The Effective Thermal Conductivity of

High Porosity Fibrous Metal Foams

Greek Symbols
e = Porosity

spheres. Under the assumption of local thermal equilibrium (Carbonell and Whitaker, 1984), they derived a set of closure equations
for the spatial deviation component of the volume-averaged temperature field in the two constitutive phases. These closure equations were solved analytically and numerically to obtain the effective thermal conductivity. Their numerical results correlated well
with the experimental data for a touching parameter, cla = 0.02
(cla is the ratio of the touching length scale to the length scale of
the particle size). Based on a similar two-dimensional study,
Sahroui and Kaviany (1993) found a value of cla = 0.002 to be
more appropriate. Hsu et al. (1995) demonstrated that a onedimensional conduction model based on in-line touching cubes
(cla = 0.13) was sufficient. It showed good agreement with the
experimental data of a packed sphere bed.
Hsu et al. (1994) based their study on the earlier work of Zehner
and Schlunder (1970). They proposed two modelsthe area contact model for packed beds of spheres and the phase symmetry
model for sponge-like materials (fibrous media). They showed that
for packed beds of spheres, their area contact model was able to
predict the thermal conductivity better than the Zehner-Schlunder
model because it took into account the finite contact area between
adjacent particles. They also developed a phase symmetry model
for sponge-like porous media (e.g., metal foams). However, they
did not validate their phase symmetry model due to lack of
experimental data. Bauer (1993) generalized Maxwell's classical
theoretical result to pores of any shape and concentration. He
further discussed cases in which other phenomena-like radiation
can be included into the analysis. Tien and Vafai (1979) derived
statistical bounds for the thermal conductivity of microsphere and
fibrous insulations based on cell geometries. Their model for
fibrous insulations was used by Hunt and Tien (1988) to study
forced convection in metal foams. Its validity is discussed later in
this paper.

Introduction
Starting with the pioneering work of Maxwell (1891), heat
conduction in fully saturated porous matrices (e.g., sand, packed
beds of cylinders and spheres, fibrous insulations, etc.) has been
studied in detail over the past several decades. Kaviany (1995) has
provided an extensive review of the available literature on the
subject along with a number of correlations and their range of
applicability.
Under simplified one-dimensional conduction conditions, two
extremes can be considered. One in which the thermal resistances
offered by the solid and fluid phases are in series (lower bound)
and the other in which they are in parallel (upper bound). The
upper bound, given by Eq. (1),

Our survey indicates a lack of experimentally validated studies

that attempt to predict the thermal conductivity of metal foams,
although they have been the topic of investigation for quite a while
(Yokoyama and Mahajan, 1995; DuPlessis et al, 1994; Hunt and
Tien, 1988). Further, in order to explore their use as highperformance heat sinks, an accurate estimate of the effective
thermal conductivity is necessary.
In this study, we report our measurements of effective thermal
conductivity of metal foams (e > 0.9) made of aluminum. Experiments were performed with air and water as the fluid phase
separately. An empirical relation that correlates the experimental
data to within an accuracy of 97.5 percent is developed. Taking the
foam structure to be hexagonal, an analytical model for the effective thermal conductivity is derived and validated with the experimental data.

V. V. Calmidi1 and R. L. Mahajan2

Nomenclature
Am
b
k
L
L,
1
r
t
T
AT
V
w
z

=
=
=
=
=
=
=
=
=
=
=
=
=

area of cross section of metal foam :samples (m2)

half-thickness of bump (Fig. 5)
thermal conductivity (W/m-K)
half-length of fiber (Fig. 5)
height of metal foam sample (m)
heat input to the patch heaters (W)
area ratio (Eq. 11)
half-thickness of fiber (Fig. 5)
temperature
temperature difference across metal foam sample (C)
volume
width of fiber perpendicular to planes of paper (Fig. 5)
direction of macroscopic heat flow

