I n d . Eng. Chem. Res.

1987, 26, 857-859

857

D = dimensionless diffusion coefficient

f = Fourier coefficient (function of time)
F = arbitrary function
g = complementary function
h = particular solution
k = odd integer
L = thickness
n = odd integer
t = dimensionless time
x = dimensionless distance
z = distance from surface
Greek Symbols
a = matrix element
/3 = matrix element
y = diffusion coefficient
X = dimensionless constant

'

0.01
0

0.2

0.4

0.6

t
Figure 1. Fourier coefficients,fi(t) and f3(t),and average concentration, CA, for D = 1 + 10C or X = 10. Solid curves: analytical
solutions from (lo), (20), and (21). Crosses, circles, and solid triangles: numerical solutions of Fick's diffusion equation. Open triangles
with spikes: numerical solutions for CAgiven by Crank (1975).

function, ( 2 / 1 r ) ' /sin

~ nx, was dictated by the slab geometry
and the boundary conditions. For geometries such as
cylinders and spheres, different types of eigenfunctions are
necessary (Crank, 1975).

Acknowledgment
The partial financial support by National Aeronautical
and Space Administration Grant NAG-5-156 is gratefully
acknowledged.

Nomenclature
c = integration constant
C = dimensionless concentration
CA = average dimensionless concentration

T = time
4 = concentration of sorbent
&, = sorbent concentration in the environment
= initial sorbent concentration

Literature Cited
Carslaw, H. S.; Jaeger, J. C. Conduction of Heat in Solids; Clarendon: Oxford, 1959; Chapters 3-5.
Crank, J. The Mathematics of Diffusion, 2nd ed.; Clarendon: Oxford, 1975; Chapters 4-9.
Lowdin, P. 0. Adu. Chem. Phys. 1959,2, 260.
Tsang, T. J. Appl. Phys. 1961, 32, 1518.

Tung Tsang*
Department of Physics and Astronomy
Howard University
Washington, D.C. 20059
Carol A. Hammarstrom
Code 6120, Naval Research Laboratory
Washington, D.C. 20375
Received for review November 21, 1985
Accepted December 12, 1986

A Consistent Rule for Selecting Roots in Cubic Equations of State

In several papers by Robinson (1979), Peng and Robinson (1976), and Soave (1972), it has been taken
for granted when selecting roots of cubic equations of state (EOS) that the largest root is for the
compressibility factor of the vapor, while the smallest positive root corresponds to that of the liquid.
Some recent studies by Lawal (1985) reveal that this assumption is only valid for van der Waals
(VDW) EOS. Therefore, the purpose of this communication is to illustrate the reason why this
assumption does not always hold for all cubic EOS of van der Waals type. Both the analytical and
graphical techniques are presented to illustrate a consistent rule to be applied in choosing the smallest
positive root of EOS. Examples of applying the rule in Peng-Robinson (PR) EOS for the high-density,
single-phase fluid are illustrated. T h e error in predicting the fluid density with the selection of an
incorrect compressibility factor resulted in an average absolute percent error of 17 142%.
Let us consider a cubic EOS of the form
a(T)
p = - RT
u - b v 2 abu pb2

Equation 1 can be rewritten as

2 3 - [I + (1 - 413122 + [ A - a~ + (p - 4~212
[AB + p ( ~ +
2 ~ 3 1 1=

(2)

where

B = -bP
RT

(4)

0888-5885/87/2626-0857$01.50/0

By assigning certain integral values to a and 0, eq 1 and

2 take the same form as the familiar cubic EOS. Examples
of these are VDW (a = 0, p = 0, a = a ( T ) ) , RKS (a = 1,
p = 0), PR ( a = 2 , p = -l),and Harmens (a = 3, /3 = -2).
Equation 2 yields one or three roots depending upon the
temperature of the pure component. In the two-phase
region as dictated by the pressure being less than the vapor
pressure when T < T,,the largest root is for the compressibility factor of the vapor. The smallest positive root,
which must be greater than or equal to B of eq 4, corresponds to that of the liquid. This is the criterion for testing
the roots of eq 2, but when this criterion is ignored, the
use of the absolute value becomes as necessary in the fugacity equation as it is in the GPA program (Bergman,
1976). Therefore, 'complex roots and roots with a value
0 1987 American Chemical Society

858 Ind. Eng. Chem. Res., Vol. 26, No. 4, 1987

Table I. Comparison of Experimental with Computed Densities for Methane"
T,O F
P, psia
De, lb/ft3
roots of eq 7
9 000
10 000
6 000
7 000
8 000
9 000
10 000

400
400
460
460
460
460
460

12.19
13.00
8.55
9.62
10.57
11.43
12.23

1.216, -0.6374,
1.260, -0.7390,
1.102, -0.3657,
1.138, -0.4526,
1.175, -0.5426,
1.215, -0.6350,
1.255, -0.7293,

0.002799
0.013 66
0.002 84
0.01083
0.01966
0.029 18
0.03931

eq 4

D,,lb/ft3

D,, lb/ft3

0.4184
0.4648
0.2607
0.3042
0.3476
0.3911
0.4345

12.87
13.80
8.85
10.01
11.06
12.04
12.95

5 590.66
1273.14
3 432.04
1050.26
661.54
501.30
413.49

4.43

17 142.57

av abs % dev
eMethane parameters: T,= -116.68

P, = 667.8 psia,

OF,

w = 0.008.

less than B, including negative roots obtained from eq 2,

have no significant meaning and should be discarded.
Application of this criterion (eq 4) in EOS will be illustrated for VDW and PR EOS.

