125

Ind. Eng. Chem. Fundam. 1986,25, 125-129

Subscripts

Matljevic, E.; Couch, J. P.; Kerker, M. J . Phys. Chem. 1962, 66, 111.
Matljevlc, E.; Mathal, K. G.; Otterwill, R. H.; Kerker, M. J . Phys. Chem. 1961,
65, 826.
Matljevlc, E.; Tezak, B. J . Phys. Chem. 1953, 57,951.
Napper, D. H. Trans. Faraday Soc. 1968, 6 4 , 1701.
Napper, D. H. J . Colloid Interface Sci. 197Oa, 32, 106.
Napper, D. H. J . Colloid Interface Sci. 197Ob, 33,384.
Napper, D. H.; Netschey, A. J . Colloid Interface Sci. 1971, 37,528.
Rles, H. E.; Meyers, B. C. Science 1968, 760, 1449.
Sarkar, N.; Teot, A. S. J . Colloid Interface Sci. 1973, 43,370.
Saunders, F. L.; Saunders, J. W. J . Colloid Sci. 1956, 77, 260.
Senju, R. "Koroldo Tekitei-ho"; Nankodo: Tokyo, 1969.
Sommerauer, A.; Sussmann, D. L.; Stumm, W. Kolloid-Z. 1968, 225, 147.
Stumm, W.; O'Melia, C. R. J . AWWA 1968, 60,514.
Suk, V.; Malat, M. Chemist-Analyst 1958, 4 5 , 30.
Suzuki, A,; Kashiki, I . Kagaku Kogaku Ronbunshu 1982, 6 , 728 (in Japanese).
Vincent, B. Adv. Colloid Interface Sci. 1974, 4 , 193-277.

a = anions taking part in colloid formation

c = cations taking part in colloid formation
Registry No. PVSK, 26837-42-3; MGCh, 88650-88-8; AgI,
7783-96-2; Zn(OH)2,20427-58-1; A1(N0J3, 13473-90-0; Hi-floc
(SRU), 32037-29-9; Hi-floc (co-poly), 82013-86-3.
Literature Cited
Bratby, J. "Coagulation and Flocculation"; Uplands Press: Croydon, U.K.,
1980a; p 142.
Bratby, J. "Coagulatlon and Flocculation"; Uplands Press: Croydon, U.K.,
1980b; p 84.
Flory, P. J. "Principles of Polymer Chemistry"; Cornel1 University Press: New
York, 1953; Chapter 13.
Kashikl, I.; Suzuki, A. Kagaku Kogaku Ronbunshu 1982a, 6 , 722 (in Japanese).
Kashiki, I.; Sauzuki, A.; Gotoh, K. Kagaku Kogaku Ronbunshu 1982b, 8 , 7 3
(in Japanese).
La Mer, V. K. J . Colloid Sci. 1964, 19, 291.
Matljevlc, E.; Abramson. M. B.; Schulz, K. F.; Kerker, M. J . Phys. Chem.
1960, 6 4 , 1157.

Received f o r review May 31, 1984

Accepted April 29, 1985

Porosity Estimation from Particle Size Distribution

Norlo Ouchlyama
Natlonal Industrial Research Institute of Kyushu, Tosu, Saga-ken, Japan

Tatsuo Tanaka"
Department of Chemical Process Engineering, Hokkaido University, Sapporo, Japan

The packing porosity of a bed of randomly placed spherical particles is generally described as a function of the
size distribution of particles. This treatment takes into account the probable effect of macropores on the mixture
porosity. For mixtures having the Gaudin-Schuhmann size distribution of particles, computed estimates of the
average porosity are found to agree in general with known data.

Introduction
Packing problems of solid particles are frequently encountered in a wide field of science and technology. Both
the packing porosity and the number of contacts between
neighbor particles have an essential relation to the material
and the process properties of solid particles. We have
already proposed a simple expression for the number of
contacts (Ouchiyama and Tanaka, 1980).
Many experimental studies have been done of the
fractional void volume of a bed of solid particles, and it
is a well-known empirical fact that the packing porosity
varies with the size distribution of the materials involved.
From a theoretical point of view, on the other hand, much
work has been devoted to the regular packings of solid
spheres, and a few investigators such as Furnas (1931),
Westman and Hugill (1930), Tokumitsu (1964), and Kawamura et al. (1971) have examined the mixture porosity
of a bed of solid particles of different sizes.
Until now, however, nobody has succeeded in generally
describing the mixture porosity as a function of the size
distribution of particles.
Recently we have proposed a theoretical formula to
estimate the packing porosity of a mixed bed from the size
distribution function of the solid particles (Ouchiyama and
Tanaka, 1981, 1984). We have tested the theory with
mixtures of the binary, ternary, Gaudin-Schuhmann,
Gaussian, and log normal particle size distributions.
0196-4313/86/1025-0125$01.50/0

