3rd International Conference on Fluid Mechanics and Fluid Power
December 7-9, 2006, IIT Bombay, India

Paper No. NCFMFP2006-1301

NUMERICAL STUDY OF FLOW DISTRIBUTION INSIDE COOLED

NON-ROTATING GAS TURBINE BLADE WITHOUT PIN FINS, RIBS
AND FILM COOLING HOLES

Sachin L. Borse
Ph.D. student ,
Dept. of Mechanical Engineering,
Indian Institute of Technology,
Bombay, Mumbai-76
sachinlb@yahoo.co.uk

ABSTRACT edge passage is always provided with film cooling holes. In

A typical internally cooled gas turbine blade consists of
three passages viz, leading edge passage, middle serpentine
passage and trailing edge passage. Accurate estimation of the
heat transfer coefficient in each passage is required to
determine turbine blade temperature distribution. One of the
factors on which heat transfer coefficient depends is the mass
flow through the given passage. Numerical study is carried out
on simplified cold stationary model of a turbine blade. Four
cases with different tip hole diameters are studied. This work
focuses attention on capability of CFD with model considered
to predict flow distribution in model of turbine blade and, also
study to flow pattern which otherwise difficult to obtain
experimentally. Here results are obtained with CFD code
FLUENT. Turbulence is modeled with standard k-ε model and
standard wall function. Results are compared with experimental
values. It shows good agreement with mass flow distribution
but pressure prediction is poor.
INTRODUCTION
Modern gas turbine blades are cooled internally and externally
to ensure longer life of the turbine blade. Figure 1 shows a
typical cooled turbine blade. Cross section of the blade consists
of three distinct passages viz., a) Leading edge coolant passage,
having holes for impingement cooling, b) middle serpentine
passage with turbulence promoters like ribs and c) Trailing
edge passage with pin fins and lateral ejection holes. Leading
some cooled blades, serpentine and trailing edge passages also
have film cooling holes. Film cooling holes are provided to
protect the blade thermally by extracting heat as coolant passes
through holes and by forming a protective layer of relatively
cold air on external surface. Optimum design of the cooling
system requires knowledge of external heat load to the blade
and heat transfer coefficient distribution in internal passages.
With this information, temperature distribution in the blade can
be estimated. The passage heat transfer coefficient, in turn,
depends on mass flowing through the passage. Experimental
studies on flow distribution are reported by Borse and Date
(2006).To the best of author’s knowledge, no such numerical
study on flow distribution without cooling holes is reported in
the open literature.
Design of cooled gas turbine the blade changes from
one manufacturer to the other. Information on exact dimensions
of a modern cooled turbine is scanty. The available information
from the literature is shown in Table 1. Han et al. (2000) give
dimensions of blade tip holes. This information clearly
indicates that the tip hole at the end of the serpentine passage is
bigger than other tip holes. The tip holes are meant for cooling
the tip area of the blade. Compared to the leading- and the
trailing edge passages, the serpentine passage with two 180o
bends is most tortuous and therefore it is necessary to have
relative bigger tip-hole for this passage to achieve reasonable
flow rate through this passage.

Copyright © 2006 by NSFMFP

1
 Table 1 Typical aircraft turbine blade cooling parameters for
serpentine passage [Hwang et al (2001)]

Parameters Values

Coolant passage hydraulic diameter, mm ~4

Coolant passage length, mm 48-70

Rib height to diameter of passage ratio, e/dh 0.01-0.15

Pitch to height ratio of ribs, P/e 7-13

Mean coolant velocity (m/s) 34.5-69

Flow Reynolds number, Re 20,000-

40,000

.
m TT Mass flow rate- trailing edge passage tip holes, kg/s
.
m TL Mass flow rate-trailing edge passage lateral ejection
Fig. 1 Schematic of a Modern Cooled Gas Turbine Blade holes, kg/s
[Han et al (2000)] .
m tot Total mass flow, kg/s
Several turbulence numerical models are used to predict .
fluid flow and heat transfer in gas turbine blade. Han et al % m tot Percentage of total inlet mass flow going to given
(2000) describes numerical modeling related to turbine blade stream
cooling. In the present study, standard k-ε model with standard .
wall function is used. %∆ m Percentage difference in prediction of mass flow rate
%∆p Percentage difference in predicted and measured
NOMENCLATURE supply pressure taking measured pressure as reference.