Subscripts
e = effective
/ = fluid
s = solid

ke=(l

- e)ks + skf,

(1)

has been successfully used in the past for packed bed studies where
the solid and fluid phases have similar conductivities. However,
the error in the prediction of ke using Eq. (1) can be considerable
as the difference in the conductivities of the phases increases.
There have been several studies that attempt to predict the thermal
conductivities of packed beds by invoking the structure of the
medium, apart from its porosity. Some of these two and onedimensional studies are reviewed below. Many of them are related
to packed beds of spheres and granular materials.
Nozad et al. (1985) solved the two-dimensional heat conduction
equation in a spatially periodic two-phase system of touching
' IBM Microelectronics, Endicott, NY 13760. e-mail: vara@us.ibm.com.
Professor of Mechanical Engineering, CAMPmode, Department of Mechanical
Engineering, University of Colorado at Boulder, Boulder, CO 80309-0427. e-mail:
mahajan@colorado.edu. Fellow ASME.
Contributed by the Heat Transfer Division for publication in the JOURNAL OF
HEAT TRANSFER. Manuscript received by the Heat Transfer Division, July 17, 1997;
revision received, Nov. 3, 1998. Keywords: Conduction, Porous Media, Thermophysical Properties. Associate Technical Editor: M. Kaviany.
2

466 / Vol. 121, MAY 1999

Medium
Figure 1 shows a picture of the metal foam medium. It has an
open-celled structure composed of dodecahedron-like cells which
have 12-14 pentagonal or hexagonal faces. The edges of these
cells are composed of the fibers and, typically, there is a lumping
of material (intersection) at points where the fibers intersect. The
fibrous matrix is made of aluminum alloy T-6201 which has a
thermal conductivity of 218 W/m-K. The matrix is brazed to
aluminum base plates on two sides. Two quantities, the porosity
and the pore density are used to describe the material. The porosity, e, is the ratio of the void volume to the total volume of the
medium and the pore density is the number of pores present per
unit length of the material. The latter is typically expressed in units
of pores per inch (ppi), and is roughly constant in the three
directions. The first three columns of Table 1 summarize the
properties of the metal foam samples which were used in the
experimental study. The porosities (provided by the manufacturer)
were determined using the weight of the sample with the density of
aluminum. Note that the pore density is a nominal quantity used to

Copyright 1999 by ASME

Transactions of the ASME

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

MM%WrV?itiM

Fig. 2

Schematic of setup used for the metal foam and air experiments

Fig. 1 Photograph of metal foam matrix

classify a particular type of metal foam. In reality, it varies about

a mean value from sample to sample.

Experiment
(Metal Foam and Air): A schematic of the experimental
setup is shown in Fig. 2. The physical dimensions of the metal
foam are 6.3 cm X 6.3 cm X 4.6 cm. The metal foam is brazed at
the top and bottom to two aluminum plates (4.7 mm thick). Four
T-type thermocouples are attached to each of these two plates at
points of varying distance (see Calmidi, 1998, for details). The top
plate is heated from above using patch heaters connected to a DC
power supply. The surfaces above the heaters and the four sides are
insulated using very low conductivity styrofoam. The bottom plate
is cooled from below by immersing its bottom surface into a tank
which contains cooling water maintained at constant temperature.
A silicone-based sealant is used to ensure that the water does not
leak out of the tank.
Since the metal foam sample is heated from above, cooled from
below and insulated on the rest of the four sides, there is heat
conduction in one direction only (as shown in Fig. 2) and buoyancy effects are negligible. Thus, if "q" is the net heat input
through the top aluminum plate, then
q =

kcAmAT/Lm.

(2)

During a typical experimental run, the power input to the patch

heaters was set at a desired value and the temperature difference
between the top and bottom plates was monitored till it reached
steady state. This took approximately seven to ten minutes depending on the porosity of the sample. The temperature was monitored
for five additional minutes to confirm that steady state had indeed
been attained. Then, the temperature difference, AT between the

Table 1 Metal foam properties and thermal conductivities

No.

