International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

Contents lists available at ScienceDirect

International Journal of Heat and Mass Transfer

journal homepage: www.elsevier.com/locate/ijhmt

Theory and experimental validation of cross-flow micro-channel heat

exchanger module with reference to high Mach aircraft gas turbine engines
Robert Nacke, Brittany Northcutt, Issam Mudawar ⇑
Boiling and Two-Phase Flow Laboratory (BTPFL), Purdue University International Electronic Cooling Alliance (PUIECA), Mechanical Engineering Building,
585 Purdue Mall, West Lafayette, IN 47907, USA
Rolls-Royce Purdue University Technology Center (UTC) in High Mach Propulsion, Maurice Zucrow Laboratories, 500 Allison Road, Purdue University, West Lafayette, IN 47907, USA

a r t i c l e i n f o a b s t r a c t

Article history: This study explores the design, analysis, and performance assessment of a new class of heat exchangers
Received 18 February 2010 intended for high Mach aircraft gas turbine engines. Because the compressor air that is used to cool tur-
Received in revised form 8 October 2010 bine blades and other components in a high Mach engine is itself too hot, aircraft fuel is needed to precool
Accepted 8 October 2010
the compressor air, cooling is achieved with a new heat exchanger. The heat exchanger consists of a large
Available online 2 December 2010
number of miniature, closely-spaced modules. Within each module, the fuel flows through a series of par-
allel micro-channels, while the air flows externally over rows of short, straight fins perpendicular to the
Keywords:
direction of fuel flow. A theoretical model was developed to predict the thermal performance of the mod-
Micro-channel
Heat exchanger
ule for various operating conditions. To confirm the accuracy of the model, a single module was con-
Turbine engines structed and tested using water to simulate the aircraft fuel. The theoretical model was used to
High Mach predict the air temperature drop, water temperature rise, and heat transfer rate for each fluid stream.
Comparisons between theory and experiment show good overall agreement in exit temperatures and
heat transfer rates. This study shows the theoretical model is a reliable tool for predicting the perfor-
mance of heat exchanger modules under actual fuel and air turbine engine conditions and for the design
of aircraft heat exchangers of different sizes and design envelopes.
Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction the fuel sides individually. Most heat exchanger designs involve
flow of compressor bleed air across a series of circular tubes carry-
Supersonic gas turbine engines pose unique thermal manage- ing the cooler fuel, similar to what is typically encountered in a
ment challenges not encountered in subsonic engines. In both conventional shell-and-tube heat exchanger. Huang et al. [3] used
cases, compressor bleed air is used to cool various downstream metal foam to enhance heat transfer on the airside. Their tests
engine components such as turbine blades. For subsonic engines, revealed that foam is too heavy for turbine engines and recom-
the compressor bleed air is cool enough to be used directly to cool mended that other designs with better heat capacity-to-weight
the turbine blades [1]. However, for supersonic engines, the com- ratios be developed. Kibbey [4] investigated the merits of enhanc-
pressor bleed air temperature is quite high and a heat exchanger ing heat transfer on the fuel side by fitting the circular fuel tube
is needed to precool the air before it is introduced to the turbine with an inner coaxial tube that featured holes to supply jets of fuel
blades. against the wall of the outer tube. While jet impingement did
Cooling of the compressor bleed air can be achieved by rejecting produce high convective heat transfer coefficients, interactions
heat to either the engine fan’s bypass air or the fuel [2,3]. In some between the jets and the spent fuel flow in the annulus between
cases, depending on engine cycle requirements and thermal man- the two tubes, as well as the need to utilize a large number of clo-
agement architecture, using fuel as the heat sink may allow a more sely spaced jets, rendered this design impractical and inefficient.
compact and lightweight heat exchanger than using air. Herring and Heister [5] also investigated heat transfer enhance-
Previous efforts to optimize air–fuel heat exchanger design have ment on the fuel side by inserting wire coils inside the fuel tubes.
been focused on investigating heat transfer schemes for the air and The inserts provided appreciable fuel-side heat transfer enhance-
ment with JP-10 jet fuel, but still further enhancement is desired.
Developing a fuel–air heat exchanger that is more compact,
⇑ Corresponding author at: Boiling and Two-Phase Flow Laboratory (BTPFL),
lightweight and effective than current cross-flow topologies is
Purdue University International Electronic Cooling Alliance (PUIECA), Mechanical
Engineering Building, 585 Purdue Mall, West Lafayette, IN 47907, USA. Tel.: +1
the primary motivation for this work. The new heat exchanger is
(765) 494 5705; fax: +1 (765) 494 0539. comprised of many modules that can be arranged to suit a variety
E-mail address: [email protected] (I. Mudawar). of engine envelopes. Fig. 1(a) illustrates the basic construction of a

0017-9310/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijheatmasstransfer.2010.10.028
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1225