d Hydraulic diameter of passage, mm

dA Diameter of tip hole, mm
dB Diameter of serpentine passage end hole, mm EXPERIMENTAL SET-UP
dh Hydraulic diameter of passage, mm Figure 2 shows details of the experimental set-up. The test
dimp Diameter of leading edge impingement hole, mm section is a scaled up (approximate 5 times) simplified cold
e Height of rib, mm model of the turbine blade. Air to the test section is supplied
k Turbulent kinetic energy, m2/s2 from 1 hp blower through a flow control valve, a diffuser and a
p Pressure, Pa settling chamber. The test section, whose internal cross section
pin Inlet pressure, Pa area is much smaller than that of the settling chamber is fed
pexit Pressure in header, Pa directly from the settling chamber to simulate sudden
P Pitch of ribs, mm contraction condition prevailing in the plenum of the gas
rd dB/dh, dimensionless turbine blade. Unlike the real blade, the test section is
Re Reynolds number based on hydraulic diameter, symmetric. It is made in split form, from two Perspex slabs (see
dimensionless Figure 3). On these slabs, passages are milled in such way that
. half portion of the passage lies in lower slab and the other half
m Mass flow rate, kg/s in the upper slab. These two slabs are bolted together by means
.
m LL Mass flow rate-leading leading most passage, kg/s of allen screws. Neoprene rubber gasket (1mm thick) is used to
. prevent leakage from all joining surfaces. The chord and the
m LS Mass flow rate-leading supply passage, kg/s span of the test section are 283 mm and 475 mm respectively.
. Square cross section (25mm X 25 mm) is used for the
m SE Mass flow rate –serpentine passage end hole, kg/s serpentine passage and for the leading edge passage.
.
Mass flow rate-serpentine passage tip holes, kg/s Trapezoidal cross section is used for trailing edge passage. No
m ST
ribs or pin fins are used in passages. No film cooling holes are

Copyright © 2006 by NSFMFP

2
provided in current setup. Dimensional details of test section Table 2 Dimensional details of test section
are shown in Table 2.
The tip of the test section is having detachable tip Parameter Value Passage
plates. Four such tip plates having different tip hole diameters Hydraulic diameter of leading 25
are selected (see Table 3 and Figure 3). Diameter of tip holes edge passage, mm, [SQUARE]
dA is same for all tip holes except in the serpentine passage Diameter of impingement hole, 5 Leading
where the end hole dB is bigger than dA, as in a real gas turbine mm edge
blade. Length of leading edge passage, 287.5 passage
Flows from different holes are collected in six mm
different headers. The positions of headers are shown by capital
Spacing of impingement holes, mm 20
letters. Following six flows are collected,
Hydraulic diameter of passage, 25
LL- tip holes of leading most passage of the leading mm, [SQUARE]
edge passage receiving impinging jet from impingent Diameter of film cooling hole, mm 2 Serpentine
holes. Length of single pass for serpentine 300 passage
LS-tip holes of supply passage of the leading edge passage, mm
passage. Thickness of divider wall of 12.5
SE-serpentine passage end hole. passage,
ST-serpentine passage tip holes. mm
TT-trailing edge passage tip holes. Hydraulic diameter of trailing edge 27.76
TL- trailing edge passage lateral ejection holes. passage, mm, [TRAPEZOIDAL]
Diameter of trailing edge ejection 3.175
holes Trailing
, mm edge
Length to diameter ratio for trailing 6.6 passage
edge ejection holes, mm
Number of trailing edge lateral 37
ejection
holes
Spacing of lateral ejection holes, 7.5
mm

Table 3 Dimensional details of tip plates

Fig. 2 Schematic of the set-up. Tip Diameter Diameter of rd dA/dB
plate of tip serpentine
No. hole, dA, passage end
mm hole, dB, mm
1 2.5 10.0 0.04 0.25
2 2.5 16.7 0.76 0.15
3 4.0 12.5 0.50 0.32
4 4.0 14.0 0.56 0.29

Table 4 Number of cells in grid

Plate no Number of elements used
1 12,50,019
2 12,74,368
3 9,93,370
. 4 9,97,123
Fig. 3 Sectional view of the test section.
computational domain. Figure 5 through figure 7 computational
grid.
NUMERICAL APPROACH AND PROCEDURE Computational Grid
The computational fluid dynamics code selected to solve this Computational domain as shown in Figure 4 is consisting
problem is FLUENT version 6.1. Pre-processor of FLUENT of various three passages with tip holes, impingement holes and
called GAMBIT is used for development of the computational lateral ejection holes (not labelled). Uniform grid, except in
geometry and subsequently 3D grid. Figure 4 shows holes is made up of tetrahedral elements. Table 4 shows
number of elements for four different cases with different tip

Copyright © 2006 by NSFMFP

3
plates. Wall functions are used to avoid fine grid at wall. Grid
is selected in such way that average y+ for wall bounded cell is
greater than 5 (beyond laminar sublayer), since wall function
cease to be valid in the viscous layer. Here values of average y+
existing for four cases are ranging from 11.6 to 15.7. Figure 5
to figure 7 shows grid used for current computation.