General agreement has been observed between the theory

and experiments. For the continuous size distribution of
particles such BS that of Gaudin and Schuhmann, however,
some remarkable deviations of computed values were
recognized from experiments. It was inferred that the
existence of macropores would have caused the discrepancy. As to the discrete size distribution of particles, on
the other hand, the effect of macropores has satisfactorily
been taken into the theory.
This paper presents a modified relation regarding the
effect of macropores on the packing porosity. This
treatment enables one to estimate more precisely than
before the packing porosity of a mixture having an arbitrary particle size distribution. The theory will be applied
to mixtures having the Gaudin-Schuhmann size distribution.
Theoretical Treatment
Porosity Estimation Based on a Simplified Model.
Our theoretical consideration starts with a simplified model
for the coordination number. See Figure 1. A central solid
sphere of diameter D is in direct contact with a fixed
number of surrounding spheres, each having the average
diameter D. Introducing the surface porosity, eA, which
means the void area fraction on a spherical surface of
diameter D + D and which is assumed to be independent
of the sizes of the particles involved, the coordination
0

1986 American Chemical Society

120

Ind. Eng. Chem. Fundam., Vol. 25, No. 1, 1986

Figure 1. A simplified model for the coordination number.

Hatching calls attention to a spherical shell considered in the theory.

number, C(D),around the central particle of diameter D

is exactly expressed as

where

D = $bf(D) dD

porosity of a mixed packing without macropores.

Restrictions on t h e Minimum B u l k Volume Attainable in a Mixed Packing. If macropores exist, the
above theory would underestimate the total bulk volume
of the space in a packing. In order to consider the probable
effect of macropores, we introduced an assumption which
imposed restrictions on the minimum bulk volume attainable in a mixed packing. In our previous paper (Ouchiyama and Tanaka, 1984),we imagined that some constituent particles of a mixture can be embedded in the
packing interstices between the other residual particles.
Thinking the matter over, however, we realize that the
larger particles cannot be embedded in the interstices of
the isolated packing of smaller particles although the reverse seems possible. We now assume that a multicomponent mixture never has a packed bulk volume smaller
than the total volume of both the pore space and the solid
space, VB,when smaller particles are taken away from the
original mixture in question. On the other hand, the total
volume above, VB,cannot be smaller than the bulk volume
computed from the theory for a bed without macropores.
Therefore, the following inequality should stand for an
arbitrary size of D.

Here, f ( D ) is the number-frequency size distribution of

particles and D, and D1 are the smallest and the largest
sizes of particles, respectively.
According to the above model, each hypothetical sphere
of diameter D D in a packing has to share, in part, the
space in common with the other hypothetical spheres. The
space commonly occupied lies within the spherical shell
having inner and outer diameters equal to ( D D ) and
( D D),respectively. Here, the abbreviation ( D D ) is
defined as
( D D ) = 0 for D d D
(34
=D-D forDID
(3b)
and the volume of the spherical shell, V,(D), is given as

vBT

D)3 - T ( D D)3
(4)
6
6
If we let the average number of hypothetical spheres taking
part in the common occupancy be ii, then the total volume
of the space allocated to a specified solid sphere of diameter D , V,(D), can be expressed as
V m ( D )= -(D

7T

V,(D) = s ( D

- D)3 + Vm(D)/17i

(5)

Where no macropores exist, we can represent the total

bulk volume of the space in the packing, V B T , as

VBT=

L, V , ( D ) N f ( D )dD
Dl

(6)

where N is the total number of particles in the packing.

