Experimental Study on Heat

Transfer and Pressure Drop of

Y. R. Lin Recuperative Heat Exchangers
J. H. Du Using Carbon Foam
W. Wu This work focuses on the development of high-effectiveness recuperative heat exchangers
1 using solid and corrugated carbon foam blocks. Characterization tests were conducted
L. C. Chow on heat transfer and pressure drop for a single carbon foam block with different sizes.
e-mail: [email protected]
Results show that carbon foam can be an effective medium for heat transfer enhance-
ment, and a short length in the flow direction yields a high heat transfer coefficient. The
Department of Mechanical, Materials and
corrugation can enhance heat transfer and reduce pressure drop at the same time. A
Aerospace Engineering,
recuperative heat exchanger with carbon foam, which consists of separate blocks of
University of Central Florida,
carbon foams packed between thin sheets of stainless steel, was designed. The hot and
Orlando, FL 32816-2450
cold flow paths were arranged in counterflow in the recuperator. The heat exchanger was
designed in a modular manner so that it can be scaled up for a larger heat transfer
requirement or a higher overall effectiveness. The anisotropic property of carbon foam
W. Notardonato was exploited to achieve higher effectiveness for one pair of foam blocks. Experiments
Cryogenic Test Laboratory,
with four pairs of carbon foams were conducted to evaluate the performance of carbon
Kennedy Space Center,
foam used in the recuperative heat exchanger. Measurements were made for both solid
Titusville, FL 32899
and corrugated foams for comparison. With four pairs of carbon foam blocks, an overall
effectiveness ␧total greater than 80% was achieved. This paper demonstrates an approach
to reach an effectiveness ␧total of 98% by placing many pairs of carbon foams in series.
关DOI: 10.1115/1.4001625兴

Keywords: heat transfer, recuperators, cryogenics

1 Introduction changer was developed for applications in reverse Brayton cycle

systems, but this design is difficult to form a counterflow type heat
NASA Kennedy Space Center 共KSC兲 关1兴 has proposed a cryo-
exchanger, resulting in a reduction in effectiveness 关3兴. McCor-
genic system that incorporates integrated refrigeration using a re-
mick et al. 关4兴 developed a radial flow heat exchanger 共RFHX兲 for
verse Brayton cycle with liquid oxygen storage to enable ad-
their turbo-Brayton cryocoolers, and their RFHX has an effective-
vanced operations on the lunar surface. Efficient liquefaction, ness of 0.89. Hill et al. 关5兴 described the slotted-plate heat ex-
storage, and transfer of oxygen are essential for a sustainable con- changer 共SPHX兲, which is the current state-of-the-art recuperator
sumable transfer station for use on the lunar surface. The recu- for space systems. It is comprised of a stack of copper plates
perative heat exchanger is one of the critical components of a containing concentric rings of slotted flow passages separated by
cryocooler system. New heat exchanger designs that are modular, low-conductivity stainless steel spacers brazed between the con-
easily scalable, highly effective, compact, and lightweight must be centric rings of slots. The effectiveness could reach 0.997 at a 1
developed. In order to remove 48 W of thermal energy at 90 K g/s mass flow rate of neon in the system. They also developed a
required for liquefaction and a zero loss storage of oxygen, the lighter and smaller recuperator by replacing the copper–stainless
heat exchanger must have a 2000 W capacity for heat transfer. The steel 共Cu-SS兲 SPHX with a silicon micromachined recuperator
heat exchanger must be designed to be modular to scale up for a 共SMR兲. Hoch et al. 关6兴 reported a modular design of a high-
high capacity. A high-effectiveness 共⬎98%兲 recuperative heat ex- effectiveness recuperative heat exchanger for a pulse tube/reverse
changer is required to maintain the cryocooler at a high opera- Brayton hybrid cryocooler. The design utilized etched cooper
tional coefficient of performance 共COP兲. Zhou et al. 关2兴 investi- plates interleaved with a stainless steel axial conduction barrier.
gated the effectiveness of the recuperator on the COP for a reverse The measured effectiveness reached 0.93 at a 1 g/s mass flow rate
Brayton cryocooler used for liquefaction of oxygen. A pressure of helium. Another design was suggested for a counterflow micro-
ratio of 1.8 was used, and the isentropic efficiencies of the com- heat exchanger with low pressure loss 关7兴. The square grooves
pressor and turbine were assumed to be 65% and 75%, respec- were etched in polymer wafers, and the wafers were stacked one
tively. The result indicated that the COP increases from 0.07 to by one to form a rectangular heat exchanger. Special consideration
0.11 when the heat exchanger effectiveness increases from 0.96 to was made to reduce the axial conduction in order to achieve high
0.98. Therefore, it is very critical to develop a high-effectiveness effectiveness.
recuperative heat exchanger. In a recent study 关8兴, we proposed a design for a compact,
Some developmental efforts for recuperative heat exchangers lightweight, recuperative heat exchanger with an effectiveness of
for cryocoolers have been made. The microtube strip heat ex- 98% using 50 pairs of carbon foam blocks. It was shown that
while the carbon foam blocks did provide superior heat transfer
1
performance, the pressure drop across the recuperator could be
Corresponding author.
Contributed by the Heat Transfer Division of ASME for publication in the JOUR-
excessively high. One attractive suggestion to reduce the pressure
NAL OF HEAT TRANSFER. Manuscript received July 23, 2009; final manuscript received drop while improving heat transfer performance by a corrugated
April 2, 2010; published online June 29, 2010. Assoc. Editor: W. Q. Tao. cutting of the carbon foam was made by Gallego and Klett 关9兴.

