A Combined Numerical and Experimental Study of Heat...
Transcript of A Combined Numerical and Experimental Study of Heat...
Proceedings ofISROMAC 2016
Honolulu, Hawaii,USA,April 10-15,2016ISROMAC-2016-44042
A Combined Numerical and Experimental Studyof Heat Transfer in a Cooling Channel
Roughened with 90◦ Ramped Ribs
M.E. Taslim and B. Ren
Mechanical and Industrial Engineering Department
Northeastern University
Boston, MA, USA
Abstract
Varieties of cooling methods have been used to protect
the hot sections in modern gas turbine engines so as to allow
a higher inlet temperature for increasing turbine efficiency.
One such method is to route the cooling air through passages
roughened with ribs within the airfoils in order to enhance
the convective heat transfer coefficients in those passages.
Much work is reported in open literature dealing with a va-
riety of rib geometries and flow arrangements. This study’s
focus, however, is on the ramped-ribs. Both upstream- and
downstream-ramped ribs are investigated. The presence of
these ramps could be by design, or due to casting imperfec-
tions, or by the accumulation of micron-size sand particles
that find their way from the engine inlet to the cooling pas-
sages when the engine is operated in harsh environments.
The ramp can be formed on both sides of the ribs in the
flow direction. A square channel roughened with 90◦ ribs
of three geometries was tested. Square as well as ramped
(with decreasing or increasing slopes in the flow direction i.e.
ramping up or down) ribs in a staggered arrangement were
studied. Square ribs of the same height as the ramped-ribs
were tested first to which the ramped-rib results were com-
pared. The numerical models contained all the features of the
tested geometries. The applied thermal boundary conditions
to the CFD models matched the test boundary conditions.
Numerical results were obtained from a three-dimensional
unstructured computational fluid dynamics model with over
13 million hexahedral elements. For turbulence modeling,
the realizable k − ε was employed in combination with the
standard wall functions. In the experimental part, all these
geometries were built and tested for heat transfer coefficients
at a Reynolds numbers range from 10,000 to 60,000, using
steady state liquid crystal thermography. Comparisons were
made between the test and numerically-obtained results in
order to evaluate the employed turbulence models and vali-
date the numerically obtained results. The test and numer-
ically evaluated results showed reasonable agreements be-
tween the two for all rib geometries. Friction factors were
also measured, and both heat transfer and friction factor re-
sults for the three rib geometries were compared. The re-
sults showed that there is a considerable penalty, up to about
28%, in heat transfer coefficients for the ramped ribs. Pas-
sage pressure drop (friction factor) could decrease by up to
43%. However, the deficit in heat transfer coefficients could
have significant impact on the airfoil life.
Nomenclature
A channel area, 25.81cm2
Dh hydraulic diameter of test section, 5.08 cm
e rib height, 0.5715 cm
P rib pitch, 5.715 cm
w rib width, 0.5715 cm
f Darcy friction factor=∆P(Dh/L)
12 ρU2
m
fs smooth wall friction factor
h heat transfer coefficient, W/(m2K)L rib-roughened length of the channel, 28.575 cm
1
i current through the surface foil heater, amps
k air thermal conductivity, W/(mK)m air mass flow rate, Kg/s
N̄u area-averaged Nusselt number based on the channel hy-
draulic diameter,hDh/k
Nus all smooth channel Nusselt number from Dittus-
Boelter correlation
Pamb ambient (lab) pressure, Pa
P channel perimeter, 20.32 cm
Pr Prandtl number
q′′
total heat flux generated by the foil heater (= vi/Aheater)
q′′
b total heat loss through the polyurethane back wall to the
ambient
q′′
r total radiational losses from the heated measurement
wall to the surrounding unheated walls
rrib rib top corner radii, 0.127 cm
Re Reynolds number based on the channel hydraulic diam-
eter, 4m/Pµ)Tf film temperature where the heat transfer coefficient is
measured, (Tm +Ts)/2
Tm Air mixed mean temperature where the heat transfer co-
efficient is measured
Ts surface temperature where the heat transfer coefficient
is measured
Um air average velocity
v voltage across the surface foil heater, volts
α rib angle with the flow direction, 90◦
∆P pressure drop across the rib-roughened length of the
channel, L
ρ air density, Kg/m3
1 Introduction
Various methods have been developed over the years to
keep turbine airfoil temperatures below critical levels. The
main purpose in turbine airfoil cooling is achieving maxi-
mum heat transfer coefficients while minimizing the coolant
air flow rate. One such method is to route the air through
rib-roughened serpentine passages within the airfoil and re-
move heat from the airfoil by convection. The coolant is
then ejected either at the tip of the airfoil, through the cool-
ing slots along the trailing edge or cooling holes along the
airfoil surface. Shed vortices from the rib tip and formation
of secondary flows give rise to the level of mixing of the
coolant air in the cavity core with the warmer near-wall air
thus increase the heat transfer from the airfoil walls. Addi-
tional heat transfer gain is accomplished due to the increased
extended surface area. However, roughening the walls in-
creases the friction factors in the cooling cavities which re-
sults in higher pressure drop penalties. Figure 1 shows a typ-
ical internal cooling arrangement for a multi-pass turbine air-
foil (Ligrani[1]).
