PSD Fly Ash

Download as pdf or txt
Download as pdf or txt
You are on page 1of 8

hr. J. Hrur Muss TrmJcr.

Primed m Great Britain

Vol. 36. No. 7. pp. lYO5-1912.

1993

M)l7-Y310/93$6.00+0.00
c I993 Pergamon Press Lid

The effects of particle size distribution


and
refractive index on fly-ash radiative properties
using a simplified approach
FENGSHAN

LIUt

Department of Fuel and Energy, University of Leeds, Leeds LS2 9JT, U.K.
and

JIM SWITHENBANK
Department of Mechanical and Process Engineering. University of Sheffield, Sheffield SI 3JD, U.K.
(Receioed

28 March

1992)

Abstract-The radiative properties of a fly-ash polydispersion are calculated using a simplified approach
based on Mie theory. The experimental data of fly-ash complex refractive index of Goodwin (Infrared
optical constants of coal slags, Technical Report T-255, Stanford University, California) are employed in
the calculation to take into account the wavelength-dependence of optical constants. The uncertainty in
representing the particle size distribution is addressed explicitly. Due to this uncertainty, the uncertainty
of the wavelength-integrated Planck mean absorption and scattering coefficients can be over 10%. The
use of wavelength-independent optical constants for fly-ash yields unacceptable results of Planck mean
coefficients.

tion of ash, ash loading and ash particle size dis-

1. INTRODUCTION

tribution.
to predict radiative heat transfer in pulDue to the lack of reliable and sufficient experverized coal-fired systems, it is essential to know the imental data, most of the up to date studieshave
radiative properties of combustion products. Indeed, assumedthat the optical constantsn and k are indeIN

ORDER

it has beenpointed out that accurate knowledgeof


the radiative propertiesof combustionproducts can
be as important as the solution method [I]. Except
radiation by water vapour and carbon dioxide, the
particles present in a coal-fired furnace, including
soot,coal, char andfly-ash, absorband emit radiation
in a continuousform covering the entire spectrum.It
is believed that the major contributors to the continuum radiation are the fly-ashparticlesbecausechar
and soot particlesare usually presentin a relatively
smallfraction of the entire furnace volume and therefore they are not important in evaluating the overall
heat transfer performance; however, fly-ash particles
exist in almostthe entire furnace volume [2].
In recentyears,due to the important role played by
fly-ashin radiative heattransferin coal-fired furnaces,
considerableattention has been paid to predicting
their radiative propertiesand their effectson radiative
heat transfer [3-71. Thesestudieshave indicated that
neglectof the ash radiation may underestimatetotal
heat transfer by up to 30%. It should be noted that
the predicted effectsof fly-ash particleson radiation
depend strongly on the absorption and scattering
coefficientsusedin the calculations.Thesecoefficients
arein turn dependenton the complexindex of refrac-

t Present address : Department of Chemical Engineering,


Queens University, Kingston, Ontario, Canada K7L 3N6.

pendentof wavelength,with n taken in eachcaseas


1.5 and k ranging from 0.005to 0.05 [4-6, 81. Only
recently, Goodwin [9] haspresentedsomemorecompleteoptical constantsfor homogeneous
bulk samples
with compositionssimilar to thoseof fly-ash. In the
wavelengthrangefrom 0.5 to 12pm which is of interestat furnacetemperatures,their resultsindicatethat
the real part n does not exhibit strong wavelengthdependence;whilst the imaginary part k varies by
severalordersof magnitude.
By assumingfly-ashparticlesarehomogeneous
and
spherical,the radiative propertiesof a singleparticle
can be obtained from Mie theory [lo]. However, the
application of Mie theory is neither simplenor practical for engineeringcalculations.Fortunately, a simplified approach derived by van de Hulst [lo] can be
readily usedto calculate the radiative properties of
fly-ashparticlesassuggested
by Mengucand Viskanta
[6]. Almost without exception,it isassumedthat each
particle in the furnace scattersand absorbsradiation
unaffectedby the presenceof other particles.Then the
absorptionand scatteringof radiation by the particle
suspension
iscalculatedby a simplealgebraicaddition
of the radiation absorbed and scattered by each
particle. By expressingthe size distribution of a flyashpolydispersionas a three-parameterskeweddistribution function [I 11,Mengucand Viskanta [6] were
able to obtain a simpleclosedform solution for the
spectralextinction and absorption coefficientsusing