Here, macroporemeans the void space which could exist
away from solid surfaces a t a distance greater than half
an average diameter of the particles. The overall average
volume porosity of the packing, Z, should therefore be
estimated from
VBT
-1
- (7)

Without going into detail, we want to point out that the

value of ii in eq 5 can be evaluated approximately, as
summarized later, when its size dependence is discarded.
Equation 7 is a general expression for the overall average

(8)

D, d D 5 D1

--

where

7T

L,V,(D)NY(D)d~
D1

D,

~ Y D=) MD)/s:~(D)
D

(94

d~

(9c)

Note that when V,(D) of inequality 8 is evaluated from eq

5, the values of ti and D should be based on the above size
distribution function, f(D),instead of f ( D ) in the related
equations. Inequality 8 places restrictions on the minimum
bulk volume attainable in a mixed packing. The existence
of macropores is suggested when eq 6 gives a value smaller
than the above minimum volume. In such a case, the
minimum value should be adopted as the total bulk volume
of the packing.
Summary of t h e Theoretical Relations. Equation
6 for no macropores is included in inequality 8, and all the
size variables of the theory can be expressed in a dimensionless form by defining
x = D/D,; k = D,/D,; x = D/DI
(10)
Denoting the number-frequency size distribution of particles as a function of the dimensionless size x by g ( x ) , the
theoretical relations are summarized as follows:

- 11-Z

for arbitrary x; k 5 x 5 1 (11)

where

Ind. Eng. Chem. Fundam., Vol. 25, No. 1, 1986 127

I '

Equation 12a is a rearranged version of eq 16a of our

previous paper (Ouchiyama and Tanaka, 19811, where zo
is the porosity value for a packing of uniformly sized
spheres.
For the discrete size distribution of particles, the above
equations can be rewritten. By numbering the size components of particles from the largest to the smallest in a
mixture, we obtain

- 1- 1-S

0.2

i=l

for arbitrary p ; 1 I p Im (13)

where

0,4

0,6 0.2 0,4 0.6 0,2 0.4 0,6

Distribution Constant, 2'
Figure 2. Effects of the distribution constant, q, on the overall
average porosity, z (replotted from experiments of Kawamura et al.
(1973)). A-2, B-2, and C-2 mean the three different powders used.
Maximum sizes were 2.98, 4.17, 6.33, and 8.50 mm. Minimum sizes
were fixed (-0.08 mm) but not specified.

1u,

2"

.-w

e
Here, m is the total number of size components of particles
in a mixture and gi is the fractional number of particles
of ith component.

0.000683
0.001367
0.002737

Discussion
First, the new restrictions represented in this paper will
be discussed in connection with binary and ternary mixtures of particles which were already examined in our
previous papers (Ouchiyama and Tanaka, 1981,1984). For
a binary mixture (m = 2), eq 13 is reduced to the relations
corresponding to p = 1 and 2. The relation for p = 2 is
the one for the packing without macropores. The relation
for p = 1 is rewritten as

0,2

0.4
0,6
0.8
Distribution Constant, a

1.0

Figure 3. Effects of the distribution constant, q, on the overall

average porosity, z (replotted from the packing data reported by
Dinger et al. (1982)).

Next, let us apply the theory to mixtures having the

Gaudin-Schuhmann size distribution of particles. The
number-frequency size distribution function of the dimensionless size x is
1

g(x) = x q - ' / L
(15)
where u1 is the fractional solid volume of the larger particles. This restriction completely agrees with eq 18 of our
previous treatment (Ouchiyama and Tanaka, 1981),which
was satisfactorily used to account for the effect of macropores in a binary mixture. For a ternary mixture ( m
= 3), eq 13 presents two relations for p = 1and 2 in addition to the one for no macropores (p = 3). According
to our preceding theory (Ouchiyama and Tanaka, 1984),
on the other hand, the mixtures of six combinations of the
size components had to be additionally examined in the
ternary mixture, three for m = 3 and three for m = 2.
Using eq 13 together with eq 14a and 14b, we estimated
packing porosities for the mixtures corresponding to Figures 4 and 7 of our previous paper (Ouchiyama and Tanaka, 1984). AU the computed values coincided with those
already reported. This suggests that the previous restrictions included some redundant conditions with regard
to the effect of macropores.

xq-*

dx

(16)

We tried in our first paper (Ouchiyama and Tanaka, 1981)

with this size distribution of particles to estimate packing
porosities by discarding the effect of macropores. In the
higher range of q values, combined with the smaller size
ratio of k,however, we recognized an opposite dependence
between the theory and experiments. In the meantime,
further experimental data have become available on the
Gaudin-Schuhmann size distribution of particles. Figure
2 shows the experiments of Kawamura et al. (1973), who
measured packing porosities varying q values a t several
different size intervals. Their experiments clearly reveal
that the average porosity decreased with increasing q
values in the lower range of q values, and the reverse was
true in the higher range. The latter dependence was also
observed in the experiments of Andreasen (1930). The
dependences similar to Figure 2 were also found in the
computer packing which was reported by Dinger et al.
(1982) as Figure 3.
Returning to our present theory, we can now estimate
the packing porosity in consideration of the effect of ma-