Porosity

Pore density
(pores/inch)

1
2
3
4
5
6
7
8
9
10
11

0.971
0.946
0.905
0.949
0.909
0.978
0.949
0.906
0.972
0.952
0.937

5
5
5
10
10
20
20
20
40
40
40

Journal of Heat Transfer

Conductivity
(air + foam)
(W/m-K)

Conductivity
(water + foam)
(W/m-K)

2.70
4.60
6.70
4.00
6.70
2.20
3.90
6.90
2.50
3.90
4.50

3.70
5.40
7.65
4.95
7.60
3.05
4.807.65
3.30
4.75
5.35

top and bottom plates, was noted. This procedure was repeated for
different values of heat input. The maximum heat input to the
patch heaters was restricted to about 5-8 W depending on the
porosity of the sample being measured. Since the thermal conductivity varies with porosity, the maximum heat input to the heaters
was set such that the temperature difference across the plates did
not increase beyond ~15C. The variation in temperature across
each of the two plates was less than 0.3C, which is the accuracy
of the temperature measurement. At lower temperature differences, the variation was even lower. Thus the two plates were
essentially isothermal.
The temperature difference between the hot and cold plates
obtained as described above was plotted as a function of the heat
input. All data points lied on a straight line passing through the
origin. Further, the y-intercept was less than 0.2C in all cases
indicating that the relationship is linear and that buoyancy effects
and other nonlinear effects are indeed negligible. From the slope of
AT/q, the measured value of the thermal conductivity of the
sample is

*, = L m /(A,(Ar/ ? ) s]ope ).

(3)

These values are listed in column 4 of Table 1.

Experiment (Metal Foam and Water): Experiments were
also performed using water as the fluid phase. An experimental
setup similar to the one used for the metal foam and air experiments was used for this purpose (see Calmidi, 1998). The main
difference between the two was a thin plexiglass enclosure that
was used to contain the water within the void spaces of the metal
foam.
The experimental procedure was the same as that for air. The
main difference was in the time taken to reach steady state which
was around two hours. The measured thermal conductivity values
are listed in column 5 of Table 1.
Error Analysis: The main sources of error in the experiment
are due to the errors in measurement and those due to conduction
losses through the thermocouples and the insulating styrofoam. As
noted earlier, buoyancy effects are negligible since the heating is
from above. Also, the contact resistance due to the brazing can be
neglected because of the high effective conductivity of the brazed
layer. The effect of brazing is to introduce an error in the estimation of L the distance between the two aluminum plates. This
error is estimated to be one percent of the dimension. The errors in
the estimation of q and A, were 2 percent and 1.6 percent,
respectively. The combined error (thermocouple calibration and
the resolution of the data acquisition device) in the estimation of
AT, the temperature difference is 0.3C. However, the temperature
difference was varied up to a maximum of about 15C. Thus, the
percentage error in the estimation of AT is 2.0 percent. Based on
the conductivity of the styrofoam (0.029 W/m-K), the heat loss
through the insulation was estimated to be less than 0.5 percent of
MAY 1999, Vol. 121 / 467

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

water

Empirical correlation

j _
0.02

0.04

0.06

0.08

0.10

0.12

1-e
Fig. 3

Experimental data given in Table 1 plotted along with empirical correlation (Eq. (4))

the heat input. The heat conducted away by the thermocouples is

insignificant and is neglected. Based on the preceding errors, the
total error in the measurement of the thermal conductivity is
estimated, using the quadrature sum (Taylor, 1980), to be 3.5
percent. For the experiments conducted with water as the fluid
phase, there is an additional error introduced due to heat conduction through the plexiglass enclosure. Based on the thermal conductivity and the dimensions of the plexiglass enclosure, the total
error in the measurement of the thermal conductivity is increased
to 3.6 percent.
Empirical Correlation: Based on the experimental data collected, as described above, an empirical correlation was developed.
When the heat conducted through the solid and fluid phases is in
parallel, the effective thermal conductivity is given by Eq. (1).
Based on Eq. (1), the following equation was postulated:
^= e +A(l

- e)"'

(4)

The best fit was obtained for n = 0.763. However, the value of
A was found to be 0.181 for air and 0.195 for water. The maximum
and average absolute errors were found to be 6.9 percent and 3.7
percent for air and 7.5 percent and 3.1 percent for water, respectively. The experimental data in Table 1 along with the empirical
correlation given in Eq. (4) are shown in Fig. 3.