1905

F. LIU and J.

1906

SWITHENBANK

NOMENCLATURE
a, b

DlO
D 20

1,
k

m
M
Mi
n

Qa

parameters in particle size distribution


function
particle diameterkm]
size-weighteddiameterkm]
r.m.s.diameterbrn]
black-body radiation intensity
[w m-* sr-1
imaginary part of refractive index
complexrefractive index, n - ik
total numberof particlesanalysed
differential numberof particles
real part of refractive index
particle numberdensity [m-1
absorptionefficiencyfactor
extinction efficiency factor
scatteringefficiencyfactor

the simplified approach. However, one of the three


parametersin the particle size distribution function
was assignedarbitrarily in the work of Menguc and
Viskanta [6].
In this paper, the effect of the undeterminedparameterin the particle sizedistribution on the fly-ash
radiative properties is studied using the simplified
approach. The wavelength-dependenceof fly-ash
optical constantsis taken into account by employing
the experimental data of Goodwin [9]. Numerical
results show that the undeterminedparameter can
causemore than 10% uncertainty in the Planckmean
absorption and scattering coefficients.A new condition is suggested
to determinethe arbitrary constant
in the particle sizedistribution function usedby Mengut and Viskanta [6]. This work also demonstrates
that the Planck meancoefficientscalculatedusingthe
wavelength-dependentoptical constants differ significantly from those using wavelength-independent
optical constants.
2. FLY-ASH

n = 1.5-0.35(1-6.0)
n = 0.8+0.5(1-8.0)
n = 2.3-0.5(1-11.0)

0.5 -e 1 c 6.0 pm
6.0 < L c 8.0pm
8.0 < 1 < ll.Opm
11.0< 1 < 12.0pm.

The imaginary part k is given as


k =

lo-4.6+2.2(1-0.5)

Greek symbols
ci
parameterin particle sizedistribution
function
absorptioncoefficient [m- 1
Ka
KC extinction coefficient[m- 1
KS scatteringcoefficient [m- 1
L
wavelengthbm].
Subscript
1
spectralquantities.
Superscript
*
meansingleparticle properties.

k = ~CJ-~.~
k = lo-3.5+(A-4.0)
k =

0.5 < I < 1.0pm

1.0 < I <4.0pm


4.0 < 1 < 5.0 pm
5.0 < 1 < 7.5 pm

lo-2.5+0.24"-50'

k = lo- 1.9+l.S(i-7.5)

7.5 < 1 < 8.5 pm


k = 10-O.
8.5 < 1 < 10.5pm
k = 10-~1-o~733(A-o~5)
10.5< 1 < 12.0pm.
2.2. Particle size distribution
The fly-ash particles collected from a small-scale
pulverized coal-fired furnace [12] are consideredin
this work. The particle sizedistribution, analysedby
a Coulter counter, is summarizedin Table I. In this
table, f(D) is defined
f(D)

= &

x 100

(1)

where M is the total number of particles analysed.


The meanfly-ash particle diametersD,, and D20 are
calculatedas

MODEL

2.1. Optical constants


The optical constantsof fly-ashpresentedby Goodwin [9] (seealsoFig. l(b) in ref. [7]) are employedin
the presentcalculations.For the purposeof computer
modelling,their experimentaldata of fly-ash optical
constant are approximated by piecewiselinear functions of wavelength.The real part n is written as
n= 1.5

temperature[K]
particle sizeparameter.

T
X

TableI. Ashparticlesizedistribution
Cumulative Differential ADi
volume(%) No., M,
@m)
4
5.04
6.35
8.00
10.1
12.7
16.0
20.2
25.4
32.0
40.3
50.8
64.0
80.6

100
98.3
94.8
90.0
83.3
72.6
60.6
48.5
34.3
19.5
8.2
4.0
2.4
0.0

63 087
44035
32 778
23 975
16081
9498
5155
3045
1577
629
110
21
6
3

4.52
1.175
1.48
1.875
2.35
2.95
3.75
4.7
5.9
7.45
9.4
11.84
14.9
-

gi o
6.98
18.74
11.07
6.39
3.42
1.61
0.69
0.32
0.13
0.04
0.0
0.0
0.0
-

Particle size distribution and refractive index


Table 2. Parameters of particle size
distribution
a

2
3
4
5

0.04293
0.01996
0.00896
0.00393

0.44118
0.58824
0.73529
0.88235

D
D

-Cf(Di)DiADi
10xf(Di)ADi

= 6.80 pm

(2)