Nomenclature

A airside parameter defined in Eq. (5a) Rc,2 surface 2 base convective resistance
Ac,csf fuel (or water) fin cross-sectional area Rc,3 surface 3 base convective resistance
Ah,csf air fin cross-sectional area Rcond module’s outer wall conduction resistance
Ah,f airside finned area Rc,sw1 first fuel (or water) sidewall resistance
Ah,uf airside unfinned area Rc,sw2 second fuel (or water) sidewall resistance
b fuel-side (or waterside) parameter defined in Eq. (5b) Rh,1 airside resistance
cp specific heat at constant pressure Rh,4 airside base resistance
Dc,eq fuel-side (or waterside) micro-channel hydraulic diam- Rtot total (equivalent) resistance
eter Rec fuel-side (or waterside) Reynolds number
Dh,eq airside hydraulic diameter on back of module Reh,b airside Reynolds number on back of module
Hc,ch fuel-side (or waterside) micro-channel height Reh,f airside Reynolds number on finned side of module along
Hc,w module’s outer wall thickness fins
Hh,f airside fin height Reh,uf airside Reynolds number on finned side of module be-
Hh,fe corrected airside fin height tween fins
c
h average fuel-side (or waterside) heat transfer coefficient s1 flow clearance above airside fin tip (0.0635 mm)

h average airside heat transfer coefficient on back of mod- s2 flow clearance along back of module (0.0635 mm)
h;b
ule T temperature

h average airside fin heat transfer coefficient T1 surface 1 temperature
h;f

h average airside heat transfer coefficient along surface T2 surface 2 temperature
h;uf
between fins T3 surface 3 temperature
kc thermal conductivity of fuel (or water) T4 surface 4 temperature
kh thermal conductivity of air Tc fuel (or water) temperature
ks thermal conductivity of heat exchanger module Tc,in,exp measured waterside inlet temperature
L length of module in direction of fuel (or water) flow T c;o mean outlet fuel (or water) temperature
Lh,f airside fin length Tc,o,exp measured waterside mean outlet temperature
mc fuel-side (or waterside) fin parameter Tc,o,th theoretical waterside outlet temperature
mh airside fin parameter Th air temperature
m_c fuel (or water) mass flow rate Th,in,exp measured airside inlet temperature
m_h air mass flow rate T h;o mean outlet air temperature
Nc,ch number of fuel-side (or waterside) micro-channels Th,o,exp measured airside mean outlet temperature
Nh,f number of airside fin rows Th,o,th theoretical airside mean outlet temperature
Nh,r number of small fins in airside fin row U overall heat transfer coefficient
Nuc average fuel-side (or waterside) Nusselt number Vc fuel-side (or waterside) mean velocity
Nuh;b average airside Nusselt number on back of module Vh airside mean velocity
Nuh;f average airside Nusselt number on finned side of mod- W width of module in direction of air flow
ule along fins Wc,ch fuel-side (or waterside) micro-channel width
Nuh;uf average airside Nusselt number on finned side of mod- Wc,w fuel-side (or waterside) micro-channel wall thickness
ule between fins Wh,ch airside channel width
Pc,f fuel (or water) fin perimeter Wh,f airside fin width
Ph,f air fin perimeter x fuel (or water) direction coordinate
Pr Prandtl number x0 dimensionless fuel (or water) direction coordinate
q heat exchanger module’s heat transfer rate y air direction coordinate
q00 heat flux across module y0 dimensionless air direction coordinate
qa,b back surface heat transfer rate
qc,exp measured water side heat transfer rate Greek symbols
qc,th theoretical water side heat transfer rate dh,f airside finned boundary layer thickness
qc,sw fuel-side (or waterside) micro-channel sidewall heat dh,uf airside unfinned boundary layer thickness
transfer rate gh,f airside fin efficiency
qc,2 surface 2 convective heat transfer rate h local dimensionless temperature difference between air
qc,3 surface 3 convective heat transfer rate and fuel (or water)
qh,exp measured airside heat transfer rate l dynamic viscosity
qh,th theoretical airside heat transfer rate m kinematic viscosity
qh,1 surface 1 heat transfer rate u ratio of mean to initial temperature difference
q00s heat flux across module’s outer wall
RA thermal resistance of branch A of module’s equivalent Subscripts
resistance c cold fuel stream (or simulated water stream)
RB thermal resistance of branch B of module’s equivalent h hot air stream
resistance s solid surface

single module. The fuel is routed through a series of parallel micro- compact and lightweight design, these devices greatly increase
or mini-channels formed in a thin monolithic metallic structure. both heat transfer area to volume ratio and convective heat trans-
The fuel-side construction of the module is reminiscent of micro- fer coefficient on the fuel side. The new module design also en-
channel heat sinks intended for cooling of high-heat-flux electronic hances heat transfer on the airside with the use of short straight
chips as well as laser and radar devices [6]. Aside from their highly fins as illustrated in Fig. 1(a). The fins are formed in rows and
1226 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

Fig. 1. (a) Cross-flow micro-channel air-to-fuel heat exchanger module. (b) Stacking of modules in radial in-flow turbine engine heat exchanger.

aligned with the airflow but perpendicular to the direction of the mance of the heat exchanger module using actual aircraft fuel,
fuel flow. These fins enhance airside heat transfer in two ways. the model validation was assessed by substituting the fuel with
First, they greatly increase heat transfer area compared to a bare water. The experimental facility developed for the study, the heat
surface. Second, by using many short fins as opposed to continuous exchanger module construction and instrumentation, and experi-
fins, they produce large heat transfer coefficients by reinitiating mental methods used are described in detail. Finally, experimental
airside boundary layer development. results and comparisons with the analytical/numerical predictions
Another key advantage of the new module is that it allows the are presented.
air–fuel heat exchanger to be configured in variety of design enve-
lopes, such as annular or rectangular, depending on volume,
weight or other constraints set by the engine manufacturer. 2. Experimental methods
Fig. 1(b) shows an annular design, in which the air is introduced
radially inwards across two of radial stages of closely spaced 2.1. Test facility
cross-flow micro-channel modules. Not shown in this figure are
the shared inlet and outlet headers for the modules. The test facility developed for this project incorporated an air
The present study represents the first phase in the design of the loop and a water loop to simulate, respectively, the air and fuel
air–fuel heat exchanger. A detailed analytical/numerical model is flows that the heat exchanger would encounter in a supersonic tur-
constructed to characterize the thermal performance of the heat bine engine. These loops are illustrated schematically in Fig. 2(a).
exchanger module. Given the complexity in testing the perfor- In the air loop, compressed air flows through a series of filters to