128

Ind. Eng. Chem. Fundam., Vol. 25, No. 1, 1986

I.
I

0,6

0.2
0.4
0.6
0.8
Distribution Constant, q

I
1.0

Figure 4. Theoretical estimates of the overall average porosity (z0

= 0.4).
0 5

0,4

0.4
0,6
0,8
Distribution Constant, 4

0,2

1,0

Figure 6. Theoretical estimates of the overall average porosity (so

= 0.6).

0.3

lu,
>

-0,2

0,7

0
L

a
a

0,l

0,6

?,2

0.4

0,E

0.8
Distribution Constant, c

1.0

Figure 5. Theoretical estimates of the overall average porosity (zo

= 0.5).

0,5

0.4

lW
>I

Table I. Values of the Size Ratios k in Figures 4-6 and 7

integral Darameter
k
integral Darameter
k
1
0.500
8
0.003 91
2
0.250
9
0.001 95
3
0.125
10
0.OOO 977
4
0.0625
11
0.000 488
12
0.000 244
5
0.0313
6
0.0156
13
0.000 122
7
0.00781

cropores. The overall average porosity of mixtures having

the Gaudin-Schuhmann size distribution of particles is a
function of (1)the porosity value for a single component,
e,, (2) the size ratio, k , and (3) the distribution constant,
q. Assuming suitable values and varying the lower integral
limit xof eq 11 stepwise between k and 1, we calculated
the porosity values from eq 11to 12a. Several theoretical
results are illustrated in Figures 4,5, and 6, where values
of e, equal to 0.4, 0.5, and 0.6, respectively, are assumed.
In each figure, the integral parameters are the assumed
k values which are summarized in Table I. In contrast
with our previous calculations (Ouchiyama and Tanaka,
1981),Figures 4-6 point out that, for the smaller range of
k , there exists a certain critical value of q, qc, which gives
the most dense packing for the Gaudin-Schuhmann size
distribution having a fixed size interval. This new theoretical trend fully explains the opposite dependence on the
q value in Figures 2 and 3. An interesting theoretical
indication in Figure 7 can be obtained by replotting the
lowest porosity against qc with constant parameters of k
and q,. Figure 7 shows that the q value giving the densest
packing a t a fixed size interval should depend on the porosity value of the single component, eo. Unfortunately,
no exact data were given for values of k and zo regarding

+-

Z0.3
0

CL
0

0.2

0.1

0
0

0,2
0,4
0.6
0,8
Distribution Constont, 4,

Figure 7. Theoretical relationship between

most dense packing.

Z,

1.0

qe, k , and z,, in the

Figure 2, and details were not reported of the packing

procedures as to Figure 3. For a quantitatively precise
discussion, further investigations are needed. However,
general agreement can be observed between the theory and
the reported data. These results suggest wide application
of the proposed theory.

Nomenclature

C(D)= coordination number, dimensionless

D,D= diameter of particle, m
D = average diameter of particles, m
D,= diameter of smallest particles, m
D1 = diameter of largest particles, m
f(D)= number-frequency size distribution of particles, m-l
f(D) = defined by eq 9c, m-l
g(x) = number-frequency size distribution of particles, dimensionless
gi = fractional number of ith component, dimensionless
i, p = integer, dimensionless
k = D,
f D1,
dimensionless

129

Ind. Eng. Chem. Fundam. 1986,25, 129-135

m = number of components of particles of different sizes,

dimensionless
N = total number of particles, dimensionless
N = defined by eq 9b, dimensionless
r i = average number of hypothetical spheres, dimensionless
q = distribution constant, dimensionless
V,, = total bulk volume of mixture, m3
V,(D) = total volume of space allocated to a specified sphere,
m3
V,(D) = volume of spherical shell, m3
vi = fractional solid volume of ith component, dimensionless
x , xy DID,, DlD,, dimensionless
f = DID,, dimensionless