_ Zf(Di)D;ADi I'*
= 7.85 ,um.
10 xf(Di)ADi >

(3)

1907

sitive to the value of a. The larger the value of a, the


largerthe maximumoff(D) and the particle diameter
at which the maximum occurs.A larger a, however,
yields a smaller particle diameter at which the
maximumof D'f (D) occurs,seeFig. 1(b).
In the prediction of radiative propertiesof a polydispersion,the size-weightedmeandiameterD ,. and
the area-weightedmeandiameterD2,, are two important parameterssincethe size of the particles characterizessingleparticle propertiesand the areaof the
particles is requiredin evaluating the polydispersion
properties.In this work, the following additionalcondition is suggestedto determineuniquely the value
ofa

The particle sizedistribution given in Table 1 can


be expressedin a more concisefunctional form by
usinga three-parameterskeweddistribution function
[6, 111

a(a+2)! = D2
20

bf3

i.e. the mean diameter D,, of the size distribution


function, equation (4), is assumedto be equalto the
f(D)= aD" exp (-bD).
(4)
measuredvalue. The values of 6, a and a can be
In the work of Menguc andViskanta [6], CIwastreated evaluatedfrom equations(5) to (7) suchthat
as a positive integer and its value wasassignedarbitrarily; whilst the constantsa and b were calculated
from overall mean diameter D, o and the normalization of f(D), i.e. the constantsa and b obey
the following relations:
(9)
b+
a=-.

a(cr+l)!
---=
b+2

D .

(6)

10

The values of a and b can be determinedonly if the


value of a is assigned.For a = 2, 3, 4 and 5, the
corresponding
valuesof a and b are given in Table 2.
The effect of a on the particle sizedistribution isshown
in Fig. 1. The sizedistribution function f(D) is sen-

(10)

a!

For the particle size distribution given in Table 1, it


is found that
a = 2.0158
u = 0.04356
b = 0.4433.

(b)
a=5

16
6

10

15
D,

pm
FIG.

20

25

30

10

15
D,

1. Effect of OTon the particle size distribution.

pm

20

25

30

F. LIU

1908

and J.

SWITHENBANK

It is seen that this size distribution follows closely that


with c( = 2 given in Table 2.
3. SPECTRAL

RADIATIVE

PROPERTIES

3.1. Single particle characteristics

Basedon Mie theory, the radiative characteristics


of a singleparticle is determinedby the particle size
parameter, x = KD/I, and the complex refractive
index, nz= n - ik.
As derived by van de Hulst (Sections 11.22and
11.23 in ref. [IO]), the extinction and absorption
efficiency factorscan becalculatedas

0.01
0.006
0.003

cos B .

Qe=2-4exp(-~tanp)--

p sin (P-P)
0

-4exp (-p tan/l)

FIG. 2.

+4 cosp
2cos2j
~
( P >
Q, = 2K(4xk)

10

12

A, wa

cm (P - 28)

AbsorptionefficiencyfactorQ,. D

= D,,.

(11)
(12)

where
tan/I=&,

xc-

p=2x(n-I),
--

e-:-

XD
L

K(z)=:+~+~.

2.5-1

di 2.0The scatteringefficiency factor QI is then calculated


by subtracting Q, from QC.Essentially,theseequa1.5;
$
tions areapplicableover the entire domainof the Mie
theory if the real part of the refractive index n is close
LOto 1.Oand even for valuesof n aslarge as2.0 and the
imaginarypart k negligiblysmall.
0.5
It is clear that the complex refractive index of flyash presentedby Goodwin [9] meetsthe conditions
0.0
. , I , I ,
, . , .
required by the simplifiedexpressionsfor extinction
2
4
6
8
10
12
0
and absorption efficiency factors except for the
imaginary part k in the wavelength range 8.0 <
FIG. 3. Scattering
efficiencyfactor Ql. D = D,,
1< ll.Opm.
The absorptionand scatteringextinction efficiency
factors for a particle having diameter D = D ,,, are
shownin Figs.2 and 3, respectively.The featuresseen
in the absorption efficiency spectrumare similar to respondingefficiency factor, N the particle number
those in the spectrumof the imaginary part of the density and f(D) the normalized particle size disrefractive index (seeFig. 1(b) in ref. [7]).
tribution function given in equation (4).
Dividing equation (14) by N, the following
3.2. Spectral coejicients of thefly-ash polydispersion
expressions
are obtained:
The spectralabsorption, scatteringand extinction
coefficientsof a cloud of fly-ash particlescanbe evalu* -!%dD (15)
GA
- N- I 0m QC(D,l,m)~f(D)
ated from