Journal of Heat Transfer Copyright © 2010 by ASME SEPTEMBER 2010, Vol. 132 / 091902-1

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 However, an explanation for the surprising heat transfer improve-
ment with a lesser amount of carbon foam due to corrugated cut-
ting was not provided.
In this study, we measured the heat transfer between a heated
substrate and a gas passing through a solid carbon foam block and
a corrugated carbon foam block. The goal was to provide an ex-
planation for the physical mechanism for the large reduction in
pressure drop and an increase in heat transfer performance due to
corrugation. Also in this paper, we confirmed that the use of cor-
rugated foams could lead to a highly effective recuperator with
better heat transfer performance but a very large reduction in the
overall pressure drop compared with a recuperator where solid
carbon foams are used 关8兴.
The carbon foams used in this study have an anisotropic ther-
mal conductivity. The bulk thermal conductivity is 135 W/m K in
the out-of-plane direction and is 45 W/m K in the other two in-
plane directions 关10兴. To increase heat transfer between the hot
and cold fluids, the out-of-plane direction 共with higher thermal
conductivity兲 is in the transverse direction of the recuperator. This Fig. 1 SEM picture of carbon foam
study experimentally investigated the performance in heat transfer
and pressure drop of a single carbon foam block. Tests were con-
ducted on different-sized blocks with different lengths in the flow dium with 350 ␮m pores connected with 110 ␮m holes between
direction. The effects of corrugation on heat transfer and pressure pores. The distance between two connecting holes represents the
drop were investigated for two different samples of carbon foam. characteristic length of the pore structure in the carbon foam. The
Based on single carbon foam block results, a recuperative heat Poco carbon foams used in this work have a density of 0.6 g / cm3
exchanger, which consists of separate blocks of carbon foam and a porosity of 0.75. The open-cell structure of the foams pro-
packed between thin sheets of stainless steel, was designed. In- vides a large surface area to volume ratio 共⬎40,000 m2 / m3兲.
stead of using a continuous piece of carbon foam, multiple pairs Table 1 lists some typical properties of carbon foams and alumi-
of carbon foam blocks were used in the design. Spaces between num foams. In comparison with aluminum foams, carbon foams
the blocks were used to reduce heat transfer down the length of have a smaller pore size and a larger surface area per unit volume.
the heat exchanger. The anisotropic feature of carbon foam was Carbon foams have thick ligaments and high thermal conductivi-
exploited in the design for each pair of foam blocks. The relation ties. The high thermal conductivity of carbon foams with the open
between the overall effectiveness of the recuperative heat ex- interconnected cellular structure provides a potential for effective
changer and the effectiveness of one pair and the number of pairs heat transfer enhancement. Carbon foam has been identified as a
of carbon foams was given. Experiments were also conducted on potential replacement for metal fins for heat transfer enhancement
four pairs of stacked carbon foams to verify the feasibility of the in heat transfer equipment such as heat exchangers, radiators, and
approach. Air was used to test the performance of the heat ex- thermal protection systems.
changer in this study. The corrugation effects on heat transfer and There is a limited amount of information on heat transfer and
pressure drop were also investigated to show that corrugation can pressure loss for gases in carbon foams. Klett et al. 关15兴 and
enhance heat transfer and reduce pressure drop at the same time. Gallego and Klett 关9兴 reported the performance tests for air pass-
This design of recuperative heat exchanger can have an effective- ing through a channel filled with carbon foams and aluminum
ness of more than 98% with multiple pairs of foam blocks. foams with several different configurations at the front velocity
from 1.5 m/s to 4.4 m/s. They reported that the heat transfer be-
2 Heat Transfer and Pressure Drop of One Foam tween a heated substrate and a moving air stream when carbon
Block foam was used could be ten times higher compared with the heat
Heat transfer between the hot and cold gases must be signifi- transfer when aluminum foam was used. At an air velocity of 4.37
cantly improved in order to increase the effectiveness of the heat m/s, their study showed that carbon foam in the bulk solid form
exchanger and reduce the size of the heat exchanger. Carbon foam has a heat transfer coefficient of 2600 W / m2 K, while for the
was used for this purpose in this study. Carbon foam has an ef- aluminum foam, the heat transfer coefficient is 250 W / m2 K.
fective thermal conductivity of 100–230 W/m K in the out-of- Mahjoob and Vafai 关16兴 also investigated extensively on the per-
plane direction 关10兴, which is much higher than the effective ther- formance metal foam heat exchangers and reported a heat transfer
mal conductivity of aluminum foam 共5–9 W/m K兲 关11兴. The coefficient of 363 W / m2 K at a gas 共90% air, 10% CO2兲 flow
compressive strength of the carbon foam is 1–3 MPa 关12兴. In velocity of 5 m/s.
addition, it has been shown 关13兴 that the thermal conductivity of While carbon foams yield much higher heat transfer coeffi-
carbon foam increases as its temperature changes from room tem- cients compared with metal foams at a flow velocity of 4.37 m/s,
perature to 80 K. We have also submerged carbon foams in liquid the pressure drop across carbon foams is more than 40 times
nitrogen for a long period of time and have also found no notice- higher compared with aluminum foams 关9兴. Gallego and Klett 关9兴
able physical changes. Yu et al. 关14兴 reported an application of also found that the pressure drop for solid foam could be signifi-
carbon foam in a compact recuperator for gas turbine systems.
Their investigation was focused on heat transfer enhancement us-
ing carbon foam. Based on their thermal analysis using a full Table 1 Properties of carbon foam and aluminum foam †10,11‡
length foam block, they reported that a heat exchanger effective-
Property Al foams Carbon foams
ness of 100% can be achieved. No consideration was given on the
effect of axial conduction along the carbon foam on the effective- Ligament thickness 共␮m兲 100–420 ⬃500
ness. Pore size 共␮m兲 1–2 0.350
Figure 1 shows a scanning electron microscopy 共SEM兲 picture Porosity ⬎0.9 0.7–0.8
we took from a carbon foam, which was originally developed by Effective k 共W/m K兲 5–9 100–230
Oak Ridge National Laboratory 共ORNL兲 and licensed to Poco Specific area 共m2 / m3兲 800–2800 ⬎40,000
Graphite, Inc. 关10兴. Carbon foam has the feature of a porous me-