Geometric parameters such as rib height to passage hy-
draulic diameter or blockage ratio e/Dh, rib angle with the
flow direction(α), the manner in which the ribs are posi-
tioned relative to one another (in-line, staggered, crisscross,
etc.), rib pitch-to-height ratio P/e, and rib corners (round vs
sharp corners, fillets, skewness towards the flow direction)
have pronounced effects on both local and overall heat trans-
fer coefficients. Some of these effects were studied by dif-
ferent investigators such as Burggraf [2], Webb et al. [3],
Han et al. [4], Metzger et al. [5], Han [6], Han et al. [7],
Metzger et al. [8], Metzger and Vedula [9], Chandra [10],
Chandra and Han [11], Han et al. [12], Zhang et al. [13],
Dutta and Han [14], Taslim and Spring [15], Taslim et al.
[16], Taslim et al. [17]. Numerous experimental and numeri-
cal studies have been performed in rectangular rib-roughened
cooling channels. Arts et al. [18] compared the computa-
tional results from a three-dimensional Navier-Stokes solver
for heat transfer of a steady viscous compressible flow in a
square channel with one rib-roughened wall with detailed ex-
periments. The three-dimensional computations captured the
correct position of the reattachment point using the k− l tur-
bulence model. Gupta et al. [19] measured the local heat
transfer distributions in a double wall ribbed square chan-
nel with 90◦ continuous, 90◦ saw tooth profiled and 60◦ V-
broken ribs. Taslim and Spring [20] used liquid crystals to
study the effects of rib profile and spacing on heat transfer
coefficient. They concluded that low aspect ratio ribs, es-
pecially with round corners, produced lower heat transfer
coefficients. They also found an optimum pitch-to-height
ratio for 90◦ square ribs was around 8. Gao and Sunden
[21] investigated the thermal and hydraulic performance of
three rib-roughened rectangular ducts. The ribs were ar-
ranged staggered on the two wide walls. They employed
liquid crystal thermography in the heat transfer experiment
to demonstrate detailed temperature distribution between a
pair of ribs on the ribbed surfaces. They found that the sec-
ondary flows caused by the inclined ribs created a significant
spanwise variation of the heat transfer coefficients on the rib-
roughened wall with high heat transfer coefficient at one end
of the rib and low value at the other. In the streamwise di-
rection between two consecutive ribs, the temperature dis-
tribution showed a sawtooth variation because of flow reat-
tachment. Lu and Jiang [22] experimentally and numerically
investigated forced convection heat transfer of air in a rectan-
gular channel with 45◦ ribs on one wall. They compared the
experimental and numerical results and showed that the SST
k−ω turbulence model was more suitable for the convection
heat transfer in such channels than the RNG k−ε turbulence
model. They found that the average heat transfer coefficients
increased with increasing mass flow rates and decreasing
spacings. Peng et al. [23] experimentally and numerically
studied convection heat transfer in a channel with 90◦ ribs
and V-shaped ribs. They found that both the 90◦ ribs and
V-shaped ribs enhanced the convection heat transfer com-
pared with a flat wall without ribs, but the pressure drop also
increased. They also compared continuous ribs and inter-
rupted ribs and concluded that the heat transfer with the 90◦
interrupted ribs is more than with the 90◦ continuous ribs.
Promvonge and Thianpong [24] conducted experiments to
assess turbulent forced convection heat transfer and friction
loss behaviors for air flow through a constant heat flux chan-
nel fitted with different shaped ribs. The rib cross-sections
used in this study were triangular (isosceles), wedge (right-
2
triangular) and rectangular shapes. They introduced two rib
arrangements, namely, in-line and staggered arrays. The ex-
perimental results showed a significant effect of the presence
of the ribs on the heat transfer rate and friction loss over the
smooth wall channel. They found that the in-line rib arrange-
ment provided higher heat transfer and friction loss than the
staggered one for a similar mass flow rate. Kamali and Bi-
nesh [25] developed a computer code to study the turbulent
heat transfer and friction in a square duct with various shaped
ribs, mounted on one wall. The simulations were performed
for four rib shapes, i.e., square, triangular, trapezoidal with
decreasing height in the flow direction, and trapezoidal with
increasing height in the flow direction. The prepared algo-
rithm and the computer code were applied to demonstrate
distribution of the heat transfer coefficient between a pair of
ribs. The results showed that features of the inter-rib distri-
bution of the heat transfer coefficient are strongly affected by
the rib shape and trapezoidal ribs with decreasing height in
the flow direction provide higher heat transfer enhancement
and pressure drop than other shapes. Taslim and Liu [26]
showed that CFD could be considered as a viable tool for the
prediction of heat transfer coefficients in a rib-roughened test
section. In the numerical part, a square channel roughened
with 45◦ ribs of four blockage ratios (e/Dh) of 0.10, 0.15,
0.20, and 0.25, each for a fixed pitch-to-height ratio (P/e)
of 10, was modeled. Sharp as well as round-corner ribs (r/e
= 0 and 0.25) in a staggered arrangement were studied. In
the experimental part, a selected number of these geometries
were built and tested for heat transfer coefficients at elevated
Reynolds numbers up to 150000, using a liquid crystal tech-
nique.
As for the heat transfer coefficient data on the sur-
faces of the rib (in contrast to the heat transfer coeffi-
cient on the area between the adjacent ribs), several stud-
ies are reported. The reported work of Metzger et al.