Fig. 2. (a) Schematic of flow loop and schematic (b) and photo (c) of flow straightener and test section.
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1227

remove any impurities such as water, oil, solid particulates. The air 2.2. Test module
passes through a venturi flow meter to measure the air flow rate.
Two different sonic venturis are used to provide broad coverage Shown in Fig. 4, the test module was fabricated from stainless
of flow rates. The air stream is heated by an inline heater to simu- steel and measures 76.2 mm long and 15.24 mm wide. The airside
late air heating in a turbine engine, albeit at lower temperatures. A fins cover one side of the test module excepting two 5.08 mm end
solid-state controller with an accuracy of ±2 °C is used to regulate regions of the same surface to allow a secure press fit into the insu-
the air temperature to the desired level between room tempera- lating PEEK housing. Since the two end regions do not contribute to
ture and 260 °C. Exiting the heater, the air enters a flow straight- the heat transfer between the air and the water, the module simu-
ener to provide uniform flow at the inlet to the heat exchanger lates a 66.04 mm heat exchanger module in a turbine engine.
test module. The heated air converges through a nozzle before As illustrated in Fig. 4(a) and (b), the module’s airside consists
entering the test section containing the test module as illustrated of 65 rows of fins. Each fin row consists of seven fins, six of which
in Fig. 2(b) and depicted in Fig. 2(c). measure 1.524 mm while the middle fin measures 2.032 mm. The
The water flow loop provides water flow that is regulated to the airside fin rows are angled to allow for a tighter packaging arrange-
desired flow rate and temperature as it enters the test module. ment in an annular turbine heat exchanger design (see Fig. 1(b)).
Illustrated in Fig. 2(a), the pump, reservoir, and water-to-air heat Fin height varies from 0.635 mm at the high edge to 0.127 mm at
exchanger are all parts of an integral unit acquired from Lytron the lower, forming a 2.29° angle with the surface of the test mod-
Inc. The filtered water is passed through one of three flow meters ule. As shown in Fig. 4(b), the waterside consists of 26 of 0.254 mm
before entering the test module. wide by 0.762 mm high micro-channels running the middle
As shown in Fig. 3(a), the heat exchanger test module is con- 66.04 mm of the test module’s length.
tained in a PEEK plastic housing that provides thermal insulation While a module used in a gas turbine engine heat exchanger
for both the air and the water flows. Fig. 3(b) shows the location would likely be made of a nickel alloy to withstand the engine’s
of the micro-channel heat exchanger module inside the PEEK hous- high temperatures, the test module used in the present study
ing and illustrates the paths of the air and the water streams in and was made of stainless steel because of its somewhat similar
out of the module. thermal conductivity and relative ease of machining compared to

Fig. 3. (a) Exploded diagram of air nozzle and test module. (b) Air and water flow paths inside test module housing (middle portion of housing is removed to show micro-
channel test module).
1228 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

Fig. 4. Test module (a) airside geometry and (b) fuel-side and airside geometries. (c) Various views of test module.

nickel alloys. The test module was fabricated from two flat stain- 2.3. Measurement accuracy
less steel plates. One was used to form a cover plate and the second
the main body of the test module. The micro-channels were Type K thermocouples were used throughout the system and
formed by holding the outer ends of the main plate in an alumi- read by an Omega temperature indicator with an accuracy of
num fixture and cutting 0.762 mm deep parallel grooves using a ±0.5 °C. Temperatures were measured upstream of the venturi as
series of miniature saw blades attached to an arbor, separated by well as upstream and downstream of both the air and water
thin spacers. The blade thickness equaled the width of the micro- streams in the test module.
channels, and the spacer thickness the width of solid wall between Pressures were measured by means of three absolute and three
micro-channels, both 0.254 mm. A similar technique was used to differential Druck pressure transducers, with accuracies of ±0.04%
form the airside fins. and ±0.25%, respectively, of the full-scale pressure reading. The
With both stainless steel plates fabricated, the cover plate was transducers were used in tandem to record absolute pressure and
soldered onto the base plate. The sides of the assembled module pressure drop at three locations: across the venturi, across the air-
were rounded off to allow for more streamlined airflow around side of the test module, and across the waterside of the test module.
the module. It is important to note that the solder was applied The two sonic venturis used to measure the air flow rates corre-
to the base plate before the micro-channels were cut, leaving only spond to test module air velocities of 3–33 m/s. These meters are
the appropriate amount of solder on top of the fins. This pre- manufactured to ASME standards and possess an accuracy of ±1%
vented excess solder from running off and filling the micro-chan- of reading.
nels after machining. Fig. 4(c) shows photos of the completed test Three waterside flow meters were used to provide the required
module. coverage over a broad range of flow rates. They had accuracies of
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1229