Turbulence Model
The standard k-ε model is applied. Scheme of discretization
for governing equation is finite volume approach. With this
approach, the Reynolds-averaged Navier-Stokes equations are
solved numerically along with transport equation for the
turbulent kinetic energy and dissipation rate. For near wall
treatment standard wall functions proposed by Launder and
Spalding (1974) are used. Flow here is assumed to be steady Fig. 5 Grid-tip portion
and incompressible. Properties of the fluid, air is assumed to be
constant. Since in this case air is behaving as incompressible,
continuity is satisfied using a semi-implicit method for pressure
linked equations, called as SIMPLE procedure is employed.
Equations are solved implicitly assuming steady flow.
Convective scheme used is first order upwind scheme.

Solution Convergence
Each iteration is done implicitly. The convergence of the
computational solution is determined based on scaled residuals
for governing equations, and many predicted variables. The
total residual for a given variable is based on the imbalance in
an equation for conservation of that variable summed over all
computational cells. The convergence criterion set for all
variables was 10-4. The solution is considered to be converged
when all of the scaled residuals are less than or equal to these Fig. 6 Grid-hub portion
default settings.

Computation procedure
Initially problem is solved considering laminar flow for few
iterations (70) with uniform exit pressure (zero gauge). Then
different exit pressures as existing in experimental set up are
applied. Now flow is treated as turbulent flow and inlet values
of k and ε are specified considering 5% turbulence intensity and
turbulent intensity ratio of 30.

Fig 7 Closer view of grid

RESULTS AND DISCUSSION

Table 5 to Table 9 shows comparison for given case between

numerical and experimental values. Error in prediction of mass
.
flow %∆ m and pressure at inlet (%∆p) is evaluated based on
measured value. Error in prediction of major mass flow rates
found to be of range 1 % to 14%. In majority of case it is
below 6 %. Experimental data is having maximum uncertainty
of ± 4%, uncertainty is larger in smaller mass flow rate
measurements than large mass flow rate measurements. For
Fig. 4 Computational domain smaller mass flow rates error in prediction is 8% to 24% this
may be because of larger uncertainty associated Thus

Copyright © 2006 by NSFMFP

4
prediction of mass flow rate is reasonably good with this
standard k-ε model. Prediction of pressure at inlet is poor
having error of range 33.5 % to 48 %. It predicts higher
pressure at inlet. This may due to fact that standard k-ε model
is high Reynolds number model and Reynolds number existing
in computational domain are relatively lower Reynolds number
(6500-9000)at inlet and in domain. Also selected y+ is having
smaller value than desired (~30). Secondary flows in bend will
affect prediction.

.
Table 5 Comparison between numerical and experimental results for plate 1 for m tot = 0.017 kg/s
Exit . . . • pexit, Pa
m, kg/s % m tot % m tot % ∆m
location
(Experimental) (Experimental) (FLUENT)
LL 0.000468 2.75 2.36 14.21 -759.3
LS 0.000456 2.68 2.34 12.72 -692.8
SE 0.00398 23.39 26.76 -14.4 -588.5
ST 0.000665 3.91 3.38 13.53 -684.7
TL 0.01066 62.72 61.7 1.62 0
TT 0.00078 4.58 3.45 24.65 -686.7
.
Table 6 Comparison between numerical and experimental results for plate 2 for m tot = 0.0215 kg/s
Exit . . . • pexit, Pa
m, kg/s % m tot % m tot % ∆m
location
(FLUENT) (Experimental) (FLUENT)

LL 0.000398 2.2 1.85 17.2 -724

LS 0.000395 2.2 1.83 16.1 -673
SE 0.0091 41.4 42.33 -2.25 0
ST 0.000532 2.74 2.47 9.83 -673
TL 0.01049 47.9 48.79 -1.83 -12
TT 0.00058 3.3 2.69 17.19 -664
.
Table 7 Comparison between numerical and experimental results for plate 3 for m tot = 0.0214 kg/s
Exit location . . . • pexit, Pa
m, kg/s % m tot % m tot % ∆m
(Experimental) (Experimental) (FLUENT)

LL 0.0013 6.07 5.35 12 -262.9

LS 0.00138 6.45 5.23 18.8 -133.4
SE 0.0061 28.5 28.97 -1.64 -214.8
ST 0.00148 6.92 7.48 -8.1 -104
TL 0.0092 42.99 45.33 -5.4 0
TT 0.00185 8.64 7.48 13.5 -70.7
.
Table 8 Comparison between numerical and experimental results for plate 4 for m tot = 0.0234
Exit location . . . • pexit, Pa
m, kg/s % m tot % m tot % ∆m
(Experimental) (Experimental) (FLUENT)