091902-2 / Vol. 132, SEPTEMBER 2010 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 2 Experimental setup of heat transfer and pressure drop of one foam block

cantly reduced by modifying the geometry of solid carbon foam as ing through it, which were measured by a multimeter 共Keithley
finned, pin-finned, and corrugated while keeping the heat transfer 2001兲. The substrate temperatures were measured by three 0.6 mm
coefficient very high. With the corrugated geometry, compared Omega type E thermocouples, which were inserted through three
with solid carbon foam, the heat transfer coefficient increases, 0.88 mm holes. The inlet and outlet temperatures were measured
while the pressure drop decreases. Though the reason for this with the same kind of thermocouples. The pressure difference
unusual characteristic of carbon foam was not explained, it offers across the foam samples was measured by an Omega RX139 se-
a very promising potential to use corrugated foam for heat ex- ries pressure differential gauge. The flow rate was measured by a
changer applications. float flow meter, which was calibrated with a high-precision mass
In this work, we explore heat transfer and pressure drop of air flow meter. The velocity of air entering the foam was calculated
passing through a single carbon foam block to evaluate the heat from the flow meter readings, with adjustments made based on the
transfer performance of open-cell carbon foams when used as a inlet and outlet pressures and temperatures. The error in power
gaseous heat exchanger at a low velocity 共⬍1 m / s兲. A low ve- calculation from the voltage and current measurements can be
locity is needed to ensure a low pressure drop to avoid a signifi- ignored in the experiments. The error in the flow rate and velocity
cant reduction in COP for a reverse Brayton cryocooler. A low was estimated to be less than 3%. The error in pressure difference
velocity was also chosen because this would lead to higher overall was less than 3%. The error in temperature measurement was
heat transfer effectiveness. Experiments were conducted at much within 0.2° C. The error in h was less than 6%.
lower velocities than those conducted by Gallego and Klett 关9兴. Performance tests were conducted on Poco foam samples with
There are no available data in this range of velocity in the litera- two different flow lengths. One foam sample was
ture. Performance tests were evaluated on two different sizes of 5 cm 共length兲 ⫻ 5 cm 共width兲 ⫻ 1 cm 共height兲, while the sec-
foam samples. The effects of corrugated configuration on heat ond was 1 cm 共length兲 ⫻ 5 cm 共width兲 ⫻ 1 cm 共height兲. In one
transfer and pressure drop were investigated. Data for each foam experiment with the first foam sample, the temperature distribu-
configuration were correlated at different volumetric flow rates. tion on one side of the foam was monitored with an infrared 共IR兲
Physical mechanisms underlying the phenomena were analyzed. camera. In order to do this, the channel wall on this side was
The results of this study can serve as a reference for further opti- replaced by a thin 共0.1 mm兲 sheet of stainless steel foil, which is
mization of the configuration to enhance the heat transfer coeffi- bonded to the side of the foam block with the Pyro-Duct epoxy
cient and reduce the pressure drop at the same time. 关17兴. Since the heat loss from this side wall was very small, the
The test setup shown in Fig. 2 consists of an air tank at a temperature distribution of the foam should be approximately one-
constant pressure and a channel for tests with an inlet and an dimensional 共in the direction normal to the heater兲. Figure 3
outlet. Cooling air was forced through the test section. A control shows the IR image obtained with an air velocity of 0.5 m/s. The
valve was used to regulate the flow rate, which was measured by substrate temperature was measured to have an average tempera-
a flow meter at the outlet. The foam samples were glued to a ture of 53° C with a 1 ° C variation. The IR image showed that the
5 cm⫻ 5 cm⫻ 1.2 mm thick copper plate with a silver loaded entire foam has a very uniform temperature throughout, with an
epoxy 共Pyro-Duct兲 from Aremco 关17兴. This epoxy has a thermal
conductivity of 9 W/m K. Heat was provided from the bottom
with a foil heater purchased from Minco. The heater has the same
footprint and area as the copper plate. A cushion was used to press
the heater up against the plate in order to make good thermal
contact. A guard heater was used to prevent heat loss through the
bottom of the experimental apparatus. There were two thermo-
couples, one on each side of the cushion. The voltage for the
guard heater was regulated to keep the two thermocouples at the
same temperature in order to reduce heat loss. The foam was
placed and secured in a Plexiglas chamber. The test chamber was
insulated and sealed.
The variables measured in this experiment included power sup-
plied to the heaters, the substrate temperatures, the temperatures at
the inlet and outlet, the pressure difference across the sample, and
the volumetric air flow rate. The power output of the heaters was
calculated from the voltage across the heater and the current pass- Fig. 3 IR image indicating carbon foam temperature

Journal of Heat Transfer SEPTEMBER 2010, Vol. 132 / 091902-3

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 4 „a… Heat transfer and „b… pressure drop versus velocity for 5 Ã 5 Ã 1 cm3 solid and corrugated foams

average temperature of 51.5° C. This image demonstrated that car- heat transfer coefficient varies with velocity. These results follow
bon foam is a good medium for heat transfer enhancement be- the same trend as those reported in Klett et al. 关15兴 though the air
cause of its high thermal conductivity. The carbon foam acts es- velocity reported in Ref. 关13兴 is much higher. It should also be
sentially as a fin with a very large surface area and a fin efficiency noted that the height of the carbon foam used in Ref. 关13兴 is 31.75
of near unity. mm, while the height of the samples in our experiments is 10 mm.
The heat transfer of foam is characterized by the heat transfer While the foam structure provides an excellent thermal path to
coefficient, which is defined as reduce the temperature difference between the heater surface and
the flowing fluid, the pressure drop is too high for many applica-
Q
h= 共1兲 tions, as shown in Fig. 4共b兲. To reduce the flow resistance, Gal-
A⌬TLMTD lego and Klett 关9兴 reported heat transfer and pressure drop results
where Q is the heat absorbed by the air as it passes through the for a corrugated configuration of carbon foam. Figure 5 shows a
carbon foam. It is given by picture of a piece of 5 ⫻ 5 ⫻ 1 cm3 corrugated foam used in our
experiments. There is a 2 mm layer of foam that is not cut at the
Q = ṁc p⌬T 共2兲 bottom in order to keep good contact between the foam and the
where ⌬T is the temperature increase of air 共To − Ti兲, A is the substrate. As shown in Fig. 4共b兲, the pressure drop is reduced by a
footprint area of the foam sample, and ⌬TLMTD is the log mean factor of about 30 for a corrugated cutting, while the heat transfer
temperature difference given by coefficient increases by a small amount at a velocity of 1 m/s 共see
Fig. 4共a兲兲. Figure 5 also illustrates the flow paths when air enters
共Tw − Ti兲 − 共Tw − To兲 the corrugated foam. The flow path of air in the foam material is