 
±6%, ±3%, and ±3%, respectively, over flow rate ranges of 0–441, 0– _c
m @T c
dq ¼ þ cp;c dx dy; ð3bÞ
3784, and 0–18,930 cm3/min. W @x
where m _ h and m_ c are the mass flow rates of the air stream and the
3. Heat exchanger model fluid stream, respectively, for the entire module. Eqs. (3a) and (3b)
are based on the assumption that the flow rates are uniformly dis-
3.1. Overall heat exchanger modeling approach tributed for both fluid streams. These equations are also based on
the assumption that all the heat lost by the air across the differen-
The approach used to model the thermal performance of the tial area is absorbed by the fluid; i.e., they assume both negligible
air–fuel heat exchanger consists of first determining the tempera- heat loss and negligible axial conduction effects.
ture distributions for both the air and fuel streams across a single Mason [7] obtained a series solution for the dimensionless tem-
module. The method used here uses a mean overall heat transfer perature difference h between the air and the fluid as defined in
coefficient, U, between the air and the fluid streams that is as- Fig. 5 by combining and differentiating Eqs. (2), (3a), (3b), followed
sumed constant everywhere across the module. This coefficient is by the use of Laplace transforms.
a function of the convective heat transfer coefficients for the air
and the fluid, as well as the conduction resistances associated with 0 0 X
1 0
ðabx y0 Þn
hðx0 ; y0 Þ ¼ eðay þbx Þ ; ð4Þ
the micro-channel plate and the air fins. Averaging the effects of n¼0 ðn!Þ2
the tapered air flow on the finned side of the test module will be
discussed later. where
Fig. 5 shows a schematic of a single module and the correspond-
ing coordinate system and nomenclature. The complex geometrical UWL
a¼ ð5aÞ
features of the module are purposely omitted for now to help ex- _ h cp;h
m
plain the differential model used to determine the temperature
distributions of the two fluids. First, the temperatures of the hot and
air stream and cold fluid stream are defined, respectively, as UWL
b¼ : ð5bÞ
T h ¼ T h ðx; yÞ ð1aÞ _ c cp;c
m

Integrating Eq. (2) over the entire surface area WL of the module
and
yields an expression for the total heat transfer rate for the module,
T c ¼ T c ðx; yÞ: ð1bÞ Z W Z L
q¼U ðT h  T c Þdxdy
Under steady-state conditions, the rate of heat transfer between y¼0 x¼0
Z 1 Z 1     
the air and the fluid across a heat exchanger differential area dx dy Th  Tc x y
¼ UWL½T h ð0; 0Þ  T c ð0;0Þ d d :
is given by 0 0 T h ð0;0Þ  T c ð0; 0Þ L W
ð6Þ
dq ¼ UðT h  T c Þdx dy: ð2Þ
Substituting the dimensionless definitions from Fig. 5 in Eq. (6)
This rate is equal to both the differential sensible heat loss by the air gives
and the differential sensible heat gain by the fluid across dx dy, Z 1 Z 1
  q 0 0
_h
m @T h /¼ ¼ hdx dy ; ð7Þ
dq ¼  cp;h dx dy ð3aÞ UWL½T h ð0; 0Þ  T c ð0; 0Þ 0 0
L @y
where / is the ratio of mean outlet to inlet temperature difference.
and Using the solution for h from Eq. (4), Eq. (7) can be expressed as

Fig. 5. Schematic and nomenclature for cross-flow micro-channel heat exchanger module with uniform inlet fluid temperatures.
1230 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