LL 0.00125 5.34 5.11 4.4 -458

LS 0.0013 5.6 4.92 11.46 -257
SE 0.00735 31.4 34.18 -8.8 -223
ST 0.0014 6.03 7.18 -19.15 -221
TL 0.0103 44 41.44 5.8 0
TT 0.0018 7.69 7.16 6.9 -167

Figure 8 shows pressure contour for case with tip plate takes place in passages as compared to pressure drop in tip and
1at the symmetry plane. It shows that very little pressure drop lateral ejection holes. It also indicates pressure drop in

Copyright © 2006 by NSFMFP

5
serpentine passage is higher than trailing edge and leading edge
passages. Figure 9 shows pressure contour in plane
perpendicular to y-axis. In this plane different pressures are
observed. Figure 10 shows velocity contours at symmetry plane
for case with plate 1. It shows passage highest velocity in
serpentine passage whereas leading edge passage is having
lowest velocity. Trailing edge passage shows three different
zones of velocities. In first and second bends of serpentine
passage shows pockets of low velocity at corners. It clearly
shows thicker boundary layer at left side and thinner boundary
layer at right side on upstream of the bend. Figure 11 shows
vector plot for velocity at symmetry plane for plate 1. Figure 12
shows velocity contour for leading edge passage at symmetry
for plate 9. This figure clearly shows impingement. It also
indicates jet deflected up in leading edge passage showing flow
moving up towards tip. Figure 13 shows contours of turbulent
kinetic energy k. The turbulent kinetic energy is higher in area
of holes (not shown in figure). The figure shows high k in areas
like divider wall blunt portion, trailing edge below tip portion Fig. 9 Static pressure contour for case with plate 3 in plane
and leading edge passage. perpendicular to y axis at y= 60mm from inlet.

Table 9 Comparison between numerical and measured supply

pressures for different tip plates.

Case with Predicted Measured %∆p

plate no Pressure, Pa Pressure, Pa
FLUENT
Plate 1 2309.5 1559.6 -48.1
Plate 2 2146 1481 -44.9
Plate 3 1975.2 1480.4 -33.4
Plate 4 1980 1417 -39.7

Fig. 10 Velocity contours for case with plate 1 at symmetry

plane.

Fig. 8 Static pressure contour for case with plate 1 at symmetry

plane.
Fig. 11 Vector plot for case with plate 1 at symmetry plane.

Copyright © 2006 by NSFMFP

6
 Lab., IIT Bombay, are acknowledged for their help in
fabrication of the experimental setup.

REFERENCES
Borse, S. L. and Date, A. W., 2006, Flow Distribution inside
Internally Cooled Gas Turbine Blade without Film Cooling
Holes, 18th National & 7th ISHMT- ASME Heat and Mass
Transfer Conference IIT Guwahati, India.

Borse, S. L. and Date, A. W., 2006, Flow Distribution inside

Internally Cooled Gas Turbine Blade with Film Cooling Holes,
13th International Heat Transfer Conference, Sydney, Australia.

Han, J. C., Dutta, Sandip and Ekkad Srinath, 2000, Gas Turbine
Heat Transfer and Cooling Technology, Taylor & Francis Inc.
London, 1st Edition, 2-25.

Fig. 12 Velocity contour- leading edge passage for plate 4 at Hwang, G. J., Tzeng, S. C., Mao, C. P. and Soong C. Y., 2001,
symmetry plane. Heat Transfer in a Radially Rotating Four-Pass Serpentine
Channel with Staggered Half- V Rib Turbulators, J. Heat
Transfer, 123, 39-50.

Launder, B. E. and Spalding, D. B., 1974, “The Numerical

Computation of Turbulent Flows”, Computer Methods in
Applied Mechanics and Engineering, 3, 269-289.

Fig. 13 Turbulent Kinetic Energy contours for plate 1 at

symmetry plane.

CONCLUSIONS
It shows capability of standard k-ε model with standard wall
function to predict flow distribution. It shows good agreement
for mass flow prediction. The supply pressure prediction is
poor (predicts about 40% higher). Cross section of supply
plenum shows different pressures in same plane.
In future prediction with revised version of k-ε such
as renormalized group (RNG) k-ε model and realizable k-ε
model can be done to improve pressure prediction.

ACKNOWLEDGMENTS
Author is grateful to Prof. A. W. Date and Prof. R. P. Vedula,
Department of Mechanical Engineering, IIT Bombay, for useful
discussion. The staffs of Central Workshop and Steam Power

Copyright © 2006 by NSFMFP

7