冉 冊
⌬TLMTD = 共3兲 shorter in the corrugated carbon foam. Due to the larger frontal
Tw − Ti
ln area where the frontal entrance length is increased from 5 cm to
Tw − To 19 cm, the air speed entering the corrugated foam is also lower
Figure 4共a兲 shows the heat transfer coefficient for carbon foams than that in the solid foam. Since the pressure loss across foams is
共5 ⫻ 5 cm2 with a 1 cm height兲 at different air velocities. The dependent on the flow length and air velocity, the pressure drop is
variation in heat transfer coefficient with velocities for corrugated significantly lower in the corrugated carbon foam configuration
foam was also shown in this figure. These results indicate that the than it is in the solid foam configuration.

Fig. 5 Illustration of air flow in the 5 Ã 5 cm2 corrugated foam

091902-4 / Vol. 132, SEPTEMBER 2010 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 7 The temperature variation in an assumed straight tube
2
Fig. 6 1 Ã 5 cm corrugated foam

Figure 7 shows the variations in the temperature difference at

Experiments were also conducted on a solid foam sample with velocities of 0.5 m/s, 1.0 m/s, 1.5 m/s, and 4 m/s. It can be seen
the size of 1 cm 共length兲 ⫻ 5 cm 共width兲 with 1 cm 共height兲. that the air temperature rises to near the wall temperature in a
The length in the flow direction was reduced from 5 cm to 1 cm. distance of less than 1 mm if the velocity in the tubes is 0.5 m/s.
Figure 6 shows a picture of the 1 ⫻ 5 ⫻ 1 cm3 共tall兲 corrugated This corresponds to a frontal velocity of approximately 0.38 m/s
foam. Like the 5 ⫻ 5 ⫻ 1 cm3 corrugated foam, there is also a 2 in the experiments because the porosity of carbon foam ␧ is about
mm layer of foam at the bottom near the substrate. Corrugation 0.75. This length increases to 4 mm if the velocity is 4 m/s. In
has effects on both heat transfer and pressure drop for carbon experiments on the 5 ⫻ 5 ⫻ 1 cm3 foam sample, when the frontal
foam. Obviously, corrugation cutting has reduced the penetration velocity was 1 m/s, the air acquired the foam temperature in a
length of air flow within the foam. This results in a reduction in distance of 1 mm. After that flow length, the foam did little in heat
the pressure drop with a corrugated configuration. The cutting also transfer but caused more pressure loss. This explains why heat
reduces the penetration velocity 共defined as the air speed as it transfer from the foam to the air was not reduced when corruga-
enters the foam兲, which is dependent on the original dimensions tion reduces the air flow length in the foam from 5 cm to 2.5 mm
of the solid foam. In the 5 ⫻ 5 ⫻ 1 cm3 case, corrugated cutting 共see Fig. 5兲. In fact, as shown in Fig. 4共a兲, because of better air
reduced the penetration velocity by a factor of 4. Both shorter distribution over the heated surface 共illustrated in Fig. 5共c兲兲, the
flow length and reduction in air penetration velocity result in a overall heat transfer coefficient is actually higher for corrugated
very significant reduction in pressure drop. foams.
The effects of corrugated cutting on heat transfer are more com- For the shorter 1-cm-long foams, Fig. 8共a兲 shows results for the
plex. Corrugation changes the distribution of cold air in the foam heat transfer while Fig. 8共b兲 shows results for the pressure drop. It
significantly, as shown in Fig. 5共c兲. This results in a more effective can be seen that the heat transfer coefficient increases by about
and uniform cooling of the heater. In addition, corrugation reduces 10% with corrugated foam compared with the solid foam at a 1
the thermal path from the heater surface to the cold air, leading to m/s velocity, while the pressure drop reduces by 3.8 times com-
an overall smaller thermal resistance. The underlying mechanics is pared with the solid foam at the same velocity. The pressure drop
related to variations in temperature difference between the fluid reduction is not as large compared with the previous case of the
共air in this study兲 and the solid matrix of the carbon foam. The 5 ⫻ 5 ⫻ 1 cm3 sample 共30 times for the 5 ⫻ 5 ⫻ 1 cm3 foam兲
open-cell structure of foam was modeled as many parallel tubes partly because the frontal entrance length only increases from 5
along the flow direction in the solid material with a uniform tem- cm to 10.6 cm.
perature. The diameter of the tubes was assumed to be the pore It is interesting to note that the heat transfer coefficient for the
size of the foam 共0.35 mm兲. The temperature difference between 5 ⫻ 5 ⫻ 1 cm3 sample 共Fig. 4共a兲兲 is lower than that for the 1 ⫻ 5
the axial surface and the mean temperature decreases exponen- ⫻ 1 cm3 sample 共Fig. 8共a兲兲. A decrease in the flow length by a
tially with distance along the tube axis. It can be expressed as 关18兴 factor of 5 leads to an increase in the heat transfer coefficient by a