BA(WN) = omQ,@, 1, ml ~f(D)N


I

dD (14)

Q~(D,i,m,~f(D)

dD. (16)

where /?A stands for either spectral absorption Substitutionof equations(4), (11) and (12) into equacoefficient K~*, spectral scattering coefficient Key, or tions (15) and (16) yields the closedform expressions
spectral extinction coefficient Key. Qp is the cor- for K$ andK$ after somemathematicalmanipulations

Particle size distribution and refractive index


K:A

1909

(25, -4 cos /Y?J,-4 cos2 PJ,


+4cos2/?cos2~J,)
K,:

y(J,

(17)

+2CJ,+2C2J6)

(18)

where
J, = &cr+3)

(19)

(2+a) arctan g -/3


J2 =

cos (a+ 1) arctan $ -2/l


J, =

1
1

r(cr+2)

/f(~+~2)2

(20)

T(a+ 1) (21)

,42(,42+g2)l/2lQ+I)

FIG. 5. Effectof OL
on thespectralscattering
coefficient
~3..

1
--(a+
J4=Ab+l

1)

(22)

J5 =

(23)
?
$1 100
X

3 a0
00
.$ 60,
A = wn- 1)
1

B&+7,

C=&.

(25)

Results of spectral absorption, scattering and


extinction coefficientsfor the fly-ash polydispersion
aregiven in Figs.4-6, respectively,for different values

Ii
';;i 40,
3
K 20,
VI
0.

FIG. 6. Effectof u on thespectral


extinctioncoefficient~2,.

3
B
x 4o

0
0

10

12

A, Pm
FIG. 4. Effectof a on thespectral
absorptioncoefficient~2.

of a. It can be seenthat the fine ripple structure exhibited by the Lorenz-Mie solutionsat short wavelength (1 < 2.5 pm) for a singleparticle of diameter
DIO, seeFig. 3, is cancelledout by integrating over
the particle sizedistribution. The fly-ash particlesare
weakly absorbingat 1 c 5 pm and becomestrongly
absorbing at longer wavelengthswhich reflect the
spectral variation of the imaginary part k. For the
particle sizedistribution consideredin this work, the
spectral scattering and extinction coefficients first
increasegradually till reaching maxima at 6.3 pm,
then decreasesharply to minima at 7.5 pm. At this
wavelength, the scatteringcoefficient is nearly zero.
Beyond 7.5 pm, the scattering and extinction
coefficientsdiffer significantly from each other since
the absorptioncoefficient is large in this wavelength

1910

F. LIU and J.

region. It should be noted that these features of the


spectral absorption, scattering and extinction co;
efficienis are similar to those presented by Goodwin
and Mitchner [7] based on Mie calculations.
The effect of cI on the spectral absorption coefficient
is negligible at short wavelengths, 1 < 5 pm, and
becomes significant at longer wavelengths, see Fig. 4.
While the effect of a on the scattering and extinction
coefficients is considerable at wavelengths 1 < 6 pm,
see Figs. 5 and 6. At long wavelengths, 1 > 8.5 pm,
the extinction coefficient becomes sensitive again to
the value of CLThe smaller the value of CI,the larger
the spectral absorption, scattering and extinction
coefficient due to the stronger weighting of large particle diameters, where Dif(D) is high (see Fig. l(b)).

SWITHENBANK

4 4.5
X
-I
2 4.08
0
. 3.52

8 3.02

d 2.58
a 2.03
g

1.5-

iz

4. PLANCK

MEAN

PROPERTIES

For most of the engineering calculations, the


wavelength-integrated
absorption and scattering
coefficients are required since it is not always possible
to perform radiative heat transfer calculations on a
spectral level. The Planck mean coefficients defined
below are of interest when the medium under consideration is optically thin

where ZbL(7) is the Planck spectral black-body intensity. These integrations can be performed numerically
by dividing the entire wavelength spectrum into a
number of bands such that

T, K
FIG. 7. Effect of a on the Planck mean absorption coefficient
K,*.