Z Z Z Z " #
1 1 1 1 X
1 0
ðabx y0 Þn T h;o ¼ T h ð0; 0Þ  a½T h ð0; 0Þ  T c ð0; 0Þ/ ð17aÞ
0 0 ðay0 þbx0 Þ 0 0
/¼ hdx dy ¼ e 2
dx dy
0 0 0 0 n¼0 ðn!Þ and
(Z Z " # )
X
1 1 1 0
ðabx y0 Þn
¼
0
eðay þbx Þ dx dy
0 0 0
: ð8Þ T c;o ¼ T c ð0; 0Þ þ b½T h ð0; 0Þ  T c ð0; 0Þ/: ð17bÞ
n¼0 0 0 ðn!Þ2
In summary, the performance parameters of the heat exchanger
The first integral of Eq. (8) can be rearranged as module consist of total heat transfer rate, q, which can be deter-
Z " # Z mined from Eq. (15), outlet temperature of the hot stream, T h;o ,
0 0
1
ðabx y0 Þn ðay0 þbx0 Þ 0
1
ðaby Þn 0 0 0 from Eq. (17a), and outlet temperature of the cold stream, T c;o , from
2
e dx ¼ 2
x0n ebx eay dx
0 ðn!Þ 0 ðn!Þ Eq. (17b). Calculating these three parameters requires determining
0 Z a and b from Eqs. (5a) and (5b), respectively, and u from Eq. (14).
ðaby Þn 0
1
0 0
¼ 2
eay x0n ebx dx : ð9Þ
ðn!Þ 0 3.2. Overall heat transfer coefficient
The solution for the integral in Eq. (9) can be found using integra-
tion tables [8]: The only unknown in the parameter a in Eq. (5a) and b in Eq.
1 (5b) is the overall heat transfer coefficient, U. This parameter can
Z 1 Xn 
0n bx0 0 bx0 r n!x0nr  be determined by using a thermal resistance network using the
x e dx ¼ e ð1Þ  heat exchanger module geometry illustrated in Fig. 6(a). On one
0 r¼0 ðn  rÞ!ðbÞrþ1  0
" # surface of the module, the air flows over the external fins as well
X
n
ð1Þn n! n! as along the surface between the fins. The air also flows over the
¼ eb ð1Þr  1
r¼0 ðn  rÞ!ðbÞrþ1 ðbÞnþ1 back surface of the module. The fluid travels through the micro-
" # channels. Details of the airside and fluid-side boundaries are given
ð1Þn n! b X n nr
b in Fig. 6(b) and (c). The number of airside fin rows, Nh,f, and fuel-
¼ nþ1
e 1 ð10Þ
ðbÞ r¼0
ðn  rÞ! side micro-channels, Nc,ch, can be found using the following
relations
Substituting the final answer from Eq. (10) into Eq. (8) and rear-
ranging terms gives L
Nh;f ¼ ð18aÞ
(Z " # ) W h;ch þ W h;f
X
1 1 0 n n X
n nr
ðaby Þ 0 ð1Þ n! b 0
/¼ eay eb  1 dy and
n¼0 0 ðn!Þ2 ðbÞnþ1 ðn  rÞ!
r¼0
( " #Z ) W
X X nr
1
abÞn ð1Þn n! n
b 1
0 0 Nc;ch ¼ ; ð18bÞ
¼ eb 1 y0n eay dy : ð11Þ W c;ch þ W c;w
n¼0 ðn!Þ2 ðbÞnþ1 r¼0
ðn  rÞ! 0
respectively.
The integral with respect to y0 in Eq. (8) takes the same form as the For the finned airside, the heat is transferred through finned and
integral with respect to x0 and therefore can be solved using the unfinned portions of the surface. Introducing the definition of air-
same method. side fin efficiency [9], the heat transfer rate for the finned surface
Z " # (surface 1) of the module can be expressed as
1
0 0 ð1Þn n! X
n
anr
y0n eay dy ¼ ea 1 : ð12Þ
0 ðaÞnþ1 r¼0
ðn  rÞ! qh;1 ¼ q00h;1 ðWLÞ ¼ qh;f þ qh;uf
 A ðT  T 1 Þ þ N h
¼ Nh;f gh;f h 
Substituting Eq. (12) into Eq. (11) and simplifying, h;f h;f h h;f h;uf Ah;uf ðT h  T 1 Þ
( " #" #)  A þh  A ÞðT  T 1 Þ;
X1
ðabÞn ð1Þn n! ð1Þn n! a X n
anr Xn
b
nr ¼ N ðg h
h;f h;f h;f h;f h;uf h;uf h ð19Þ
b
/¼ e  1 e  1
n¼0 ðn!Þ2 ðaÞnþ1 ðbÞnþ1 r¼0
ðn  rÞ! r¼0
ðn  rÞ! where
(" #" #)
1 X X X
1 n n nr
anr b Ah;f ¼ Nh;r ½2ðW h;f þ Lh;f ÞHh;f þ W h;f Lh;f ; ð20Þ
¼ 1  ea 1  eb :
ab n¼0 r¼0
ðn  rÞ! r¼0
ðn  rÞ!
ð13Þ Ah;uf ¼ ðW h;f þ W h;ch ÞW  Nh;r W h;f Lh;f ; ð21Þ

Substituting k for n  r yields the final solution gh,f is the airside fin efficiency, hh;f and hh;uf are the average airside
heat transfer coefficients for the fin and base surfaces, respectively,
" #" #
1 X 1 Xn
ak Xn
b
k Ah,f and Ah,uf are the areas of the fin and base surfaces in a single fin
a b
/¼ 1e 1e : ð14Þ row, respectively, and Nh,r is the number of airside fins in each row
ab n¼0 k¼0
k! k¼0
k!
of fins.
Rearranging Eq. (19) yields
Rearranging Eq. (7),
Th  T1
q ¼ UWL½T h ð0; 0Þ  T c ð0; 0Þ/: ð15Þ qh;1 ¼ n o; ð22Þ
1
 A þh
Nh;f ½gh;f h 
h;f h;f h;uf Ah;uf 
Note that the module’s total heat transfer rate must also equal
the sensible heat lost by the hot stream or gained by the cold which is used to determine the airside resistance as
stream.
1
Rh;1 ¼  A þh  A  : ð23Þ
_ h cp;h ½T h ð0; 0Þ  T h;o  ¼ m
q¼m _ c cp;c ½T c;o  T c ð0; 0Þ; ð16Þ Nh;f ½gh;f h h;f h;f h;uf h;uf

where T h:o and T c:o are the mean outlet temperatures of the hot For the fluid side, the heat must first travel through the outer
stream and the cold stream, respectively. Combining Eqs. (15) and metallic wall of the micro-channels, which has a thickness Hc,w
(16) and introducing the definitions of a and b from Eqs. (5a) and and conductivity ks as illustrated in Fig. 6(c). Assuming the inner
(5b), respectively, yield the following relations for the mean outlet surface temperature of the outer wall, T2, is uniform, and the con-
temperatures. duction across the outer wall one-dimensional,
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1231

Fig. 6. Schematics and nomenclature of (a) heat exchanger module, (b) finned airside boundary and (c) fuel micro-channels and unfinned airside boundary.