␹=
Tm共x兲 − Tm共0兲
Tw − Tm共0兲
= 1 − exp −
Px
冉
ṁc p
h̄ 冊 共4兲
factor of about 3. This is because the heat transfer coefficient as
defined by Eq. 共1兲 is footprint area dependent 共the area of the 1
⫻ 5 ⫻ 1 cm3 sample is five times smaller兲. Since carbon foams
where Tm共x兲 is the mean air temperature at position x, Tm共0兲 is the have a small pore size and a very large surface area to volume
mean air inlet temperature, Tw is the wall temperature of tube 共i.e., ratio, air can pick up its temperature rapidly while moving
through the foam. As the air temperature increases, the local tem-
the solid foam temperature兲, P is the tube perimeter 共P = ␲d兲, h̄ is
perature difference between the foam and air decreases along the
the mean value of the heat transfer coefficient in the tubes, and ṁ flow direction. When carbon foam was cut shorter in the flow
is the mass flow rate, where direction, the reduction in the amount of heat transferred from the
␳ v m␲ d 2 solid substrate to air is not much, while the footprint area drops by
ṁ = 共5兲 a factor of 5. Therefore, shorter carbon foam in the flow direction
4
yields higher heat transfer coefficients.
and vm is the mean air velocity in the tubes. To get a rough It can be concluded from these experiments that carbon foams
estimate of the distance x at which the air acquires a temperature are good media for heat transfer enhancement due to their high
close to that of the foam, one can assume a heat transfer coeffi- thermal conductivity of the material and the high specific heat
cient corresponding to that of the fully developed laminar flow. transfer area. The results indicated that short-length samples in the

Journal of Heat Transfer SEPTEMBER 2010, Vol. 132 / 091902-5

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 8 Heat transfer and pressure drop versus velocity for 1 Ã 5 Ã 1 cm3 solid and corrugated foams

flow direction have a high heat transfer coefficient because air could allow a large pressure differential between the hot and cold
acquires heat in carbon foam within a short flow length. Corruga- fluids if needed. Huang and Vafai 关20兴 numerically investigated
tion increases heat transfer and reduces pressure drop at the same convection heat transfer in a partially blocked channel using po-
time. An optimal design that includes corrugation can enhance rous blocks on one side of the channel. Yucel and Guven 关21兴
heat transfer and reduce pressure drop at the same time. reported an investigation on forced convection in a partially
blocked channel using porous blocks on both sides of a channel.
3 An Approach to High Effectiveness From One Pair Both papers reported results of cross flow over the porous blocks
to Multiple Pairs of Foam Blocks in the channels. In the configuration of the current study, the chan-
nel was fully blocked by carbon foam blocks. Air is forced to
As indicated from the experiments for one foam block, a short- penetrate the carbon foam blocks uniformly to take advantage of
length foam in the flow direction has a higher heat transfer coef- the highly conducting carbon foam to achieve maximum heat ex-
ficient with less pressure drop. Therefore, short-length foam change effectiveness. Flow and heat transfer in the partially and
blocks should be used instead of a long piece of foam in the fully blocked configurations have different applications and per-
recuperative heat exchanger. While it is important to reduce the formance.
thermal resistance in the transverse direction between the hot and The basic unit of the heat exchanger is a pair of foam blocks
cold fluids as much as possible, the axial conduction in the recu- packed with a thin sheet of solid materials in between, as shown
perative heat exchanger can degrade its performance considerably in Fig. 10. Since the flow rates are the same for both the hot and
关19兴. The high effectiveness of the heat exchanger needs to be the cold sides, the effectiveness of the one-pair carbon foam heat
achieved through a series of discrete carbon foam blocks instead exchanger is defined as
of a continuous piece. Figure 9 shows a schematic of the concept
of a recuperative heat exchanger. The basic unit was comprised of TH,i − TH,o TC,o − TC,i
separate blocks of carbon foams packed between thin sheets of a ␧= or ␧= 共6兲
TH,i − TC,i TH,i − TC,i
solid material. The flow paths were piled alternately, and the hot
and cold gas paths were arranged in counterflow. The stacked Transverse heat transfer from hot to cold is enhanced through
parallel plate arrangement allows symmetrical counterflow pas- the carbon foam blocks. Heat transfer in carbon foam involves the
sages that balance flow across each heat transfer interface. This combined process of heat conduction within the carbon ligaments
design is modular and can easily be scaled up for large capacity and convective heat transfer between the ligaments and air in the
by using more stacks. open-cell carbon foam. The foam is an open interconnected
It is important to minimize axial conduction between the hot spherical structure. The orientation distribution of the carbon liga-
and cold ends of the heat exchanger to achieve high effectiveness.
In order to reduce this axial heat conduction, several measures
were taken in the design. First, the separation plates were made
thin to reduce the cross-sectional area in the axial direction. Sec-
ond, the material used should have a low thermal conductivity k.
Commercially available 100-␮m-thick stainless steel sheets were
chosen 共k = 11 W / m K兲. It should be noted that the carbon foam
blocks also act as a support for the thin stainless steel sheet and

Fig. 9 Schematic of heat exchanger Fig. 10 Heat transfer between one carbon foam pair

091902-6 / Vol. 132, SEPTEMBER 2010 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 11 Axial temperature distributions of hot and cold streams within recuperator