Y2,

2000

T, K
K: = f &b.i(T)A~i
i= I

f L0-W.i
I i= I

(28)

FIG. 8. Effect of a on the Planck mean scattering coefficient


K,*.

K: = i K?&.i(T)A~i
-f: L&Wi.
i= ,
I i= I

(29)

The wavelength summations in equations (28) and


(29) are performed numerically for the wavelength
range from 0.5 to 12.0 pm. The fraction of the total
black-body flux contained in this range is 0.971 at
1200 K and 0.990 at 1800 K with the excluded fraction
primarily on the long wavelength side.
Numerical results of the Planck mean absorption
and scattering coefficients obtained by using different
values of a are shown in Figs. 7 and 8, respectively.
The Planck mean absorption coefficient decreases
with increasing temperature due to the stronger
weighting of short wavelengths where the spectral
absorption coefficient is low (see Fig. 4). The Planck

mean scattering coefficient increases with increasing


temperature and is less temperature-dependent than
the Planck mean absorption coefficient. It is worth
noting that the variation trend of the Planck mean
absorption coefficient with temperature is opposite to
that presented by Viskanta et al. [4] using wavelengthindependent optical constants ; however, it is in agreement with the results of mean single-particle total
emissivity calculated by Goodwin and Mitchner [7l.
The figures also show that both the absorption and
the scattering coefficients are sensitive to the value of
CL.As expected from Figs. 4 and 5, the smaller the
value of a, the larger the Planck mean absorption and
scattering coefficients. By varying the value of a from

Particle size distribution and refractive index

1911

.,

m=L5-0.0015i

$1
9%
X

y(lI
1200

1400

1600 2000

1600

T. K

1200

m=L5-0.00%

1-I
1400

1600

1800

2000

T, K

FIG. 9. Comparison of the Planck mean absorption


coefficient K,* based on wavelength-independent indices with
that based on wavelength-dependent data.

FIG.
1I. Comparison of the Planck mean extinction
coefficient K: based on wavelength-independent indices with
that based on wavelength-dependent data.

2 to 5, there is more than a 10% decrease in both the


Planck mean absorption and scattering coefficients.
Calculations
have also been performed
for
wavelength-independent
optical
constants
of
m; = 1.5-iO.005 and m2 = IS-iO.0015.
The results
are compared with those obtained using spectral
optical constants in Figs. 9-l 1. The particle size distribution with a = 3 (see Table 2) is used in these
calculations. Figure 9 shows that the use of k = 0.005
overestimates the Planck mean absorption coefficient,
while the use of k = 0.0015 yields a much closer result
to that based on wavelength-dependent
optical con-

stants, except that they give different variation trends


with temperature.
In opposition to the effect of k on the mean absorption coefficient, the lower imaginary part k results in
higher mean scattering coefficient, see Fig. 10. The
smaller k (0.0015) over-predicts the mean scattering
coefficient especially at temperatures
7~ 1600 K,
while the larger k (0.005) overestimates the mean
scattering coefficient at T < 1400 K and then underpredicts the mean scattering coefficient at higher temperatures.
Figure 11 shows the Planck mean extinction
coefficients. It can be seen that the extinction
coefficient calculated using spectral optical constants
is almost constant
in the temperature
range
considered. The Planck mean extinction coefficient
based on wavelength-independent
optical constants
decreases with increasing temperature. In addition,
the mean extinction coefficient is very insensitive to the
value of the imaginary part k due to the insensitivity of
the extinction efficiency factor to the value of k [12].
The results shown in Figs. 9-l 1 indicate that it is
not acceptable to ignore the wavelength-dependence
of optical constants in the prediction of fly-ash radiative properties. The average value of the imaginary
part of the fly-ash complex refractive index k = 0.012
estimated by Gupta and Wall [13] is definitely too
high to calculate the fly-ash radiative properties.

5. CONCLUSIONS
I

!Ol I

1400

1600

1800

21

T, K
10. Comparison of the Planck mean scattering
coefficient K: based on wavelength-independent indices with
that based on wavelength-dependent data.
FIG.

