INTRODUCTION
Gas Turbine engines are designed to continuously and efficiently convert the
energy of fuel into useful power and are developed into very reliable, high
performance engines. Now, gas turbines are widely used in power plants, marine
industries as well as for aircraft propulsion.
Starting from early 1970s, many researches were carried out by aircraft and
power generation gas turbine designers aiming at increase the combustion chamber
exit and hence the highpressure turbine stage inlet temperatures. By increasing
the combustion chamber exit temperature, the engine thermal efficiency can be
improved and fuel consumption can be reduced.
The cooling technology can be divided into two techniques that are internal
and external cooling. For external cooling, a cooled fluid is injected as a
jet to interact with the hot external fluid flow (Bounegta
et al., 2010). For the internal cooling, in order to increase the
heat transfer from turbine blade to the coolant air, different technologies
were used like ribs, Pinfin and dimples cooling and impingement cooling (Gao
and Sunden, 2001; Dahlquist, 2008; Donald,
2008).
The temperature distribution of compressed air can be categorised into two
different cooling channel configurations. The first type is called smooth due
to the nonribbed channel walls. Meanwhile, for the second type, the channel
walls are artificially roughened by regular repeated ribs as shown in Fig.
1.
Factors like channel aspect ratio, rib configuration and flow Reynolds number
will affect the heattransfer performance in a stationary ribbed channel. In
general, typical ribs heights that are used for experimental studies, Fig.
2, are around 510% of channel hydraulic diameter. p/e ratio varying from
7 to 15 (Han, 2004).
AlKayiem and Ghanizadeh (2011) have published detailed
analysis on the GT blade cooling with smooth and different ribbed channel cases.
The set of equations was discretized and solved numerically by finite difference
method solved by matrix inversion. They have concluded that channels ribbed
with 60° angle and rib blockage ratio, e/D_{h} = 0.078 produced
the best performance based on highest convection heat transfer coefficient.
The objective of this paper is to simulate the gas turbine blade cooling by
using finite volume method employing the commercial ANSYSFLUENT software. The
investigation was proposed to cover the thermal analysis of the entire blade
body at smooth and ribbed channels. The same operational conditions employed
by AlKayiem and Ghanizadeh (2011) has been adopted
her.

Fig. 1(ab): 
(a) Typical coolant channel in turbine airfoils and (b) Internal
arrangement (Han et al., 2000) 

Fig. 2: 
Scheme of flow separation and rib orientations in heattransfer
coefficient enhancement (Han, 2004) 
MATERIALS AND METHODS
The methodology of the present study is entirely numerical. Some measurements
have been involved to identify the profile of the blade.
The blade model: The selected blade for the current analysis is 143
MW ABB GT13E2 supplied by Lumut Power plantMalaysia. A Laser digitizer is used
to obtain the exact profile of the blade in three dimensions, where the airfoil
surface points are captured at different 210 mm span wise locations. The matrix
of profile points is exported to AutoCad to model the blade and the model then
exported to GAMBIT for discretization.
The cooling channel: The cooling channel has rectangular cross section.
The size of the channel is selected as 18x9 mm and its height is 210 mm. The
Hydraulic Diameter of the cooling channel, D_{h} is 0.012 m.
The ribs: The specifications of the ribbed channel used are chosen based
on the highest predicted convection coefficient h_{ribbed }= 559.32
W m^{2} K case which is at rib blockage ratio, e/D_{h} = 0.078,
pitchtoribheight ratio p/e = 8 and rib angle α = 60° (6, 7). Accordingly,
the ribs dimensions are height, e = 0.936 mm and pitch, p = 7.488 mm. Hence,
the number of ribs in each side of the cooling channel becomes 26. With that,
the configuration of the ribbed side in the cooling channel is as shown in Fig.
3.
Numerical implementation: The digitized shape of the blade is converted
to 3D drawing by AUOCAD and then to 3D model in GAMBIT. Then, the channel
is introduced in the model as shown in Fig. 4. Ribbed channels
were also modeled and considered since the work requires comparison between
smooth and ribbed channels.
The grid independency was determined. This is because with a correct size of
grid meshing, the total elements can be reduced. With fewer totals of elements,
the simulation can be run smoothly without large load exerted on the computer.
This can also reduce the time used for simulation.
There are three variables in the size function to be looked into; which are
starting size, growth rate and size limit. The selection of the size function
on the blade will be stopped when the temperature differences exceed 5%. Once
the size function on the blade is determined, the size function on the channel
will be considered.

Fig. 3: 
Configuration of rib on the channel wall 

Fig. 4: 
3D model of the blade with the channel 
The boundary conditions adopted for the simulation are based on a typical gas
turbine operational environment as following:
• 
Temperature of hot gases attacking the blade is 1700 K 
• 
Temperature of compressed air (T_{a}) entering the blade root
is 400 K 
• 
Convection heat transfer coefficient (h_{8}) for the hot gases
is 1000 W m^{2} K^{1} 
• 
Mass flow rate of compressed air entering the channel is 0.01 kg sec^{1} 
• 
Thermal conductivity of the blade metal is assumed to be 25 W m^{1}
K 
RESULTS AND DISCUSSION
Thermal analysis the key to study the behaviour of the materials subjected
to high temperature medium, like the case of the GT blades. This analysis will
allow the designer to create appropriate cooling techniques to enhance the performance
of the GT. A validated CFD analysis is representing a promising tool for thermal
analysis, as the case of the present study. One of the requirements in the CFD
simulation technique is the grid independency on the results.
Grid independency test: For the grid independency, two criteria are
used to choose the size function for blade as follows:
• 
The temperature (if it does not have error in meshing) 
• 
The total number of elements 
Three points (A, B and C) on the blade, as in Fig. 5, are
selected to check on the variation and accuracy of the temperatures by varying
the grid structure and element numbers. Start size, growth rate and size limit
are the variables to be changed until a maximum temperature difference of not
more than 5% is obtained. If the blade cannot be meshed for a bigger growth
rate, then the previous size of growth rate will be chosen.
Blade size function: Table 1 shows the temperature
differences for different blade size functions. From Table 2
and 3, for a constant growth rate of 1, the temperature does
not affect much on the meshed blade although the size limit is changing. Thus,
the growth rate was increased to 3, 6, 9 and 12 for different size limit of
0.9, 3, 6, 9 and 90.