[27], Taslim and Wadsworth [28], Korotky and Taslim [29],
Taslim and Lengkong [30],Taslim and Korotky [31], Taslim
and Lengkong [32] cover a wide range of pertinent parame-
ters affecting the on-rib heat transfer coefficients.
When gas turbines are operated in harsh environments
such as sand storms or particle-laden winds, extremely fine
particles,(below 10 microns in diameter) are ingested into
the turbomachinery, compromise engine performance and
significantly reduce the normal life expectancy of the en-
gine. These particles could find their way into the airfoils
cooling cavities and block the cooling features such as film
holes, tip holes, crossover holes and the trailing-edge slots.
Compounded by high gas temperatures, the particles may go
through chemical reactions and conglomerate to make de-
posits at the turbulators’ roots, 180-degree turns in the ser-
pentine cooling cavities, around the base of the pins in the
pin banks and in the trailing-edge holes and slots, to name a
few. Similar behavior is observed around the airfoils outer
peripheries especially on their platform. Deposition of these
particles on different cooling features often has severe unde-
sirable effects on their thermal performance. The severity of
these effects is different for different cooling features and it
has to be studied one by one. Hot and harsh environment is
Fig. 1. Typical internal cooling arrangement for a multipass turbine
blade [1].
not the only cause of altering the cooling features effective-
ness. Casting imperfections and wear and tear of the core
dies could also be other causes.
In this study, downstream- and upstream-ramped ribs,
simulating the changes in rib shape due to the accumulation
of foreign particles such as sand in front or on the back side
of the ribs, are studied for their effects on the heat transfer
coefficient as well as the friction factor in the cooling cavities
of a gas turbine airfoil.
2 Test Section
Figure ?? shows schematically the test section and rib
geometries. Steady state liquid crystal thermography tech-
nique is used for the measurement of heat transfer coeffi-
cient. In this technique, the most temperature-sensitive color
displayed by the liquid crystals is chosen as the reference
color corresponding to a known temperature. By proper ad-
justment of the Ohmic power to a very thin foil heater im-
mediately underneath the liquid crystals, the reference color
is moved from one location to another so that the entire area
of interest is eventually covered with the reference color at
one time or another. This process results in a series of pho-
tographs which correspond to a certain location of the refer-
ence color. Among the advantages of liquid crystal thermog-
raphy is to depict the flow ”footprints” and local values of
heat transfer coefficient on the surface under investigation.
This simultaneous flow visualization enhances the under-
standing of the underlying physics and helps the investiga-
tor in interpretation of the results. Furthermore, unexpected
asymmetries in flow are revealed as well as the slightest heat
and flow leaks, nonuniformities in surface heat flux, imper-
fections associated with the attachment of the heater to the
3
� � � �
� ���
� �
��
������
����
����
������ !"#$
% &' ()*
+ , - . / 0 , 1 , 2 34 0 , -
5 6 7 8 8 9 6 7: ; < 8
= >?@A= >?@B= >?@C
= >?@D E
FGHI
J K L M N O P Q R S
T U V W S X N O M Y Z P M Y [ O \ P Q R S] [ S X N O M Y Z P M Y [ O \ P Q R S
^ _ ` a
b c d _ ` e f
g h i j k l m
n o p q
r s o t uv w x y z { v | } ~
� � � � � � �
� � � � � � �
� � � � � � �
� � � � � � �
� � � �
� � � � � � � � �
¡ ¢ £ ¤
¥ ¦ § ¨ ¨ © ª¥ « ¬ ® ¯ © ° ± ² ¬ ¨
³ ´µ¶
· ¸ ¹ º
» ¼ ½ ¾ ¿ À Á
Â Ã Ã Ä Å Æ Ç È É Å Ê È É Ë Ì Ç Å È Í Æ
ÎÏ Ð Ñ Ò Ó Ô Õ Ö
Ñ Ô Ò× Ø ÙÚ Û Ü Ý Ý Þ ß Þ à
Fig. 2. schematics of the test section and the ribs.