Equation of States Root Selection

For the VDW EOS, eq 2 becomes
2 3 - (1 + B ) Z ~
+ AZ - AB = o

(5)
We can explore the roots of eq 5 by applying Descartes'
rule of signs which simply states the following:
The number of positive roots of function f ( x ) = 0 is
either equal to the number of variations of sign of f ( x ) or
is less than that number by an even integer. The number
of negative roots of f ( x ) = 0 is either equal to the number
of variations of sign of f ( - x ) or is less than that number
by an even integer.
Therefore, if the terms of a polynomial, f(Z),with real
coefficients are arranged in order of descending powers of
2, a variation of sign occurs when two consecutive terms
differ in their signs. When we apply Descartes' rule to eq
5, we have three sign variations. Therefore, three real or
one real and two complex roots are possible from the VDW
EOS. But since

f(-z) = - 2 3 - (1 + B).P - AZ - AB

(6)
has no sign variation, the number of negative roots is zero.
Hence, VDW EOS will always result in the largest compressibility root for vapor and the smallest root for liquid
compressibility. This also implies that the graph of the
VDW EOS exists only in one region.
In the case of PR EOS, eq 2 becomes
Z3 - (1 - B ) P + ( A - 3B2 - 2B)Z - (AB - B2 - B3) = 0
(7)
Application of Descartes' rule to eq 7 leads to the same
conclusions as VDW EOS. But, since the signs of the
coefficients of P,Z,and ZO depend on the magnitudes of
A and B, which depend on temperature and pressure, the
result of applying Descartes' rule is inconclusive. In order
to know the kind of roots from PR EOS and the regions
where they occur, we shall apply Silberberg's principle
(1963). The Silberberg principle can be illustrated by
putting eq 1 in the reduced form of PR EOS. Thus, the
reduced PR EOS becomes
PR

3.253

TR

V R- ( b / VJ

4.839

NTR,a)

VR' -I-2 ( b / V c ) V-~ ( b / V J 2

(8)
Singularities exist for eq 8 at V R= b/ V, and V , = - (1 f
Z1I2)b/V,. The singularity at VR = b / V , leads to the

vertical asymptote, while the horizontal inflection occurs

at PR = 1. By analyzing (Silberberg, 1963) the limit as V R
approaches each of the singularities at -1 f Z1I2 positively
and negatively, we can approximately sketch the graph of

\\ I! IIIIIIini
Figure 1. Illustrating graph for Peng-Robinson equation.

PR EOS as shown in Figure 1for ( P ( T R , w ) = 1 at PR = 1

and TR = 1. However, at any other values of Q(TR,w ) , the
graph might look different, but the general pattern will
be as shown in Figure 1. Figure 1 shows more than one
region in which multiple roots can exist; they can be
positive, negative, or complex. Therefore, it is possible to
have one, two, or three positive, one or two negative, and
two complex roots from PR EOS. The possibility of some
positive roots being less than b (the covolume of the
molecule) is implicitly clear from Figure 1. This graphical
illustration of PR EOS is consistent with the observation
described for argon using the same EOS by Gomez-Nieto
(1979).

Application of Equation 4
Table I shows the density of methane at two temperatures and at several pressures in the supercritical region,
as reported by Sage and Berry (1971). At the temperatures
and pressures shown in Table I, PR EOS is expected to
yield one real and two complex roots in the single-phase
region, as does VDW EOS, but in these cases, three unequal roots are obtained. Since these roots can be positive
or negative, it is necessary to choose only the positive
compressibility root greater than B for calculating the
density. The results of selecting the minimum positive
compressibility root from eq 7 to calculate the density are
shown as D , in Table I, and the result based on criterion
B (eq 4) is also shown as D,. The average absolute percent
error in predicting the dense fluid density is 4.4%, while
that based on the minimum positive root is 17 142%.
Therefore, in the event of multiplicity of real roots, the
smallest of the positive roots larger t h a n or equal t o B
must be chosen for the compressibility of the liquid. But,
if in every EOS the smallest of the positive roots larger
than B were always the largest root, the comments by
Gomez-Nieto (1979) and the reply of Robinson (1979)

Ind. Eng. Chem. Res, 1987,26, 859-860

might be adequate. However, this is not always the case.