Fig. 5: 
Points A, B and C for grid independency checking 
Table 1: 
Size function variable for blade with different growth rate 

The final selected size function with starting size of 0.9 is 0.933. This
is because it has less total elements, so it reduces the load of the software
during fluent simulation. Starting size of 0.9 and 1.8 are used for comparison
after starting size function 0.9 has been chosen (0.933). Starting size function
of 1.8 is to be considered because (1.833) has a lower number of total elements
which is 162162, compared to size function (0.933) which has 164736 elements.
Size function of 0.933 and 1.833 will be used to determine the channel size
function.
Channel size function: The blade size function (0.933) and (1.833)
are now be used to investigate the best size function for the channel. Table
4 shows the channel size function configuration meshing. The error that
occurs during meshing is due to the size limit of the channel which does not
comply with the starting size and growth rate.
Table 2: 
Size function variable for blade with same size limit for
comparison 

Table 3: 
Channel size function configuration meshing 

Table 4: 
Size function variable for channel 


Fig. 6: 
The optimum meshing criteria with channel size function 0.931.8
and blade size function 1.833 
Table 4 shows the temperature differences for different channel
size functions. Channel size function 0.931.8 and blade size function 1.833
has been chosen for the simulation. Points A, B and C at that size function
have reached the lowest temperature difference among the configuration which
is 3.5%. This temperature difference did not exceed the limit that has been
set which is 5% temperature difference. The final optimum meshing criteria with
the selected functions sizes is shown in Fig. 6.
Simulation results: To compare between the recent simulation and the
previous works, four reference points at the centre of each side of the channel
are selected and shown as A, B, C and D in Fig. 7.
Smooth channel case: The simulation result for the case of smooth channel
is Fig. 7 as temperature contours. The temperatures on the
smooth channel blade is compared with the previously published results by AlKayiem
and Ghanizadeh (2011) which are obtained by using analytical/finite difference
coupled method.

Fig. 7: 
Temperature at 4 points near smooth channel showed in FLUENT 
Table 5: 
Comparison between analytical/FD solution and the recent
simulation at blade root. (smooth channel case) 

Table 6: 
Comparison between previous analytical/FD solution and the
recent simulation, at blade root at height, H =157.5 mm (Smooth channel
case) 

Table 5 shows that the temperature difference between the
results obtained from this study those by using the recent numerical method
is less than 5% at the root of the blade.
However, that depends on the height of the blade. While at the third section
of the height of the channel (H = 157.5 mm), as shown in Table
6, the temperature differences are bigger especially at point D which is
located toward the tail of the blade. The temperature different is 15.92% at
point D, while points A, B and C have differences less than 10% when comparing
the two method.

Fig. 8: 
Temperature distribution of 60 degrees rib 

Fig. 9: 
Temperature distribution of smooth channel 
Ribbed channel: Simulation for rib channel is carried out for the ribbed
channel with rib angle α = 60°, heighttohydraulicdiameter ratio
e/D = 0.078 and pitchtoheight ratio p/e = 8. Figure 8 shows
the temperature distribution for a ribbed channel with ribs at 60 degrees. While
Fig. 9 shows the temperature distribution in a smooth channel.
However, Table 7 shows the temperature difference between
the two findings.
To calculate the rib efficiency, the following equation is used:
Table 7: 
Temperature differences between smooth channel and rib channel 

The contribution of the ribs to enhance the cooling performance is the mean
of the three points, A, B and C. based on equation, the ribbed channels have
enhanced the cooling performance by 9.5%. Meaning to say that the ribs are able
to reduce the blade temperature by 9.5% which is considerable reduction.
CONCLUSIONS
A numerical analysis has been carried out by using Gambit modelling and FLUENT
simulation software. Gird independency test is performed to reduce the total
number of cells in order to minimize the total time and load when running the
simulation on fluent. The best size function for the channel is starting size
of 0.9, growth rate of 3 and size limit of 1.8 with blade size function starting
size of 1.8, growth rate of 3 and size limit of 3.
Comparison has been done on the smooth channel between analytical/numerical
coupled method results and numerical method (ANSYSFLUENT) results. The difference
found is 4.35% at the root of the blade. Meanwhile, on the height of 157.5 mm,
the temperature difference found is slightly higher, mostly less than 10% except
at point D which has 15.92%. This may be due to the location which is located
further behind and toward the tail of the blade.
Another comparison is carried out for the 60 degrees rib angle with pitchtoheight
ratio, p/e = 8 and heighttohydraulicdiameter, e/D = 0.078 with the smooth
channel. The results show that with the specified rib, the blade can be cooled
by 9.5% compared to the smooth channel case.
ACKNOWLEDGMENT
The authors would like to express their sincere gratitude and appreciation
to Universiti Teknologi PETRONAS for providing the necessary facilities to conduct
the experiments for this study.