test section surface and nonuniformities in wall material ther-
mal conductivity. The test section was a 5.08cm× 5.08cm
square channel with a length of 91.44cm, three walls of
which were made of 1.27cm−thick clear acrylic plastic. The
fourth wall, on which a sheet of encapsulated liquid crystals
was attached and all measurements were taken, was made
of a 10.16cm-thick machinable polyurethane board with a
thermal conductivity of 0.0576W/mK in order to mitigate
heat losses from the heaters to the ambient. Three 5.08cm
x 28.2cm custom-made etched foil heaters were glued onto
the polyurethane wall where measurements were taken. The
heaters covered the entire polyurethane wall including the
smooth entry and exit lengths. The 0.15-mm-thick etched
foil heaters were made of a 0.0127mm-thick inconel heat-
ing element and a 0.0127mm-thick electrically inactive in-
conel foil to further spread the heat uniformly over the
polyurethane surface. These two inconel foils were sand-
wiched and glued between three layers of Kapton. It is noted
that in a stationary channel, the heat transfer coefficient is
not sensitive to the number of heated walls [33]. The test
section was covered on all sides, except for a small win-
dow at the location where the pictures were taken, with a
5-cm-thick styrofoam slab to minimize heat losses to the en-
vironment. The radiational heat loss from the heated wall
to the unheated walls as well as losses to ambient air were
taken into consideration when heat transfer coefficients were
calculated. A digital camera, in conjunction with proper fil-
ters and background lighting to simulate daylight conditions,
was used to take pictures of isochrome patterns formed on
the liquid crystal sheet. A centrifugal compressor supplied
compressed air to a 0.76m3 storage tank. A combination of
an air dryer and two air filters dried and cleaned the air. A
pressure regulator was used to set the air mass flow rate for
a desired jet Reynolds number. A critical venturi, choked for
all mass flow rates, was then used to measure the air mass
flow rate before it entered a 50.8cm × 50.8cm × 53.34cm
plenum equipped with a honeycomb flow straightener, and
to the the test section at about ambient temperature. To-
tal mass flow rate entering the supply channel varied from
0.0093 to 0.056Kg/s. Heat was induced to the air in the
test section via the heaters through a custom-designed power
supply unit. Each heater was individually controlled by a
variable transformer to assure a constant heat flux over the
entire heated wall. Two thermocouples at the channel inlet
measured the air inlet temperature. Their measurements did
not differ by more than a fraction of a degree. A typical air
temperature at the channel inlet was 21oC. A pressure tap
measured the static pressure in the plenum and four pres-
sure tap readings, one on each wall at the test section inlet
were averaged for the test section inlet pressure. Five ribs,
also machined out of clear acrylic plastic, were mounted on
the liquid crystal wall using a double-sided tape with min-
imum deformation, and five ribs on the opposite clear wall
in a staggered arrangement. Ribs were normal to the flow
direction. All Ribs had a height of 0.572cm, corresponding
to a blockage ratio, e/Dh, of 0.125 and they were 5.72cm
apart, corresponding to a pitch to height ratio, P/e, of 10.
Measurements for all three cases were performed on the area
between the third and fourth ribs in the flow direction on the
liquid crystal side. This area is shown in Fig. 2 as the mea-
surement area and the reported heat transfer coefficient is the
area-weighted heat transfer coefficient for the area between
the third and fourth ribs on the liquid crystal side. The rib-
roughened length in all cases was 28.58cm, so approximately
62.87cm of the channel length at the inlet and the exit was
smooth. The smooth inlet section simulates the cooling cav-
ity in the dovetail region of an airfoil. Heat flux on the heated
wall was provided by the foil heaters through a custom de-
signed power supply unit. Before testing, the liquid crystal
sheet was calibrated as follows. A water bath was used to
attain uniform isochromes on a small sample piece of the
liquid crystal sheet used throughout this investigation. The
temperature corresponding to each color was measured using
a precision thermocouple and photographs were taken at lab-
oratory conditions simultaneously so as to simulate closely
the actual testing environment. A reference color along with
its measured temperature of 34.6oC was then chosen to be
used throughout the experiments. It should be noted that all
possible shades of the selected reference color showed a tem-
perature difference of no more than 0.3oC. A contact micro-
manometer with an accuracy of about 0.0254 millimeter of
water column measured the pressure drop across the channel
for the calculation of friction factor.
For a typical test run, the Reynolds number was set by
precisely fixing the mass flow rate. The heat flux was then
induced by turning on the main power supply and adjust-
ing heater power until the first band of reference color was
observed on the liquid crystal sheet in the area of interest.
Enough time was allowed so that the system came to thermal
equilibrium when a photograph was taken and data recorded.
The power to the heaters was then increased so that the ref-
4
erence color was moved to a location next to the previous
one and another picture was taken. This procedure was con-
tinued until the entire surface of interest on was covered by
the reference color at one time or another. The process was
repeated for all Reynolds numbers. Each photograph was
digitized in order to measure the area covered by the refer-
ence color. This was done by using a commercial software
package installed on a PC. Once the areas (pixels) were mea-
sured, the area-weighted average heat transfer coefficient for
the surface area between the third and fourth ribs was calcu-
lated. Experimental uncertainties in friction factor and heat
transfer coefficients, following the method of Kline and Mc-
Clintock [34], were ±5% and ±6%, respectively.
3 Computational Model
The computational model was constructed for the en-
tire rib-roughened channel. Figure 3 shows the mesh for
the test section and square ribs while Fig. 4 shows the de-
tails of the mesh distribution around the rib-roughened sec-
tion of the channel with ramped ribs. The CFD analyses
were performed using Fluent/UNS solver by Ansys, Inc., a
pressure-correction based, multi-block, multi-grid, unstruc-
tured/adaptive solver. Incompressible fluid with a turbulent
Prandtl number of unity is assumed. Boundary conditions for
the numerical models were identical to those of the experi-
ments. At the inlet, a total mass flow rate exactly the same
as what was measured was specified at the same temperature
(18−25 ◦C range) and pressure (101.35−105.4 KPa range)
of the air entering the rig. The turbulence intensity at the inlet
was set to 5%. The heat fluxes on the heated walls were also
identical to those of experiment (2500−4000 W/m2 range).
Channel exit had a pressure boundary condition identical to
that of the lab. The realizable k − ε turbulence model was
employed in combination with the standard wall functions.