Analytical Model

As mentioned before, the structure of the metal foam consists of

dodecahedron-like cells with 12-14 pentagonal or hexagonal
faces. The edges of the cells are formed by the individual fibers.
We represent this structure by a two-dimensional hexagonal array
where the fibers are the edges of the hexagons (Fig. 4). The
lumping of material at the points of intersection of the fibers is
taken into account in the structure by a square. It is noted here that
the area of cross section of the fibers (and of the intersection) is
more important than the exact shape itself. Since the structure
shown in Fig. 4 is periodic, it is convenient to consider a unit cell
(Fig. 5). The conductivity of the unit cell is thus representative of
the effective thermal conductivity of the metal foam. This method
has been successfully used in the past (Kunii and Smith, 1960;
Zehner and Schlunder, 1970; Hsu et al, 1994; Hsu et al., 1995) to
determine the effective conductivity of packed beds. In general,
one-dimensional heat conduction is assumed in a two (or three)
dimensional periodic structure.
Effective Thermal Conductivity: We assume one-dimensional
heat conduction in order to derive an analytical expression for the
conductivity. The direction of heat flow is as shown in Fig. 5. Note
that consistent with the assumption of one-dimensional conduction,
the side faces are adiabatic. Although this assumption may not be true

Unit Cell

In this section, an analytical model is developed based on the

structure of the metal foam matrix. First, a periodic structure for
the metal foam is postulated and then one-dimensional conduction
analysis is performed in the periodic structure to derive the effective thermal conductivity.
Representation of the Metal Foam Structure: In reality, the
structure of the metal foam is a complex one and it would be
difficult to accurately capture its nuances to every bit of detail.
However, there is a fair degree of uniformity associated with the
structure which enables it to be treated in a semi-empirical manner.
468 / Vol. 121, MAY 1999

Fig. 4

Hexagonal structure of metal foam matrix

Transactions of the ASME

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

After substituting for "t" from Eq. (11), the resulting quadratic
equation in bIL can be solved for bIL as follows:

-r + Jrl +

Fig. 5

#'

1 +

(1
r 1+

(14)

bIL is plotted in Fig. 6 for different values of r. It should be noted

that for r = 0.366, the fiber extends upto the right edge of the
intersection (Fig. 5). Hence, Eq. (14) is not valid for r > 0.366.
Equation (14) can be used in Eq. (12) to obtain the effective
thermal conductivity of the unit cell in terms of the porosity "e"
and the area ratio, "r." The final expression can be written as

Unit-cell representation of hexagonal structure

/
locally, globally the heat transfer is indeed one-dimensional. For
determining the effective thermal conductivity, the unit cell in Fig. 5
can be divided into three layers in series. The conductivity of each
layer, in turn, is derived separately by applying the parallel law of
thermal resistances.
In layer 1, the solid and fluid phases are in parallel. Their
respective volumes are given by
Vs = t(L + b)w

and

(6)

"w" in Eqs. (5), (6) is the width in the third direction (perpendicular to the plane of the paper). The conductivity of layer 1 can be
written as
t(L + b)
t(3L)

t[3L - (L + b)]
k,
f(3L)

(7)

Equation (7) can be simplified as

k{ = kf+

(*, - kf)
5

1 +b

(8)

In a similar manner, the conductivities of layers 2 and 3 can be

written as
'b

kn = kf+l(ks
km = kf+

and

kf +

(9)

It

(10)

3~7J (ks~kf\z

b\(ks-

1 +

kf)

k,+
b

\\

I
kf +

(5)

V}= t[3L - (L + b)]w.

""-

(1-r)

*,=

4r

(ks - k.