average porosity of packing of uniform sized spheres,

dimensionless

Zo =

Literature Cited
Andreasen, A. H. M. Koiiok9-2. 1930, 5 0 , 217.
Dinger, D. R.; Funk, J. E., Jr.; Funk, J. E., Sr. Proc. 4th Int. Symp. Coal
Slurry Combust. 1982, 4 , 1.
Furnas, C. C. Ind. Eng. Chem. 1931, 2 3 , 1052.
Kawamura, J.; Aoki, E.; Okusawa, K . Kagaku Kogaku 1971, 3 5 , 777.
Kawamura, J.; Hayami, H.; Ariyoshi, K . ; Hosoi, E. Yogyo Kyokaishi 1973, 81,
7.
Ouchiyama, N.; Tanaka, T. Ind. Eng. Chem. Fundam. 1980, 19, 338.
Ouchiyama, N.; Tanaka, T . Ind. Eng. Chem. fundam. 1981, 2 0 , 66.
Ouchiyama, N.; Tanaka, T . Ind. Eng. Chem. Fundam. 1984, 2 3 , 490.
Tokumitsu, 2 . Zairyo 1984, 13, 752.
Westman, A. E. R.; Hugill, H . R. J . Am. Ceram. SOC. 1930, 13, 767.

Greek Letters
z = overall average porosity of mixed packing, dimensionless
t~ = surface porosity around particle, dimensionless

Received for review June 6 , 1984

Accepted April 24, 1985

Thermal Instability of Coal-Derived Naphtha

Laurlne G. Galya*+ and Donald C. Cronauerr
Gulf Research & Development Company, Pittsburg, Pennsylvania 15230

Paul C. Painter
Department of Material Sciences, Pennsylvania State University, University Park, Pennsylvania 16057

Norman C. Ll
Chemistry Department, Duquesne University, Pittsburgh, Pennsylvania 152 19

Formation of solid deposits from SRC-I1 in heat exchangers causes plugging and results in pilot plant shutdowns.
A deposit obtained from a hydrogenation reactor was examined, and the deposit precursors were concentrated
through distillation and ion-exchange chromatography. The results indicate that phenols and aikylated nitrogenous
bases are responsible for deposit formation. Phenols react through coupling reactions to give furans and water.
Alkylated nitrogenous bases react by pyrolysis to produce free radicals which cross-link to produce carbonaceous
deposits. These reactions have been followed by thermogravimetric analysis, Fourier transform infrared spectroscopy, and gas chromatography in the temperature range 250-400 O C .

Introduction
Upgrading of SRC-I1 naphthas requires nitrogen and
oxygen removal by hydrogenation using a catalytic system.
Black deposita form in both the vaporizer and inlet regions
of the catalyst bed. This often causes plugging and reactor
instability. The overall purpose of our project is to avoid
these problems by understanding the fundamentals of the
reactions leading to deposit formation.
Two basic types of reactions have been implicated in fuel
deposit formation. Oxidation reactions lead to insoluble
sludges and acids during fuel storage. These can then
deposit in reactor lines and subsequently can form coke.
Pyrolytic reactions lead to carbonaceous deposits in hot
reactor zones. Most of the previous work done in this area
has been done on petroleum- and shale-derived materials;
relatively little has been done on coal-derived fuels.
Current address: Mobil Research and Development, Billingsport Rd, Paulsboro, NJ 08066.
$Current address: Amoco Research Center, P.O. Box 400,
Naperville, IL 60566.
0196-4313/86/1025-0129$01.50/0

This paper initially discusses previous work done in the

area of fuel deposits. A brief discussion of the chemical
nature of the deposit is then given, followed by a description of the methodology used and conclusions.
Oxidative Deposits. Hydrocarbons are known to
produce acids and sludges through autoxidative reactions
in which free radicals are formed. The relative reactivity
is determined by the ease of radical formation. This means
that benzylic, allylic, and tertiary carbons are all likely
centers for oxidation reactions to occur. Nitrogen compounds have been shown to contribute to oxidative instability in both petroleum and shale oils (Offenhauer et
al., 1957; Thompson et al., 1951; Ward and Schwartz,
1962). The most deleterious forms of nitrogen compounds
appear to be alkylated nitrogen heterocycles (Frankenfeld
and Taylor, 1978, 1982). Included in these are 2,5-dimethylpyrrole, 2-methylindole, 2-methylquinoline, and
5-ethyl-2-methylpyridine(Nowack et al., 1980). Pyrroles
have been shown to react via an autoxidative mechanism.
Nonalkylated pyrrole oxidizes in n-pentane to form a deposit even though it is reported not to cause fuel instability
(Frankenfeld and Taylor, 1982). Elemental oxygen con0 1986 American Chemical Society