ment in the foam is randomly arranged. Dispersion generally has blocks, with each pair responsible for a limited temperature
effects on the interfacial heat transfer between the solid structure change. The number of pairs required in the design of the heat
and the fluid in porous media, which is highly dependent upon the exchanger depends on the performance of each pair. The relation
specific structure of each kind of porous media. Heat transfer between the overall effectiveness, ␧total, and the effectiveness of
between solid and fluid in porous media due to temperature dif- one pair, ␧, can be expressed as follows:
ference is dependent upon the properties of the fluids and solid. As
carbon foams have a small pore size and a very large surface area N·␧
␧total = 共7兲
to volume ratio and high solid-to-fluid thermal conductivity ratio 1 + 共N − 1兲␧
共ks / k f 兲 for gas flow in carbon foams, the equilibrium model with where N is the number of carbon foam pairs.
effective properties is applicable to heat transfer in carbon foam Even if the effectiveness of each pair of carbon foam blocks is
关22兴. The major factor for the enhancement of heat transfer in this only 50%, the overall effectiveness can be as high as 98% through
case is the extended surface effect. The axial conduction in the 50 pairs of them placed in series. From the above analysis, it
flow direction will reduce the effectiveness when only one pair of appears that the design concept of using discrete 1 cm carbon
carbon foams is used in a recuperative heat exchanger. foam blocks with a 0.2 cm air gap between them can provide an
In an earlier study, Yu et al. treated the carbon foam as a overall effectiveness of 98%.
straight fin structure rather than a porous medium to simulate heat
transfer enhancement 关14兴. They used Taylor’s formulas to deter-
mine the thickness of the fins and space between them from the 4 Experiments on a Recuperative Heat Exchanger
properties of the carbon foam. However, they did not consider With Four Pairs of Foam Blocks
axial conduction in their simulation and concluded that a heat
In the development of recuperative heat exchangers with carbon
exchanger effectiveness of 1.0 can be achieved.
foam, experiments were conducted to validate the design with a
Since heat is transferred from hot fluid to cold fluid, the carbon
heat exchanger with four pairs of foam blocks. Both solid foam
foams should be oriented with their out-of-plane direction with
and corrugated foam were used in experiments. Figure 12 shows
higher thermal conductivity in the transverse direction. The in-
the configuration of a recuperator where the hot and cold flow
plane direction with lower thermal conductivity should be in the
channels have four carbon foam blocks each. A schematic of the
flow direction to lower the axial conduction effect 关23兴. Taking
experimental setup is provided in Fig. 13. The dimensions of the
into account the anisotropic properties of the carbon foam and
flow channels are 10⫻ 1.7⫻ 1 cm3 共length⫻ width⫻ height兲. The
axial conduction effects, a fin model similar to that of Yu et al.
关14兴 was developed, which showed that the effectiveness of one cross sections of the air inlet and outlet are both 1.7⫻ 1 cm2
pair of foam blocks is 0.58 with a carbon foam pair of 1 cm length
in the flow direction at an air speed of 0.67 m/s 关8兴.
The effectiveness of one pair is limited because of axial heat
conduction due to heat transfer from the hot end to the cold end
for each pair of carbon foam blocks. An overall high effectiveness
of the heat exchanger can be achieved through a series of carbon
foam pairs. Figure 11 shows the predicted stepwise change of the
hot and cold fluid temperatures in the heat exchanger. The present
requirement for the recuperative heat exchanger is for the tem-
perature of the hot stream to decrease from 298 K to 98 K 共a 200
K difference兲, while the temperature of the cold stream has to
increase from 94 K to 294 K 共also a 200 K difference兲. This can Fig. 12 Configuration of a recuperator with four pairs of car-
be accomplished with a sufficient number of pairs of carbon foam bon foams

Journal of Heat Transfer SEPTEMBER 2010, Vol. 132 / 091902-7

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 13 Schematic of experimental setup for evaluating recuperator performance

共width⫻ height兲. Carbon foam blocks of dimensions 1 ⫻ 1.7 共TH,i − TC,o兲 − 共TH,o − TC,i兲

冋 册
⫻ 1 cm3 共length⫻ width⫻ height兲 are glued to a 0.1-mm-thick ⌬Tm = 共10兲
TH,i − TC,o
stainless steel plate using a silver loaded epoxy 共Pyro-Duct™兲.关17兴 ln
Using the recommended procedure to apply the epoxy as de- TH,o − TC,i
scribed in the manufacturer data sheet, the contact resistance be- where TH,i, TH,o, TC,i, and TC,o are the hot and cold, inlet and
tween the foam block and the stainless steel sheet can be esti- outlet air temperatures.
mated to be around 10−5 K / W and can be ignored due to the low The effectiveness of the heat exchanger can be calculated based
heat flux in the present experiments. The thickness of epoxy is on the temperature drop of the hot air or the temperature rise of
250 ␮m, and the epoxy has a thermal conductivity of 9 W/m K. It the cold air,
was calculated that if the total length of the air gaps is equal to
共TH,i − TH,o兲
20% of the total length of foam blocks, axial conduction would be ␧total H = 共11兲
less than 0.1% of the amount of heat transfer between the hot and 共TH,i − TC,i兲
cold streams. The total length of the air gaps is not expected to
have much effect on the overall effectiveness of the heat ex- 共TC,o − TC,i兲
␧total C = 共12兲
changer. The carbon foam blocks were mounted and sealed inside 共TH,i − TC,i兲
a Plexiglas chamber. The carbon foam occupied the entire cross
section of the flow channel. The entire test chamber was insulated Since the flow rates were the same for both the hot and the cold
and sealed to prevent air leaks. The cold air stream inlet was sides, and the specific heat could be assumed to be constant in the
provided by passing room temperature air from the laboratory temperature range of the experiment, the hot and cold side effec-
supply system through a cold thermal bath. The air from the cold tiveness should be the same in theory. The small difference 共error兲
outlet TC,o was then heated by another hot thermal bath to become in the effectiveness calculated from the hot and cold streams is
due to measurement errors and the very small heat transfer be-
the hot air stream at the inlet temperature TH,i. In this way, both
tween the flow channels with the surrounding air,
the cold and hot air streams have the same mass flow rate.
The performance of carbon foam blocks when used in a recu- 兩␧total H − ␧total C兩
perative air heat exchanger was evaluated by the effectiveness and error = 共13兲
␧total H
pressure drop. The overall heat transfer coefficient was obtained
by measuring the inlet and outlet temperatures and flow rate. The The average effectiveness of the heat exchanger is calculated as
air temperatures between the foam blocks were measured by 0.3 兩␧total + ␧total C兩
H
mm diameter thermocouples, which were placed in the space be- ␧total = 共14兲
tween the carbon foam blocks. The temperatures at the inlet and 2
outlet 共TH,i, TH,o, TC,i, and TC,o, i.e., hot/cold air inlet/outlet tem- The air flow speed V and the inlet and outlet air temperatures
perature兲 were also measured. The flow rate was measured with a were measured in the experiment. The measured data were pro-
flow meter that was calibrated with a high-precision mass flow cessed using Eqs. 共6兲–共12兲. The uncertainty for U at V
meter. All data collection was automated using a Keithley 2700 = 0.25 m / s is 25% due to a small temperature difference, but
data acquisition system and a personal computer. The thermo- those at high velocities are small. The average uncertainties for
couples used in experiments were carefully calibrated, and the Qair, ⌬Tm, and U are 4.2%, 3.1%, and 5.9%, respectively. The
error in temperature measurement was within ⫾0.1° C. The errors uncertainty for ␧total was estimated to be 4.7% for the worst case
for pressure drop and flow rate were estimated to be less than ⫾10 with V = 0.25 m / s, and the average uncertainty for ␧total was
Pa and 1.2%, respectively. 3.2%.
The heat transfer performance of the heat exchanger with car- The heat transfer and the total pressure drop across the four 1
bon foam blocks was evaluated by the overall heat transfer coef- cm carbon foam blocks were measured at air flow speeds of
ficient U, which is defined as 0.25–1 m/s. Figure 14 shows the measured overall heat transfer
Qair coefficient and pressure drop of the recuperative heat exchanger
U= 共8兲 with four pairs of solid and corrugated carbon foam blocks. The
Ab⌬Tm measured overall heat transfer coefficient U for the solid foam
where Qair is the heat given up by the hot air that passes through was about 568 W / m2 K at the speed of 1 m/s. Without the four
the carbon foam. It was calculated by carbon foam pairs, the overall heat transfer coefficient was only
10.2 W / m2 K for the empty flow channels at V = 1 m / s. Heat
Qair = ṁc p共TH,i − TH,o兲 共9兲 transfer enhancement by the carbon foam is significant. Corruga-
Ab is the heat transfer area between the hot and cold fluids; it tion foam increases the heat transfer coefficient further to
coincides with the surface area of the dividing stainless steel 770 W / m2 K at V = 1 m / s. The pressure drop for the solid foam
sheet. ⌬Tm is the log mean temperature difference and is defined compares well with the pressure drop across a 5-cm-long solid
as carbon foam measured by Klett et al. 关15兴 when their pressure