The average y+ for the first layer of cells was calculated to
be below 5 for all cases. Other available turbulence models in
this commercial code including the k−ω with Shear Stress
Transport (SST) option and the realizable k− ε turbulence
model in combination with enhanced wall treatment were
also tested and the corresponding results are compared. Cells
in all models were entirely hexagonal, a preferred choice for
CFD analyses, and were varied in size bi-geometrically from
the boundaries to the center of the computational domain in
order to have finer mesh close to the boundaries. Mesh inde-
pendence was achieved at about 12 million cells for a typical
model. For a representative geometry, the heat transfer re-
sults of a series of meshed models with different number of
elements were compared. As the mesh became more and
more refined, the heat transfer results came closer and closer
to each other. When a difference of a fraction of a percent
between two consecutive meshes was observed, no further
refinement was done. We reached that situation when the
total number of elements was about 13 million. We started
at 8M and increased the element numbers by about 2M at a
time first, and 1M at the end. Residual sums for all variables
in all models were less than 1x10−7. Convergence, for most
Fig. 3. A typical CFD model representing the entire square-rib test
section.
cases, was achieved at around 25,000 iterations.
4 Results and Discussion
As was mentioned earlier, three turbulence models were
tested for our numerical models. The CFD results of these
three turbulence models - realizable k− ε with standard wall
functions, realizable k− ε with enhanced wall treatment and
5
Fig. 4. Details of the mesh around the ramped ribs.
k−ω with Shear Stress Transport (SST) are compared for
the square rib geometry in Fig. 5. These cases were run un-
der otherwise identical conditions i.e. same mesh size and
arrangement and same boundary conditions. Only the turbu-
lence model was changed. Measured values are also shown.
It was concluded that the realizable k− ε turbulence model
with standard wall functions produced the closest results to
the measure values, thus in all reported CFD results, this tur-
bulence model was employed.
The first test for each rib geometry was a cold test for the
measurement of the pressure drop across the rib-roughened
channel and calculation of the Darcy friction factor for a
range of Reynolds numbers. The surface heaters were off
so that the air properties could be evaluated accurately at the
measured channel inlet temperature. Figure 6 compares the
test as well as the numerically-obtained friction factor re-
sults for the three rib geometries. It should be noted that,
in these and ensuing figures, the symbols represent the mea-
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.50
50
100
150
200
250
300
350
400
450
z/Dh
Nu
Re
Area 1
Top
Back
Test
600005000040000300002000010000
kωSST
Real. kε,Std.Wall Funcs.
Front
Real. kε withWall Treat.
Floor
Area 4Area 3Area 2 Area 5
Fig. 5. Comparison of the laterally-averaged local CFD Nusselt
number variation along the rib-roughened section of the channel us-
ing three turbulence models and the measured area-averaged Nus-
selt numbers.
0 10000 20000 30000 40000 50000 600000
0.05
0.1
0.15
0.2
0.25
0.3
Re
f
Square Ribs, TESTDownstreamRamped Ribs, TESTUpstreamRamped Ribs, TESTSquare Ribs, CFDDownstreamRamped Ribs, CFDUpstreamRamped Ribs, CFD
Fig. 6. Darcy friction factor variation with Reynolds number for the
three rib geometries - a comparison between the measured and
numerically-obtained results.
sured data while the lines represent the numerically-obtained
results. Square ribs, as expected, produce a much higher
pressure drop across the channel, compared with the ramped
ribs. This is explained by the fact that the square ribs intro-
duce a rather abrupt blockage to the flow while the ramped
6
18
17
16
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
Vmag
m/s
Square Ribs
DownstreamRamped Ribs
UpstreamRamped Ribs
Fig. 7. CFD contours of velocity magnitudes around the ribs for the
three rib geometries at Re=40,000.
ribs introduce a gradual blockage to the flow. CFD con-
tours of velocity magnitudes for the three rib geometries,
shown in Fig. 7, show wider areas of high velocity mag-
nitudes for the ramped rib cases indicating less resistance to
the flow. Test results do not show a difference in pressure
drop between the two ramped rib geometries beyond the ex-
perimental uncertainties while the numerical results show an
average of about 4.5% decrease in the friction factor for the
upstream-ramped ribs compared to the friction factor for the
downstream-ramped ribs.
The heat transfer coefficient corresponding to each
recorded picture of liquid crystals display on the measure-
ment area, shown in Fig. 2, was calculated from:
h =q′′−q
′′
b −q′′
r
(Ts−Tm)
where Ts and Tm are the surface and air mixed mean temper-
atures, respectively. q′′
is the total heat flux generated by the
foil heater, q′′
b is the total heat loss from the heaters to am-
bient through the back polyurethane wall and q′′
r is the total
radiational losses from the heated measurement wall to the
surrounding unheated walls. Air properties were evaluated
at the film temperature, Tf . Heat transfer results were gath-
ered for Reynolds numbers from about 10,000 to 60,000.
Figure 8 compares the tested and numerically-obtained
area-averaged Nusselt numbers for the three rib geometries.