T1  T2 T1  T2 Pc;f ¼ 2L; ð27aÞ

q00s ðLWÞ ¼ ks ðLWÞ ¼n o; ð24Þ
Hc;w Hc;w
ks ðLWÞ
Ac;csf ¼ W c;w L ð27bÞ
Eq. (24) yields the following relation for thermal resistance of the and
outer wall,
sffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Hc;w c P
h c;f
Rcond ¼ : ð25Þ mc ¼ : ð27cÞ
ks ðLWÞ ks Ac;csf
The micro-channel sidewalls can be approximated as fins. Using Eq. (26) may be rearranged as
a prescribed fin tip temperature case (see [9]), the rate of heat flow
through the sidewalls can be broken into the difference between T2  Tc
qc;sw ¼ 8 9
two separate heat flow terms. < =
h i qffiffiffiffiffiffiffiffiffi
ffi 1 
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi coshðmc Hc;ch Þ  TT 32 T c
T c
:N phffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 c P ks A
ffi
coth
 P
hc c;f
H
;
c;ch c;f c;csf c;ch
qc;sw ¼ N c;ch h c P ks A ðT 2  T c Þ ks Ac;csf
c;f c;csf
sinhðmc Hc;ch Þ T3  Tc
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 8 9 ð28Þ
¼ N c;ch h c P ks A ðT 2  T c Þ cothðmc H Þ  N
c;f c;csf c;ch c;ch
< =
ffi qffiffiffiffiffiffiffiffiffi ffi 
1
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi :N phffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  P
hc c;f ;
 P kA
 h c P ks A ðT 3  T c Þ csc hðmc H Þ; ð26Þ c c;ch s csc h
c;f H
c;csf ks Ac;csf c;ch
c;f c;csf c;ch

where to yield the following two expressions for fluid sidewall resistances,
1232 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

1
Rc;sw1 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffi  ð29aÞ
c P
h

Nc;ch hc P c;f ks Ac;csf coth c;f
H
ks A c;ch
c;csf

and
1
Rc;sw2 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi rffiffiffiffiffiffiffiffiffiffi : ð29bÞ
 P
h
P kA
Nc;ch h c c;f
c c;f s c;csf csc h ks A
H c;ch
c;csf

Aside from the sidewalls, there is direct convective heat transfer

from surface 2 of the micro-channel to the fluid.

c ðW LÞðT 2  T c Þ ¼ n ðT 2  T c Þ o ;
qc;2 ¼ Nc;ch h ð30Þ
c;ch
1
c ðW
Nc;ch h c;ch LÞ

which yields the following expression for base convective resistance

1
Rc;2 ¼  c ðW LÞ : ð31Þ
Nc;ch h c;ch

Similarly, there is direct convection for surface 3 of the micro-chan-

nel to the fluid, which is associated with a similar convective
resistance.

c ðW LÞðT 3  T c Þ ¼ n ðT 3  T c Þ
qc;3 ¼ Nc;ch h c;ch o ð32Þ
1
c ðW
Nc;ch h c;ch LÞ

and

1
Rc;3 ¼  c ðW LÞ : ð33Þ
Nc;ch h c;ch

The heat must flow from the opposite hot airside through the
other wall of the micro-channel, whose conductive resistance is Fig. 7. Equivalent thermal resistance network representing entire micro-channel
identical to that given by Eq. (25). module.
On the backside of the heat exchanger module, heat is trans-
ferred from the air to the back surface by convection.
Rc;cw2 Rc;3
 T4Þ RB ¼ Rh;4 þ Rcond þ : ð38cÞ
qa;b  ðWLÞðT  T 4 Þ ¼ ðT
¼h nh o; ð34Þ Rc;cw2 þ Rc;3
h;b h
1
 ðWLÞ
h h;b

which yields the following expression for convective resistance 3.3. Determination of airside and fluid side heat transfer coefficients
and fin efficiencies
1
Rh;4 ¼  : ð35Þ
hh;b ðWLÞ Eq. (37) shows that calculating the overall heat transfer coeffi-
The total heat transfer between the air and the fluid for the entire cient requires determination of the airside and fluid-side heat
module can be related to both the module’s total thermal resis- transfer coefficients as well as fin efficiencies. To accomplish this
tance, Rtot, and the overall heat transfer coefficient, U. task, the module geometry, the air and fluid inlet conditions, and
the air and fluid properties must be initialized. In the present
Th  Tc study, water was used to simulate the fuel stream. The air and
q00 ðWLÞ ¼ ¼ UðWLÞðT h  T c Þ; ð36Þ
Rtot water properties were determined using EES [10].
which yields In the airside fin calculations, laminar flow over a flat plate is as-
sumed, based on the low Reynolds numbers associated with the
1 present application and the experimental validation study. For this
U¼ : ð37Þ
ðWLÞRtot assumption to be valid for the entire air passage between two adja-
cent rows of fins, the boundary layer thickness must be smaller
As shown in Fig. 7, the total resistance may be represented as the
than the spacing between fin rows. Otherwise, the boundary layers
equivalent of two parallel branches A and B. Each branch consists
would merge, and the air flow would resemble internal instead of
of a series of three resistances; the third of which is the equivalent
external flow. The airside fin efficiency can be determined by using
of two parallel resistances. Therefore,
the approximation for a fin with an adiabatic tip because
 W =ks 6 0.0625 for the present study. Laminar flow over a flat
h
RA RB h;f h;f
Rtot ¼ ; ð38aÞ plate is also assumed for the airside base calculations between fins.
RA þ RB
As with the airside fins, this assumption is valid if the boundary
where layers for the base and the adjacent module (or PEEK housing in
Rc;sw1 Rc;2 the experimental study) do not merge, which would change the
RA ¼ Rh;1 þ Rcond þ ð38bÞ flow from external to internal.
Rc;sw1 þ Rc;2
Unlike the finned side, the air flow along the back of the module
and (surface 4 in Fig. 6(c)) is internal, given the small back flow
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1233

clearance s2. Here, a correlation for laminar flow in a channel with

an equivalent hydraulic diameter is used.
The fuel-side convection coefficient is determined from correla-
tions for flow in a circular channel that are corrected for equivalent
hydraulic diameter of the micro-channel. For laminar flow condi-
tions of interest, two different correlations are recommended
based on Prandtl number range.
Table 1 provides all correlations or relations for the airside and
fluid-side heat transfer coefficients and fins [9] that are used to eval-
uate the overall heat transfer coefficient U. It should be noted that a
viscosity ratio term that appears in the heat transfer coefficient cor-
relations in Table 1 was set equal to unity in the present study.