091902-8 / Vol. 132, SEPTEMBER 2010 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Fig. 14 Overall heat transfer coefficient versus air flow speed for heat exchanger „solid and corrugated foams…

drop was scaled down linearly from 5 cm to 4 cm so that the total that of solid foam. Figure 15 also shows that though the average
lengths of the foam in the flow direction are the same. With a measured effectiveness for a one-pair foam at V = 0.25 m / s is
corrugated configuration, the pressure drop across the four pairs of around 0.57, by using four pairs of them in series, one can in-
foam blocks reduces more than five times at 1 m/s. These results crease the overall effectiveness to about 86% 共for both solid and
demonstrate that corrugated carbon foam is better both for heat corrugated foams兲.
transfer enhancement and pressure drop reduction, as demon-
strated in the experiments for the performance of one single car- 5 Conclusion
bon foam. Experiments using solid and corrugated carbon foams of differ-
Figure 15 shows the overall effectiveness with four pairs of ent sizes show that carbon foam can enhance heat transfer, and the
carbon foams for both the solid and the corrugated foam. The heat transfer coefficient increases as the foam length decreases.
average measured effectiveness values of one-pair solid foam are Corrugation can reduce the pressure drop while maintaining a
about ␧ = 0.50 and 0.57 when the flow velocities are V = 1 m / s high heat transfer coefficient. The effectiveness for one pair of
and 0.25 m/s, respectively. In a similar fashion, by using the air carbon foam blocks is limited due to the axial conduction effect of
temperatures across two and three pairs of carbon foams, one the foam. The anisotropic property of carbon foam can be used to
could measure the average effectiveness for two and three pairs of obtain a higher effectiveness for one pair of foam blocks. High
carbon foams. The average measured effectiveness for one, two, effectiveness of the heat exchanger can be achieved by placing
three, and four pairs of carbon foams is plotted in Fig. 15. The many pairs of carbon foam blocks in series. Experiments with
measured effectiveness for four pairs of solid foams decreases four pairs of carbon foam blocks demonstrated that an effective-
slightly from 85% to 80% when the air speed V increases from ness ␧total greater than 80% could be achieved. Corrugation can
0.25 m/s to 1 m/s, while for the corrugated foam, the effectiveness enhance the overall heat transfer coefficient and reduce pressure
drops from 86% to 81%. drop at the same time in recuperative heat exchangers Corrugation
It is noted that the average measured effectiveness values of does not have much effect on effectiveness though pressure drop
one corrugated carbon foam pair at V = 1 m / s and 0.25 m/s ve- is significantly reduced. Based on these test data, it is predicted
locities are about ␧ = 0.51 and 0.57, which are about the same as that an overall heat exchanger effectiveness ␧total of 98% can be
reached by using 50 pairs of carbon foam blocks.

Acknowledgment
This research was sponsored by Florida Space Grant Consor-
tium and NASA Kennedy Space Center.

Nomenclature
A ⫽ single block foam heat transfer area 共m2兲
Ab ⫽ base heat transfer area of the foam exchanger
共m2兲
h ⫽ heat transfer coefficient of a single foam block
共W / m2 K兲
k ⫽ thermal conductivity 共W/m K兲
L ⫽ length of the carbon foam block 共m兲
ṁ ⫽ mass flow rate of air 共kg/s兲
P ⫽ tube perimeter 共m兲
Q ⫽ heat transfer of a single block foam 共W兲
Qair ⫽ heat transfer from the hot to cold fluids 共W兲
TC,i ⫽ cold air inlet temperature 共K兲
Fig. 15 Comparison of the average measured effectiveness TH,i ⫽ hot air inlet temperature 共K兲
„solid and corrugated foams… TC,o ⫽ cold air outlet temperature 共K兲