The error bars on the symbols represent the experimental un-
certainties. Square ribs, as expected, produce higher Nusselt
numbers, compared with the ramped ribs. This behavior is in
agreement with Colburn Analogy of higher heat transfer co-
efficients corresponding to higher friction factors. It has been
established both experimentally and analytically that, given
10000 20000 30000 40000 50000 600000
50
100
150
200
250
300
350
Re
Nu
Square Ribs, TESTDownstreamRamped Ribs, TESTUpstreamRamped Ribs, TESTSquare Ribs, CFDDownstreamRamped Ribs,CFDUpstreamRamped Ribs, CFD
Fig. 8. Nusselt number variation with Reynolds number on the mea-
surement area for the three rib geometries - a comparison between
the measured and numerically-obtained results.
enough space between a pair of adjacent ribs in the flow di-
rection (P/e≥ 5), for the flow to re-attach after tripping over
the rib, the heat transfer coefficient reaches its maximum
value in the re-attachment zone and decreases monotonically
in the flow direction until it approaches the next rib where it
starts to increase again due to a stagnation point type of flow.
Streamlines of Fig. 9 and CFD contours of the Nusselt num-
bers, shown in Fig. 10, support this argument as well. There-
fore, air trips over the square ribs and re-attaches around the
middle of the area between the adjacent ribs. The shed vor-
tices from the rib top edges promote the mixing of the cooler
core air with the near-wall warm air thus increase the heat
transfer coefficient. This phenomenon occurs on the ramped
ribs to a lesser degree as the air climbs up or down the ramps
gradually thus the enhancement in heat transfer coefficient
is not as high as of that for the square ribs. The square and
upstream-ramped rib results are in agreement with the test
results. The downstream-ramped numerical results, however,
show a slight increase in heat transfer coefficients compared
with the square ribs. This behavior was observed regard-
less of the turbulence model employed in our CFD analyses.
Given that the pressure drop along the channel for the ramped
ribs were in agreement with the test results, it is speculated
that numerical analyses for the downstream-ramped ribs did
not capture all viscous effects. As the air Reynolds num-
ber increases, there is stronger interaction between the flow
and the ribs. These interactions increase the heat transfer
coefficients on the area in between the two ribs where the
measurement were performed.
Figure 11 shows the variation of the numerical Nusselt
numbers along the channel on the rib-roughened surface for
the square ribs. In general, the Nusselt number on the roof of
7
Fig. 9. CFD contours of streamlines around the ribs for the three rib
geometries at Re=40,000.
Fig. 10. CFD contours of the Nusselt numbers for the three rib ge-
ometries at Re=40,000.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.50
50
100
150
200
250
300
350
400
450
500
550
60000
50000
40000
30000
20000
10000
z/Dh
Nu
Re FLOW
z/Dh
Front
Top
Back
Floor
TopFloor
Back
Front
Area 1
Area 5Area 4Area 3Area 2
Area 1 Area 2 Area 3 Area 4 Area 5
Fig. 11. Laterally-averaged numerically-obtained Nusselt number
variation along the rib-roughened section for the square ribs.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.50
50
100
150
200
250
300
350
400
450
500
550
z/Dh
Nu
Re CFD Test
AreaWeightedAverage
FLOW
z/Dh
Front
Top
Back
Floor
TopFloor
Back
Front
600005000040000300002000010000
Area 1 Area 5Area 4Area 3Area 2
Area 1
Area 2 Area 3 Area 4 Area 5
Fig. 12. Numerical (laterally-averaged and area-averaged) and
measured (area-averaged)Nusselt number variation along the rib-
roughened section for the square ribs.
a rib is much higher than that on the other surfaces. The first
rib for all geometries does not benefit from the presence of
any upstream rib on the opposite wall. The second and other
ribs, however, receive the diverted air from the opposite rib-
roughened surface and produce a higher heat transfer coeffi-
cient as seen in Fig. 10. The back surface of the rib has the
lowest heat transfer coefficient due to the formation of a re-
circulating zone behind the rib (Fig. 9). The points indicated
by front, top, back and floor represent the area-averaged Nus-
selt numbers on those areas. A consistent pattern of Nusselt
number variation at different Reynolds numbers is what is
8
Z/Dh
FLOW Front RampTop Floor
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.50
50
100
150
200
250
300
350
400
450
500
550
z/Dh
Nu
Re CFD Test
AreaWeightedAverage
Front
Top
Floor
Area 1
Ramp
Area 3Area 2 Area 5Area 4
Area 1
Area 4Area 3Area 2 Area 5600005000040000300002000010000
Fig. 13. Numerical (laterally-averaged and area-averaged) and
measured (area-averaged)Nusselt number variation along the rib-
roughened section for the downstream-ramped ribs.
Z/Dh
FLOWBackRamp
TopFloor
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 60
50
100
150
200
250
300
350
400
450
500
550
z/Dh
Nu
Re CFD Test
AreaWeightedAverage
Back
TopFloor
Ramp
Area 1
Area 5Area 3Area 1 Area 4Area 4
Area 5Area 4Area 3Area 2
600005000040000300002000010000
Area 2
Fig. 14. laterally-averaged and area-averaged) and measured
(area-averaged)Nusselt number variation along the rib-roughened
section for the upstream-ramped ribs.
expected.
Figure 12 shows the same numerical results as in Fig.
11 with some new features. The parallel, almost horizontal
lines, are the ”area-weighted average” Nusselt number vari-
ations on the rib-roughed section at different Reynolds num-
bers. The symbols are the measured values corresponding to
the measurement area where the camera is located (Fig. 2).