Fig. 8. Outlet air temperature profiles for different water flow rates.
4. Experimental results and comparison with model predictions

As discussed earlier, water was used to simulate aircraft fuel in water is warmest. Also, notice the gradient of the airside tempera-
a high Mach gas turbine engine. Experiments were performed to ture profile is strongest near x = 0, especially near y = 0, where local
assess the performance of the module over a range of air and water heat transfer rate between the air and the water is greatest.
flow rates. The air flow rate was set by a combination of air tem- Since the airside fins are tapered, with the tallest fin at the air
peratures and pressures upstream of the venturi. The water flow inlet and becoming progressively shorter, the values that are func-
rate was measured by one of three water flow meters. tions of air fin height—such as air velocity and air fin efficiency—
Fig. 8 shows the measured outlet airside temperature profile for vary across the module. To arrive at an average value for the overall
a fixed inlet air temperature and fixed flow rate and four water heat transfer coefficient, U, for the test module and compare the
flow rates. Notice that the air temperature is lowest at x = 0 (see model predictions to the test module results, an average fin height
Fig. 5), where the water is coolest, and highest at x = L, where the is used. Since the height of the air fins changes linearly, the average

Table 1
Heat transfer coefficient and fin efficiency equations.

Module section Equation

Air flow on finned side of module along fins Heat transfer coefficient:

 L
hh;f h;f 1=2 1=3
Nuh;f ¼ ¼ 0:664Reh;f Prh
kh
V h Lh;f
where Reh;f ¼ mh ; valid for dh;f < W h;ch =2; Reh;f < 5  105 and Pr h > 0:6
Fin efficiency:
tanhðmh Hh;fe Þ
gh;f ¼
mh Hh;fe
where
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 P
h h;f h;f
Hh;fe ¼ Hh;f þ W h;f =2; mh ¼ ; Ph;f ¼ 2ðW h;f þ Lh;f Þ; Ah;csf ¼ W h;f Lh;f
ks Ah;csf

Air flow on finned side of module between fins Heat transfer coefficient:

 W
h h;uf 1=2
Nuh;uf ¼ ¼ 0:664Reh;uf Pr1=3
h
kh
where Reh;uf ¼ Vmh W valid for dh;uf < ðHh;f þ s1 Þ=2; Reh;uf < 5  105 and Prh > 0:6
h
Fluid flow in micro-channels Heat transfer coefficient:
 
D
c Dc;eq
h 0:0668 c;eqL
Rec Prc
Nuc ¼ ¼ 3:66 þ Dc;eq
forRec < 2300 and Pr c  5
kc 1 þ 0:04½ð ÞRec Pr c 2=3 L

 1=3 !0:14
c Dc;eq
h Rec Prc lc
Nuc ¼ ¼ 1:86 for Rec < 2300 and Prc < 5
kc L=Dc;eq lc;s
V c Dc;eq 4ðW c;ch Hc;ch Þ
where Rec ¼ mc ; Dc;eq ¼ 2ðW
c;ch þH c;ch Þ
Airflow on back of module Heat transfer coefficient:
 1=3 !0:14
 D
h Reh;b Prh lh
h;b eq;b
Nuh;b ¼ ¼ 1:86 for Reh;b < 2300 and Pr h < 5
kh W=Dh;eq lh;s
V h Dh;eq
where Reh;b ¼ mh , Dh;eq ¼ 2s2
1234 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235

_ c ¼ 0:00553 kg=s, Th,i = 90.5 °C, and Tc,i = 24.3 °C, (b) m
Fig. 9. Percent error in predicting airside and waterside temperature drop with water flow rate for (a) m _ c ¼ 0:0069 kg=s,
_ c ¼ 0:0097 kg=s, Th,i = 69.0 °C, and Tc,i = 24.0 °C.
Th,i = 93.6 °C, and Tc,i = 24.4 °C, and (c) m

fin height is the height of the fin in the middle of the module. conditions. For the airside, the percent temperature error is
Therefore, the terms calculated from the average fin height are defined as
essentially values for the middle of the module.
jT h;o;exp  T h;o;th j
Fig. 9 shows the percent error in predicting the temperature % Airside temperature error ¼ ; ð39Þ
ðT h;in;exp  T h;o;exp Þ
drop across the air and water streams for three sets of operating

Fig. 10. Comparison of predicted and measured heat transfer rate for airside, waterside and average of two sides with water flow rate for (a) m_ c ¼ 0:00553 kg=s, Th,i = 90.5 °C,
_ c ¼ 0:0069 kg=s, Th,i = 93.6 °C, and Tc,i = 24.4 °C, and (c) m
and Tc,i = 24.3 °C, (b) m _ c ¼ 0:0097 kg=s, Th,i = 69.0 °C, and Tc,i = 24.0 °C.
 R. Nacke et al. / International Journal of Heat and Mass Transfer 54 (2011) 1224–1235 1235

Fig. 11 shows good agreement between model and experiment,

with values for all except one test falling below 10% error.
These results demonstrate the effectiveness of the model as a
tool for both the design and performance assessment of cross-flow
micro-channel heat exchangers.