A simplified approach based on Mie theory has


been employed to calculate the radiative properties of
a fly-ash polydispersion.
The experimental
data of
Goodwin [9] were used in the calculations to take into
account the wavelength-dependence
of fly-ash optical

1912

F. Lru and J. SWITHENBANK

constants. A new condition is suggested to determine


uniquely the value of CI present in the particle size
distribution based on the consideration of the important role played by the mean particle diameter D,, in
evaluating the radiative properties of polydispersions.
Numerical
results show that the simplified
approach is able to produce the important features of
those based on Mie calculations presented in ref. [7].
The effect of the value of a on the fly-ash radiative
properties is studied. The results indicate that the
arbitrariness
of a can easily cause more than 10%
uncertainty in the Planck mean absorption and scattering coefficients. Based on the work of Menguc and
Viskanta [14, 151, such uncertainty in radiative properties can give rise to significant uncertainty in the
prediction of radiative heat transfer. Therefore, it is
of importance
to establish a criterion to eliminate
the uncertainty in the radiative properties of fly-ash
particles due to their size distribution.
By comparing the results of using wavelength-independent optical constants with those using measured
fly-ash optical constants, it can be concluded that it is
not acceptable to ignore the wavelength-dependence
of fly-ash refractive index. Moreover, the previous
studies employing n = 1.5 and k ranged from 0.005 to
0.05, seem to overestimate the Planck mean absorption coefficient and to underestimate the Planck mean
scattering coefficient of fly-ash particles.

REFERENCES

2. A. G. Blokh, Heal Transfer in Steam Boiler Furnaces,


Chap. 3. Hemisphere, Washington, DC (1988).
3. T. F. Wall. A. Lowe. L. J. Wibberlev. T. Mai-Viet and
R. P. Gupta, Fly ash characteristics-and radiative heat
transfer in pulverised-coal-fired furnaces, Cornbust. Sci.
Technol.

(1981).

Phys. 17, 1107-l

114 (1984).

9. D. G. Goodwin, Infrared optical constants of coal slags,


Technical Report T-255, Stanford University, California
(1986).
10. H. C. van de Hulst, Light Scafrering by Small Parficles.
Wiley, New York (1957).
I I. C. L. Tien, D. G. Doornink and D. A. Rafferty, Attenuation of visible radiation by carbon smokes, Cornbust.
Sci. Technol. 6, 55-59 (1972).
12. F. Liu, New development in pulverized coal combustion :
numerical modelling of radiative heat transfer and experimental test of a novel burner technique, Chap. 7, Ph.D.
Thesis, University of Sheffield, U.K. (1990).
13. R. P. Gupta and T. F. Wall, The complex refractive
index of particles, J. Phys. D: Appl. Phys. 14, L95-98
(1981).
14. M. P. Menguc and R. Viskanta, A sensitivity analysis for
radiative heat transfer in a pulverized coal-fired furnace,
Cornbust.

I. R. Viskanta and M. P. Mengnc, Radiative heat transfer


in combustion systems, Prog. Energy Combusl. Sci. 13,
97-160 (1987).

26, 107-121

4. R. Viskanta, A. Ungan and M. P. Menguc, Predictions


of radiative properties of pulverised coal and fly-ash
polydispersions, ASME paper 81-HT-24 (1981).
5. R. P. Gupta, T. F. Wall and J. S. Truelove, Radiative
scatter by fly ash in pulverised-coal-fired furnaces : application of the MontdCarlo method to anisotropic scatter,
Int. J. Hear Mass Transfer 26. 1649-1660 (1983).
6. M. P. Menguc and R. Viskanta, On the radiative properties of polydispersions : a simplified approach,
Cornbust. Sci. Technol. 44, 143-l 59 (1985).
7. D. G. Goodwin and M. Mitchner, Flyash radiative properties and effects on radiative heat transfer in coal-fired
systems,Inf. J. Heat Mass Transfer 32,627438
(1989).
8. S. A. Boothroyd and A. R. Jones, Radiative transfer
scattering data relevant to fly ash, J. Phys. D: Appl.

Sci. Technol.

51, 51-74

(1987).

15. M. P. Menguc and R. Viskanta, Radiative transfer in


axisymmetric, finite cylindrical enclosures, J. Hear
Transfer 108,271-276
(1986).

You might also like