There is an about 15% difference between the measured and
numerically-obtained Nusselt numbers. The CFD results for
the ramped ribs are depicted in Figs. 13 and 14. Figures
12, 13 and 14 have the same scales so the three rib geome-
tries can be compared. The downstream-ramped ribs show a
higher level of heat transfer coefficients which is not what ex-
pected. Since this rib geometry produced much lower pres-
sure drop, evidenced by Fig. 6, one would expect that the
heat transfer coefficients be lower than those for the square
ribs. Furthermore, the measured values for the downstream-
ramped ribs are lower than those of the square ribs. There-
fore, it is speculated that the CFD model does not capture the
negative effects of the recirculating zone in front of each rib.
The upstream-ramped rib results (Fig. 14), however, are in
line with the measured values as well as confirming the Col-
burn analogy of lower pressure drops corresponding to lower
heat transfer coefficients.
5 Conclusions
Downstream- and upstream-ramped ribs, simulating the
changes in rib shape due to the accumulation of foreign par-
ticles such as sand in front or on the back side of the ribs,
are studied both experimentally and numerically and results
are compared with the square rib results in the presence of
no foreign particles. Major conclusions of this study were:
1) There is a substantial decrease of about 43% in pres-
sure drop (friction factor) for the ramped ribs. The gradual
change of rib-roughened surface geometry, compared to the
square ribs, is responsible for this decrease.
2) Along with the decrease in pressure drop, there is
a deficit of about 28% in heat transfer coefficients for the
ramped ribs.
3) Numerically-obtained results using the realizable k−
ε turbulence model with standard wall functions are in rea-
sonable agreement with the measured results.
6 REFERENCES
[1] P. Ligrani, 2013, ”Heat Transfer Augmentation Technologies for
Internal Cooling of Turbine Components of Gas Turbine Engines,”
International Journal of Rotating Machinery, pp. 138-169.
[2] Burggraf, F., 1970, ”Experimental Heat Transfer and Pressure
Drop with Two Dimensional Turbulence Promoters Applied to Two
Opposite Walls of a Square Tube”, American Society of Mechanical
Engineers, Augmentation of Convective Heat and Mass Transfer,
Edited by A.E. Bergles and R.L. Webb, pp. 70-79.
[3] Webb, R.L., Eckert, E.R.G. and Goldstein, R.J., 1971, Heat
Transfer and Friction in Tubes with Repeated- Rib-Roughness, In-
ternational Journal of Heat Mass Transfer, Vol. 14, pp. 601-617.
[4] Han, J.C., Glicksman, L.R., and Rohsenow, W.M., 1978, An
Investigation of Heat Transfer and Friction for Rib Roughened Sur-
faces, International Journal of Heat and Mass Transfer, Vol. 21, pp.
1143-1156.
[5] Metzger, D.E., Fan, C.S., and Pennington, J.W., 1983,Heat
Transfer and Flow Friction Characteristics of Very Rough Trans-
verse Ribbed Surfaces With and Without Pin Fins, Proceedings of
the ASME-JSME Thermal Engineering Joint Conference, Vol. 1,
pp. 429-436.
[6] Han, J.C., 1984, Heat Transfer and Friction in Channels with
Two Opposite Rib-Roughened Walls, Journal of Heat Transfer, Vol.
106, No. 4, pp. 774-781.
9
[7] Han, J.C., Park, J.S., and Lei, C.K., 1985, Heat Transfer En-
hancement in Channels With Turbulence Promoters, Journal of En-
gineering For Gas Turbines and Power, Vol. 107, No. 1, pp. 628-
635.
[8] Metzger, D.E., Vedula, R.P., and Breen, D.D., 1987, The Ef-
fect of Rib Angle and Length on Convection Heat Transfer in
Rib-Roughened Triangular Ducts, Proceedings of the ASME-JSME
Thermal Engineering Joint Conference, Vol. 3, pp. 327-333.
[9] Metzger, D.E., Vedula, 1987, Heat Transfer in Triangular Chan-
nels with Angled RoughnessRibs on Two Walls, Experimental Heat
Transfer, Vol. 1, pp. 31-44.
[10] Chandra, P.R., 1987, Effect of Rib Angle on Local Heat/Mass
Transfer Distribution in a Two Pass Rib- Roughened Channel,
American Society of Mechanical Engineers Paper 87-GT-94.
[11] Chandra, P.R. and Han, J.C., 1989, Pressure Drop and Mass
Transfer in Two-Pass Ribbed Channels, Journal of Thermophysics,
Vol. 3, No. 3, pp. 315-319.
[12] Han, J.C., Zhang, Y.M., and Lee, C.P., 1992, Influence of Sur-
face Heat Flux Ratio on Heat Transfer Augmen- tation in Square
Channels with Parallel, Crossed, and V-shaped Angled Ribs, ASME
J. Turbomachinery, Vol. 114, pp. 872-880.
[13] Zhang, Y.M., Gu, W.Z. and Han, J.C., 1994, Heat Transfer and
Friction in Rectangular Channels with Ribbed or Ribbed-Grooved
Walls, Journal of Heat Transfer, Vol. 116, No. 1, pp. 58-65.
[14] Dutta, S. and Han, J.C., 1994, Effect of Model Orientation
on Local Heat Transfer in a Rotating Two-Pass Smooth Triangular
Duct, ASME Winter Annual Meeting.