5. Conclusions

This paper explored the design, analysis and performance

assessment of a new class of heat exchanger intended for high
Mach gas turbine engines. While the actual application will involve
dissipating the heat from hot compressor air to aircraft fuel, the
fuel in the present experiments was simulated using water. Key
Fig. 11. Percent error in predicting average heat transfer rate versus water flow rate
for different operating conditions.
findings from the study are as follows:

(1) A new cross-flow micro-channel heat exchanger, which con-

where Th,in,exp, Th,o,exp, and Th,o,th are the measured inlet temperature, sists of a number of closely spaced modules, was designed.
the measured mean outlet temperature, and the predicted mean Within each module, the fuel flows through a series of
outlet temperature, respectively. Similarly, the percent temperature parallel micro-channels, while the air flows externally over
error for the waterside is defined as rows of short, straight fins perpendicular to the direction
of fuel flow. This compact design is well suited for aircraft
jT c;o;exp  T c;o;th j
% Waterside temperature error ¼ ; ð40Þ applications where weight and volume are primary design
ðT c;o;exp  T h;in;exp Þ
concerns.
where Tc,in,exp, Tco,exp, and Tc,o,th are the measured inlet temperature, (2) A new theoretical model, which consists of a cross-flow heat
the measured mean outlet temperature, and the predicted mean exchanger submodel, an internal micro-channel fluid flow
outlet temperature, respectively. Fig. 9 shows errors for both the submodel, a resistance network submodel, and associated
air- and watersides are mostly below 10% for the two higher airside correlations and models for heat transfer coefficients and
flow rates but increase to about 15% for the lowest airside flow rate, fins efficiencies, was constructed. The overall model is
where venturi measurement error is greatest. These errors may be shown to provide closure in predicting outlet temperatures
related to several factors. First, there are uncertainties associated and heat transfer rates for the two fluid streams.
with the various instruments used such as the venturis and the (3) Comparisons between theory and experiment show good
water flow meters. Also, there were slight fluctuations in the water agreement in both exit temperatures and heat transfer rates.
flow rate, which created added difficulty in pinpointing the exact This proves the model is a reliable tool for predicting the
water flow rate and caused slight fluctuations in the air outlet tem- module’s performance under actual turbine engine condi-
perature. Third, the model calculations are based on an average fin tions and for the design of aircraft heat exchangers of differ-
height for the finned airside of the test module, which introduces ent sizes and design envelopes.
some error in predicting the module performance. Nonetheless,
the small error values in Fig. 9 support the overall accuracy of the
heat exchanger model. Acknowledgments
The module’s heat transfer rates are measured and calculated
for the cold and hot streams, respectively, as The authors are grateful for the financial support of Rolls-Royce
and LibertyWorks through the Rolls-Royce Purdue University
_ c cp;c ðT c;o  T c;i Þ
qc ¼ m ð41aÞ Technology Center (UTC) in High Mach Propulsion.
and
References
_ h cp;h ðT h;i  T h;o Þ:
qh ¼ m ð41bÞ
[1] Rolls-Royce, The Jet Engine, fifth ed., Rolls-Royce plc, Derby, England, 1996.
Fig. 10 compares predicted and measured heat transfer rates for [2] G.B. Bruening, W.S. Chang, Cooled Cooling Air Systems for Turbine Thermal
the waterside and airside calculated according to Eqs. (41a) and Management, ASME Paper No. 99-GT-14.
[3] H. Huang, L.J. Spadaccini, D.R. Sobel, Fuel-cooled thermal management for
(41b), respectively. These results show good agreement in both
advanced aero engines, J. Eng. Gas Turbines Power 126 (2004) 284–293.
magnitude and trend. There is an increase in overall heat transfer [4] T.P. Kibbey, Impinging Jets for Application in High-Mach Aircraft Thermal
rate between Fig. 10(a) and (b) mostly because of the increasing Management, M.S. Thesis, Purdue University, West Lafayette, IN, 2004.
air flow rate. An appreciable decrease in the heat transfer rate be- [5] N.R. Herring, S.D. Heister, On the use of wire coil inserts to augment tube heat
transfer, J. Enhanced Heat Transfer 16 (2009) 19–34.
tween Fig. 10(b) and (c) appears to be mostly the result of the large [6] I. Mudawar, Assessment of high-heat-flux thermal management schemes, IEEE
decrease in air inlet temperature despite the higher air flow rate. Trans. Compon. Pack. Technol. 24 (2001) 122–141.
To further assess the accuracy of the model predictions of the [7] J.L. Mason, Heat transfer in crossflow, in: Proceedings of Second US National
Congress of Applied Mechanics, Ann Arbor, MI, 1954, pp. 801–803.
module’s heat transfer rate, the following error parameter is [8] W.H. Beyer, Standard Mathematical Tables, 24th ed., CRC Press, Cleveland, OH,
defined, 1976.
    [9] F. Incropera, D. Dewitt, T. Bergman, A. Lavine, Fundamentals of Heat and Mass
 qh;exp þqc;exp q þq  Transfer, sixth ed., Wiley, Hoboken, New Jersey, 2007.
 2
 h;th 2 c;th 
% Heat transfer rate error ¼   : ð42Þ [10] S.A. Klein, Engineering Equation Solver (EES), F-Chart Software, Madison, WI.
q þq
h;exp c;exp
2