Journal of Heat Transfer SEPTEMBER 2010, Vol. 132 / 091902-9

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 TH,o ⫽ hot air outlet temperature 共K兲 “Design and Experiment of Compact and Effective Carbon Foam Recuperative
Heat Exchangers,” J. Thermophys. Heat Transfer, 23, pp. 339–345.
Tm共x兲 ⫽ the mean temperature at the x position 共K兲 关9兴 Gallego, N., and Klett, J. W., 2003, “Carbon Foams for Thermal Manage-
Tw ⫽ the wall temperature of tube 共K兲 ment,” Carbon, 41, pp. 1461–1466.
⌬Tm ⫽ log mean temperature difference 共K兲 关10兴 Poco Graphite Inc., 2000, PocoFoam Product Information.
U ⫽ overall heat transfer coefficient based on the 关11兴 Leong, K. C., and Jin, L. W., 2006, “Effect of Oscillatory Frequency on Heat
Transfer in Metal Foam Heat Sinks of Various Pore Densities,” Int. J. Heat
log mean temperature difference 共W / m2 K兲 Mass Transfer, 49, pp. 671–681.
vm ⫽ averaged small tube air flow speed 共m/s兲 关12兴 Klett, J. W., 1999, “High Thermal Conductivity, Mesophase Pitch-Derived
cp ⫽ heat capacity of the air 共J/kg K兲 Graphitic Foams,” Composites in Manufacturing, 14共4兲, pp. 1–5.
关13兴 Klett, J. W., Mcmillan, A. D., Gallego, N. C., and Walls, C. A., 2004, “The
␧ ⫽ effectiveness of one-pair heat exchanger Role of Structure on the Thermal Properties of Graphitic Foams,” J. Mater.
␧total ⫽ total effectiveness of the heat exchanger Sci., 39, pp. 3659–3676.
␹ ⫽ nondimensional temperature 关14兴 Yu, Q., Thompson, B. E., and Straatman, A. G., 2005, “Carbon Foam—New
Generation of Enhanced Surface Compact Recuperators for Gas Turbines,”
ASME Paper No. GT2005-69007.
References 关15兴 Klett, J. W., Tee, C. C., Stinton, D. P., and Yu, N. A., 2000, “Heat Exchangers
关1兴 Notardonato, W. U., 2007, “Development of Consumable Transfer Systems for Based on High Thermal Conductivity Graphite Foam,” Proceedings of the
Sustainable Lunar Exploration,” 45th AIAA Aerospace Sciences Meeting and First World Conference on Carbon, EuroCarbon, pp. 244–245.
Exhibit, Reno, NV, AIAA Paper No. 2007-0957, January. 关16兴 Mahjoob, S., and Vafai, K., 2008, “A Synthesis of Fluid and Thermal Trans-
关2兴 Zhou, L., Kapat, J. S., Chow, L. C., and Li, X., 2000, “Design of a High port Models for Metal Foam Heat Exchangers,” Int. J. Heat Mass Transfer, 51,
Effectiveness Micro Heat Exchanger for Mars Application,” SAE, Int. Journal pp. 3701–3711.
of Aerospace, 109共1兲, pp. 875–882. 关17兴 Technical Bulletin A8, Aremco Company, www.aremco.com
关3兴 Doty, F. D., Hosford, G., Spitzmesser, J. B., and Jones, J. D., 1991, “The 关18兴 Incropera, F. P., Dewitt, D. P., Bergman, T. L., and Lavine, A. S., 2007, Fun-
Microtube Strip Heat Exchanger,” Heat Transfer Eng., 12, pp. 31–41. damentals of Heat and Mass Transfer, 6th ed., Wiley, New York.
关4兴 McCormick, J. A., Nellis, G. F., Sixsmith, H., Zagarola, M. V., Gibbon, J. A., 关19兴 Demko, J. A., and Chow, L. C., 1984, “Heat Transfer Between Counter Flow-
Izenson, M. G., and Swift, W. L., 2001, “Advanced Developments for Low ing Fluids Separated by a Heat-Conducting Plate,” AIAA J., 22共5兲, pp. 705–
Temperature Turbo-Brayton Cryocoolers,” Cryocoolers, 11, pp. 481–488. 712.
关5兴 Hill, R. W., Izenson, M. G., Chen, W. B., and Zagarola, M. V., 2007, “A 关20兴 Huang, P. C., and Vafai, K., 1994, “Analysis of Forced Convection Enhance-
Recuperative Heat Exchanger for Space-Borne Turbo-Brayton Cryocoolers,” ment in a Channel Using Porous Blocks,” J. Thermophys. Heat Transfer, 8,
Cryocoolers, 14, pp. 525–533. pp. 563–573.
关6兴 Hoch, D. W., Nellis, G. F., Meagher, N. L., Maddocks, J. R., and Stephens, S., 关21兴 Yucel, N., and Guven, R. T., 2007, “Forced-Convection Cooling Enhancement
2007, “Development and Testing of a Multi-Plate Recuperative Heat Ex- of Heated Elements in a Parallel Plate Channel Using Porous Inserts,” Numer.
changer for Use in a Hybrid Cryocooler,” Cryocoolers, 14, pp. 515–524. Heat Transfer, Part A, 51共3兲, pp. 293–312.
关7兴 Zhou, L., Kapat, J., Chow, L., and Lei, S., 2000, “Design of a High Perfor- 关22兴 Lee, D. Y., and Vafai, K., 1999, “Analytical Characterization and Conceptual
mance Cryocooler for Propellant Liquefaction and Storage on Mars,” Proceed- Assessment of Solid and Fluid Temperature Differentials in Porous Media,”
ings of the 2000 International Mechanical Engineering Congress and Exposi- Int. J. Heat Mass Transfer, 42, pp. 423–435.
tion, Orlando. 关23兴 Jones, G. F., 1995, “Axial Conduction in a Thick-Wall Matrix Heat Ex-
关8兴 Wu, W., Du, J. H., Lin, Y. R., Chow, L. C., and Notardonato, W., 2009, changer,” Cryogenics, 35, pp. 581–588.

091902-10 / Vol. 132, SEPTEMBER 2010 Transactions of the ASME

Downloaded From: http://heattransfer.asmedigitalcollection.asme.org/ on 01/28/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use