[15] Taslim, M.E. and Spring, S.D., 1994,Effects Turbulator Profile
and Spacing Have on Heat Transfer and Friction in a Channel, J.
Thermophysics Heat Transfer, Vol. 8, No. 3, pp. 555-562.
[16] Taslim, M.E., Li, T. and Kercher, D.M., 1996, ”Experimental
Heat Transfer and Friction in Channels Roughened with Angled,
V-Shape and Discrete Ribs on Two Opposite Walls,” J. Turboma-
chinery, Vol. 118, pp. 20-28.
[17] Taslim, M.E., Li, T. and Spring, S.D., 1997, Measurement of
Heat Transfer Coefficients and Friction Factors in Rib-Roughened
Channels Simulating Leading-Edge Cavities of a Modern Turbine
Blade, J. Turbomachinery, Vol. 119, pp. 411-419.
[18] Arts, T., Rau, G., Cakan, M., Vialonga, J., Fernandez, D.,
Tarnowski, F. and Laroche, E., 1997, ”Experimental and numeri-
cal investigation on flow and heat transfer in large-scale, turbine
cooling, representative, rib-roughened channels,” Journal of Power
and Energy, vol. 211, pp. 263-272.
[19] Gupta, A., SriHarsha, V., Prabhu, S. and Vedula, R., 2008,
”Local heat transfer distribution in a square channel with 90◦ con-
tinuous, 90◦ saw tooth profiled and 60◦ broken ribs,” Experimental
Thermal and Fluid Science, vol. 32, pp. 997-1010.
[20] Taslim, M.E. and Spring, S. D., 1994, ”Effects of Turbulator
Profile and Spacing on Heat Transfer and Friction in a Channel,”
Journal of Thermophysics and Heat Transfer, vol. 8, no. 3, pp.
555-562.
[21] Gao, X. and Sunden, B., 2001, ”Heat trasfer and pressure drop
measurements in rib-roughened rectangular ducts,” Experimental
Thermal and Fluid Science, vol. 24, pp. 25-34.
[22] Lu, B. and Jiang, P.-X.,2006, ”Experimental and numerical
investigation of convection heat transfer in a rectangular channel
with angled ribs,” Experimental Thermal and Fluid Science, vol.
30, p. 513521.
[23] Peng, W., Jiang, P.-X., Wang, Y.-P. and Wei, B.-Y.,2011, ”Ex-
perimental and numerical investigation of convection heat transfer
in channels,” Applied Thermal Engineering, vol. 31, pp. 2702-
2708.
[24] Promvonge, P. and Thianpong, C., 2008, ”Thermal perfor-
mance assessment of turbulent channel flows over different shaped
ribs,” International Communications in Heat and Mass Transfer,
vol. 35, p. 13271334.
[25] Kamali, R. and Binesh, A., ”The importance of rib shape ef-
fects on the local heat transfer and flow friction characteristics of
square ducts with ribbed internal surfaces,” International Commu-
nications in Heat and Mass Transfer, vol. 35, p. 10321040, 2008.
[26] Taslim, M.E. and Liu, H., 2005, ”A Combined Numerical and
Experimental Study of Heat Transfer in a Roughened Square Chan-
nel with 45◦ Ribs,” International Journal of Rotating Machinery,
vol. 1, pp. 60-66, 2005.
[27] Metzger, D.E, Chyu, M.K. and Bunker, R.S., 1988,”The Con-
tribution of On-Rib Heat Transfer Coefficients to Total Heat Trans-
fer from Rib-Roughened Surfaces”, Transport Phenomena in Rotat-
ing Machinery, Edited by J.H. Kim, Hemisphere Publishing Co.
[28] Taslim, M.E. and Wadsworth, C.A., 1997, ”An Experimen-
tal Investigation of the Rib Surface-Averaged Heat Transfer Coef-
ficient in a Rib-Roughened Square Channel,” J. Turbomachinery,
Vol. 119, pp. 381-389.
[29] Korotky, G.J. and Taslim, M.E., 1998, Rib Heat Transfer Coef-
ficient Measurements in a Rib-Roughened Square Passage, J. Tur-
bomachinery, Vol. 120, No. 2, pp. 376-385.
[30] Taslim, M.E., and A. Lengkong, 1998, ”45◦ Staggered Rib
Heat Transfer Coefficient Measurements in a Square Channel,” J.
Turbomachinery, Vol. 120, pp. 571-580.
[31] Taslim, M.E., and G.J. Korotky, 1998, ”Low-Aspect-Ratio Rib
Heat Transfer Coefficient Measurements in a Square Channel,” J.
Turbomachinery, Vol. 120, pp. 831-838.
[32] Taslim, M.E., and A. Lengkong, 1999, 45◦ Round-Corner Rib
Heat Transfer Coefficient Measurements in a Square Channel, J.
Turbomachinery, Vol. 121, pp. 1-9.
[33] El-Husayni, H.A., Taslim, M.E., and Kercher, D.M., 1994, An
Experimental Investigation of Heat Transfer Coefficients in a Span-
wise Rotating Channel With Two Opposite Rib-Roughened Walls,
J. Turbomachinery, Vol. 113, pp. 75-82.
[34] Kline, S.J. and McClintock, F.A., 1953, Describing Uncer-
tainty in Single-Sample Experiments, Mechani- cal Engineering,
Vol. 75, January, pp. 3-8.
10