3~J3

(15)

II

where bIL is given by Eq. (14).

To assess the validity of this model, we first consider the
experimentally measured values for metal foam and air given in
column 4 of Table 1. They are plotted in Fig. 7 as a function of
1-e, the solid fraction. Different symbols have been used for
samples of different pore sizes. Clearly, an excellent fit between
the experimental data and the predicted values is obtained for r =
0.09.
Figure 8 shows the effective thermal conductivity of foamed
materials (kjkf) as a function of the porosity for different values
of kjkf. Also shown, along with experimental data, are curves for
air (kjkf = 8226), and water (kjkf = 357). As expected, for
kjkf = 1, ke = kf for all porosity values. As kjkf increases, kjkf
also increases and all curves converge at kjkf = 1.

Discussion
As was seen in the previous section, the area ratio r = 0.09
results is excellent agreement with experiment results for both
aluminum/air and aluminum/water systems. Due to the two-

In Eq. (10), an "area ratio" r is defined as

0.40

tw
bw

(11)

By combining the three layers which are in series, the effective

thermal conductivity of the unit cell can be written as
L, + L2 + L i
kc.

i-i\

LJO

L-/-1 ,

kn

km

=++~
k\

(12)
0.20 -

where, ku kn, and km are given by Eqs. (8), (9), and (10), respectively, and L,, L2, and L3 are the heights of the three layers in
Fig. 5.
The solid volume fraction, 1-e, is the ratio of solid volume to
the volume of the unit cell. For the assumed hexagonal geometry,
it can be easily shown to be

t(L + b) + (b - t)2b + [L^-(l-e) =

Journal of Heat Transfer

7 ^
V3
3L L

0.30

bj

0.10

0.90

0.92

0.94

0.96

1.00

Porosity

d3)
Fig. 6

ML. (Eq. (14)) as a function of e

MAY 1999, Vol. 121 / 469

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

0.04

0.06

1-8
Fig. 7

Comparison of experimental results with model (Eq. (15)) for air (k, = 0.0265 W/m-K)

dimensional nature of the assumed geometry, the ratio of the

length scales of the fiber thickness to the intersection size is r1'2 =
0.3. A close examination of the metal foam structure indices that
r is not unique and it, in fact, varies. However, r"2 = 0.3 appears
to be a representative value. Due to the one-dimensional conduction analysis, it is expected that the value of r is slightly underpredicted since spreading effects are not included. Nevertheless,
these effects are expected to be small due to the high conductivity
of aluminum. The experimental data for both air and water suggest

000.0

1 '

'

' 1

that there is no systematic effect of the pore density variation on

the effective thermal conductivity. The implication is that the
structure of the metal foam is more or less constant over the range
of pore densities considered. This is in sharp contrast to the
experimental data for air-saturated polyurethane foams from the
Scott Paper Co. reported in the monograph (p. 137) by Kaviany
(1995). However, the reported conductivity values are higher than
the theoretical maximum given in Eq. (1). It is possible that
buoyancy and radiation effects may not have been negligible for

|<

ks/kf-10000

Data (air)

Data (water) _

^
100.0

^X\

ks/kf=1000

10.0

=
-

1.0

"

o^

"-*<K),_
ks/kf= 100

^ " " \

~--V. 9 ,

X ::

"~~~

ks/k(=10
ks/kf=1

^v

__is)

0.1
0.88

0.90

0.92

0.94
Porosity

0.96

0.98

1.00

Fig. 8 Effective thermal conductivity of foamed materials using Eq. (15) with r -- 0.09. Data for
aluminum/air and aluminum/water is also shown, with model predictions.