Conclusions
I t has been shown briefly in this communication that
the smallest compressibility root must be greater than B
to have physical significance in a cubic EOS. The same
rule applies for a mixture of components. However, for
this case, the b term in eq 4 has to be replaced with the
mixture parameter, b,. It was also shown that only the
VDW EOS can yield a single value of the compressibility
factor in the supercritical region.
Nomenclature
A = constant defined by eq 3
a, b = equation of state parameters (EOS)
B = constant defined by eq 4
De = experimental density, lb ft3
D, = calculated density, lb/ft
D, = calculated density based on smallest positive root, lb/ft3
P = pressure, psia
PR = reduced pressure, PIP,
R = gas constant
T = temperature O F
T R = reduced temperature, T I T ,
V R= reduced volume, V I V ,
u = molar volume, ft3/(lb.mol)
2 = compressibility factor, PV/RT

859

Greek Symbols
cy, @ = characteristic coefficients of EOS
w = acentric factor
@ = temperature function in PR EOS

Literature Cited
Bergman, D. F. Ph.D. Dissertation, University of Michigan, Ann
Arbor, 1976.
Gomez-Nieto, M. Ind. Eng. Chem. Fundam. 1979, 18, 197.
Lawal, A. S. L. Ph.D. Dissertation, University of Texas, Austin, 1985.
Peng, D.-Y.; Robinson, D. B. Ind. Eng. Chem. Fundam. 1976,15,59.
Robinson, D. B. Ind. Eng. Chem. Fundam. 1979,18, 197.
Sage, B. H.;Berry, V. M. Phase Equilibrium in Hydrocarbon Syst e m . Behavior of the Methane-Propane-n-Decane System; API
Monograph Research Project 37; American Petroleum Institute:
Washington, DC, 1971.
Silberberg, I. H.PEN 384, 1963; Petroleum Engineering Department, The University of Texas at Austin, 1963.
Soave, G. Chem. Eng. Sci. 1972,27,1197.
*Present address: SOHIO Petroleum Company, Anchorage, AK
99508-4254.

Akanni S.Lawal*
SOHIO Petroleum Company
Sun Francisco, California 94105
Received for review September 3, 1985
Accepted November 19, 1986

Comments on Recent Publications on Minimum Fluidization Velocity at

High Temperatures
Recent controversies reported in the literature by Wen and Yu and Botterill et al. concerning the
variation of bed voidage at minimum fluidization with temperature are discussed. These are explained
and reconciled on the basis of the powder classification scheme of Saxena and Ganzha and the
corresponding fluid flow field around the particles, giving rise to the interparticle forces as elaborated
by Mathur and Saxena. Thus, the apparent contradictions in the experimental findings are consistent
with each other as well as with the hydrodynamical picture of the gas-solid system.
Pattipati and Wen (1981) reported measurements on
minimum fluidization velocity of sand beds in the size
range 240-3376 pm at ambient pressure and in the temperature range 298-1123 K and concluded that the Wen
and Yu (1965) correlation could successfully represent the
temperature dependence of minimum fluidization velocity,
UFP This implied that the bed voidage at minimum
fluidization, emf, is independent of temperature and the
values determined at lower temperatures could be used for
making predictions at higher temperatures. This was
questioned by Botterill et al. (1982a) who emphasized that,
in conformity with their earlier observations (Botterill et
al., 1982b), emf varies with temperature and that the conclusion of Pattipati and Wen (1981) became increasingly
inaccurate as the temperature was increased. They found
that predictions based on temperature-invariant E& values
and corresponding experimental values agreed in most of
the cases within a maximum scatter of f34%. Pattipati
and Wen (1982) refuted this claim of Botterill et al. (1982a)
by reporting two sets of measurements of emf for a 250-ym
sand bed in the temperature range 293-1123 K. They
concluded that emf does not change with temperature and
the results obtained by Botterill et al. (198213) were due
to an inadequate heating arrangement leading to bed
temperature nonuniformity. The purpose of this communication is to provide explanations which resolve all
these apparent controversies raised by these works.
First, we would like to point out that two recent independent researchers (Lucas et al., 1986; Saxena et al., 1986)

have found that emf varies with temperature and have

reported the same for sand beds of different sizes. The
qualitative nature of this variation is different for beds of
particles of different sizes, and further, in each case it also
depends upon the temperature range. Second, as emphasized by Mathur and Saxena (1986), Emf for a given
gas-solid system depends upon the nature of operating
interparticle forces which in turn depend upon the fluid
flow field around the particles. Consequently, the characterizing parameter for the system is Reynolds number,
Re, and not such variables as temperature, pressure,
particle size, etc. It is shown in the following that all the
measurements mentioned above are adequately explained
on the basis of these two facts.
The emf values of Pattipati and Wen (1982) correspond
to Red values lying in the range 0.04-0.87. The fluid flow
around particles for this range of Re is known to be laminar
(Clift et al., 1978). For laminar flow, the interparticle
forces are negligible, and therefore, emf has a constant value.
In developing a classification scheme based on their hydrodynamic and themal behaviors of powders, Saxena and
Ganzha (1984) placed such powders in group I and, based
on existing literature, proposed an upper limit for the
characteristic Re as 1or even as high as 10. Particle shape
and its morphology will play a very significant role in
establishing the flow field, and until such controlled experiments become available, it is not possible to uniquely
specify the value of Re where initial departure from laminar flow field will be observed. The constant values of

0888-5885187 /2626-0859$01.5010 0 1987 American Chemical Society