470 / Vol. 121, MAY 1999

Transactions of the ASME

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

low-conductivity polyurethane foams resulting in higher thermal

conductivity values at larger pore densities.
A comparison of the analytical (phase-symmetric) model of Hsu
et al. (1994) with our experimental data indicates that their model
overpredicts the thermal conductivity for metal foams. Further, the
error systematically increases with an increase in the solid fraction,
1-e. A possible explanation for this might lie in the extension of
the experimentally observed low solid conductivity asymptote
result of Zehner and Schlunder (1970) to the low fluid conductivity
asymptote by invoking phase symmetry among the solid and fluid
phases. Tien and Vafai (1979) have derived statistical upper and
lower bounds for the effective thermal conductivity of fibrous
insulations. Their three-dimensional model for randomly dispersed
fibers was also compared with the experimental data obtained in
this study. It was found that, in general, the data were close to the
upper bound and orders of magnitude different from the lower
bound. However, the recommended value of the cell geometry
factor, G = j , was found unsuitable for accurate predictions of the
thermal conductivity over the entire range of porosities considered.

Summary
The effective thermal conductivity of high-porosity fibrous
metal foams has been investigated experimentally. Experiments
have been conducted using both air and water as the fluid phases.
An empirical correlation has been developed based on the experimental data. A theoretical model has also been derived based on
the hexagonal structure of the metal foam matrix. The model
matches extremely well with the experimental results for both air
and water for an area ratio value of 0.09. It is expected that the
experimentally validated model for the thermal conductivity will
be helpful in the evaluation of metal foams as possible candidates
as heat sinks in electronic cooling applications.

Acknowledgments
This work has been funded by CAMPmode, The Center for
Advanced Manufacturing, and Packaging of microwave, optical,
and digital electronics at the University of Colorado. The help
received from Mr. Kyoung-Hwan Ahn for experiments with water
is gratefully acknowledged. The authors would like to thank Dr.
Yoichi Yokoyama for the many useful discussions and Mr. Bryan
Leyda of ERG (Energy Research and Generation), Inc., for providing the samples used in the experimental study and for helpful
discussions during the course of this work.
References
Bauer, T. H., 1993, "A general approach toward the thermal conductivity of porous
media," Int. J. Heat Mass Transfer, Vol. 36, No. 17, pp. 4181-4191.
Calmidi, V. V 1998, "Transport phenomena in high porosity fibrous metal forms,
" Ph.D. dissertation, University of Colorado, Boulder, CO.
Carbonell, R. G and Whitaker, S., 1984, "Heat and mass transfer in porous
media," Fundamentals of Transport Phenomena in Porous Media, Bear and M. Y.
Corapcioglu, eds., Martinus Nijhoff, Dordrecht, The Netherlands, pp. 123-198.
DuPlessis, P., Montillet, A., Comiti, J., and Legrand, J., 1994, "Pressure Drop
Prediction for Flow through High Porosity Metallic Foams," Chem. Engng. Sci., Vol.
49, pp. 3545-3553.
Hsu, C. T., Cheng, P., and Wong, K. W 1994, "Modified Zehner-Schlunder
models for stagnant thermal conductivity of porous media," Int. J. Heat Mass
Transfer, Vol. 37, No. 17, pp. 2751-2759.
Hsu, C. T Cheng, P., and Wong, K. W 1995, "A Lumped-Parameter Model for
Stagnant Thermal Conductivity of Spatially Periodic Porous Media," ASME JOURNAL
OF HEAT TRANSFER, Vol. 117, pp. 264-269.

Hunt, M. L., and Tien, C. L 1988, "Effects of Thermal Dispersion on Forced

Convection in Fibrous Media," Int. J. Heat Mass Transfer, Vol. 31, pp. 301-309.
Lee, Y. C , Zhang, W., Xie, W., and Mahajan, R. L., 1993, "Cooling of a FlipChip
package with a 100 Watt, 1 sq. cm chip," Proceedings of the ASME International
Electronic Packaging Conference, Vol. 1, pp. 419-423.
Kaviany, M 1995, Principles of Heat Transfer in Porous Media, Springer-Verlag,
New York.
Kunii, D., and Smith, J. M., 1960, "Heat Transfer Characteristics of Porous Rocks,"
AIChEJ., Vol. 6, pp. 71-78.
Maxwell, J. C , 1891, A Treatise on Electricity and Magnetism, Vol. 1, Oxford
University Press, reprinted by Dover, New York (1954).
Nozad, I., Carbonell, R. G., and Whitaker, S 1985, "Heat conduction in multi-

Journal of Heat Transfer

phase systems I: Theory and Experiments for two-phase systems," Chem. Engng Sci.,
Vol. 40, pp. 843-855.
Sahraoui, M., and Kaviany, M., 1993, "Slip and no-slip temperature boundary
conditions at interface of porous, plain media: Conduction," Int. J. Heat Mass
Transfer, Vol. 36, No. 4, pp. 1019-1033.
Taylor, T. R., 1980, An Introduction to Error AnalysisThe study of uncertainties
in physical measurements, University Science Books, Mill Valley, CA.
Tien, C. L., and Vafai, K., 1979, "Statistical Bounds for the Effective Thermal
Conductivity of Microsphere and Fibrous Insulation," AIAA Progress Series, Vol. 65,
pp. 135-148.
Zehner, P., and Schlunder, E. U., 1970, "Thermal conductivity of granular materials at moderate temperatures," Chemie. Ingr.-Tech, Vol. 42, pp. 933-941.
Yokoyama, Y., and Mahajan, R. L., "Non-Darcian Convective Heat Transfer in a
Horizontal Duct," Proceedings of the 30th National Heat Transfer Conference,
Portland, OR, HTD-Vol. 309, ASME, New York, pp. 83-91.

Dimensional Analysis in Heat Transfer

T. M. Dalton1 and M. R. D. Davies1
Introduction
The object of dimensional analysis is to reduce the number of
parameters in a problem for ease of calculation and experimental
economy. It follows that the best analysis will produce the minimum number of nondimensional groups. As these groups have
physical significance, the nondimensionalizing technique must always produce the correct scaling groups, and it might also be
expected that this method should be applicable to a wide range of
problems. The technique must, in short, be concise, correct, and
consistent.
The genesis of dimensional analysis is outlined by White (1994)
and Ensault-Pelterie (1950), it can be traced back to Fourier (1822)
through the work of Rayleigh (1877) and the Pi theorem of
Buckingham (1914).
Ehrenfest-Afanassjewa (1916) is attributed with first suggesting
the method of expressing the governing equations in nondimensional form to derive the governing nondimensional constants,
subsequently named inspectional analysis by Ruark (1935). It is a
modification of this method which is the subject of this paper. In
inspectional analysis each parameter in the governing equations is
nondimensionalized by a reference parameter. It is based on the
principle that two systems with identical governing equations and
boundary conditions will have identical solutions. This is the
condition of similarity. If the equations and boundary conditions
are written without dimensions, the principle still holds and great
generality is achieved. The method is now widely found in college
level texts. In all the published work, the reference parameters
include boundary conditions and therefore for each new boundary
condition type the process must be repeated. In this analysis, the
rescaled groups which arise from the governing equations are
separated from those arising from the boundary conditions. This is
shown to have a number of advantages over any previously reported techniques: First, if possible, it derives the reference parameters for each variable from the governing equations, thereby
avoiding any confusion in their choice; secondly, it represents a
straightforward method to produce the minimum number of scaling groups; thirdly, it reduces the effort of inspectional analysis to
just nondimensionalizing the boundary conditions; and, finally, it
1
PEI Technologies, Thermofluids Research Centre, Department of Mechanical
and Aeronautical Engineering, University of Limerick, Limerick, Ireland.
Contributed by the Heat Transfer Division for publication in the JOURNAL OF
HEAT TRANSFER and presented at '97 NHTC, Baltimore. Manuscript received by
the Heat Transfer Division, Mar. 6, 1998; revision received, Jan. 1, 1999.
Keywords: Analytical, Heat Transfer, Natural Convection, Scaling. Associate
Technical Editor: A. Lavine.

MAY 1999, Vol. 121 / 471

Downloaded From: https://heattransfer.asmedigitalcollection.asme.org/ on 01/14/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use