Chid

Engirrcering Science. Vol. 45. No. 7, pp. 1933-1935,

1990.

Printed in Great Britain.

$3.00 + 0.00 caJ9-2509/90 0 1990 Pergamon Press plc

The influence of feeding irregularities on the through-circulation drying of loose materials

(Received 25 August 1989; accepted 20 December 1989)

Schliinder (1977) has pointed out the significance of nonuniformities in particulate beds of materials in which mass transfer is taking place between the particles and a throughflowing medium. In the through-circulation drying of loose materials over perforated bands, such non-uniformities can arise because of inadequacies in the feeding arrangements to produce material which is uniformly spread over the whole band width. Such irregularities are common in the handling of loose., fibrous materials when individual fibres can tangle together, thus producing high and low spots over the layer of material. Operators, in the wool-drying industry for instance, tend to compensate by using extra drying capacity to smooth out consequential moisture-content variations in the stuff. This policy is costly both in terms of hardware and operating costs. Keey and Wu (1989) have recently examined this problem in relation to the drying of loose scoured wool over a train of suction drums. They introduced a transformation of the thickness profile across the dryer as a means of estimating the effect of irregularities on the drying performance of the equipment. This communication sets out a. generalization of the concept. Consider the drying of loose solids over a perforated band of width w. The depth y of the material on the band will vary from zero (or a small residual value yO) to a maximum s which is set by the feeder. A rcprcsentatlve profile is shown in Fig. 1. This profile can be transformed into a cumulative probability distribution which is illustrated in Fig. 2. To a first approximation the distribution can be approximated by an exponential function: 6 =
(Y y,)l(s Y,) = exp -

set by the feeder. The coefficient k has been called the unevenness factor (Keey and Wu, 1989) and is a measure of the steepness of the cumulative probability curve. As k approaches zero, so the distribution becomes more uniform. The mean depth of solids on the band is given by <6> = 1exp ( - k/w)dw I0 (3)

from which the hold-up of solids on the band of length L and breadth B can be found as H = PB<B>BL. (4)

The through-circulating air velocity through the bed varies from place to place to produce a uniform overalf pressure drop AP,. The gas rate itself may be described by a function of the kind

(5)
Fluid-dynamic considerations yield a relationship between AP, and the local velocity, which is a function of the path length 6 and the unevenness factor k. The local velocity Y& increases over the limiting value u1 for the maximum possible depth as the unevenness becomes less and the bed shallower. The changes are shown qualitatively in Fig. 3; the actual numerical values depend upon the size of the band s perforations relative to the particle dimensions and the relative thickness of the band compared to the maximum solids depth. For example., in the drying of wool fibres for which k = 0.2 is a representative value for the unevenness factor, the velocity ratio is about 1.2 at 6 = 0.5 but is 1.85 at 6 = 0.1. Thus, for over one-tenth of the band, the air velocity is not less than about twice the magnitude it would have been if the wool had been uniformly piled to the maximum possible depth. The value of the velocity ratio ud/ul at zero 6 reflects the air rate through the solids-free band to give the same overall pressure drop. The velocity ratio for S = 1 is slightly greater than 1 to reflect the small difference between the actual and theoretical maximum solids depth.

k/w

(1)

over the region 1 > P(w) > PO, where o is essentially the relative effective width evaluated in terms of the probability P(w) that the depth is y or less: w = [P(w) - POl lCl - P&J_ (2)

The probability P, is the fraction of the bed which is covered by some residual depth y,. When o = 1, the relative depth 6 takes a value slightly less than 1, depending on the magnitude of k, to reflect the observation that the solids rarely are piled to the exact limit

! s-

Y
Y

/
k
1b-

P(w)
Fig. 1933 2. Cumulative probability depth. distribution of material

Fig. 1. Variation

of material depth over the band width.

1934

Shorter Communications rate NW per unit exposed surface A:

e=
d

NO.4
-X,)

F(X.,

I5

where LI is exposed surface per unit volume. Equations (12a)-(12c) when substituted in eq. (10) lead to the non-dimensional equation dO -=-. ae I
I

r=II PA1
an

&a
-#2)(X,, -A-,)

cu2c)
(13)

ae

Consideration of the drying rate within volume in the layer yields the equation CP,U -

the incremented

Fig. 3. Variation of local velocity ratio with relative depth of material on the band.

d~,dw+l~

=fS#&dwd~d4CYw

- YeI (14)

wherefis some function of Q and accounts for the influence of the fibres themselves in restricting moisture movement. Equation (14) reduces to the non-dimensional expression If a linear relationship between the mass-transfer coefficient p and the gas rate G is assumed for the bed (Kunii and Suzuki, 1967), then B = B,G(WG, (6)

A!!+

(15)

where & is some base value at a standard gas rate G,. The number of transfer units (NTU) t(o) or mass-transfer extensiveness in the streamwise direction is given by t;(a) = LWMMG = &d,a~/G, (7)

where 0, is the humidity-potential coefficient, u is the interfacial area per unit volume, and y is the solids depth in the streamwise direction (Keey. 1978). It follows that
T(o) =

!!$e css + (1 0

4YJ

(8)

which reduces to

when y, is zero, which is often the case. The consideration of the exchange of moisture from an elemental volume of width dw, depth dy and extent dz leads to the expression d[GY,](dwdz) = - [p,(l - s)u,(dwdy)]dX (10)

where X is the local moisture content (dry basis), Y, is the bulk-air humidity, E is the porosity or openness of the material, ps is the true solids density, and u, is the solids velocity in the longitudinal (z) direction. Equation (10) simplifies to p,(l - e)g= G2

on substitution with values from eqs (12a)-(12c). In the first drying period f= 1 and 0 3 1, but in the so-called falling rate period, 0 < f 6 1 and 0 -z QI 6 1. Solutions to eqs (13) and (15) are given elsewhere (Keey, 1978) for the case when the falling rate is linear with moisture content (f= (0). These equations may be applied to the case of uneven feeding if it is assumed that the solids do not redistribute themselves once they enter the dryer. The unit may thus be regarded as an infinite set of parallel chambers, each of which is fed uniformly to a particular depth 5. (In practice, the use of finite differences would be adequate.) Although the thickness profile is likely to change with time, one may assume to a first approximation that the cumulative probability distribution [and thus eq. (l)] does not. By way of example, the moisture-content profiles as a function of relative contact time B and thickness t(w) are given in Fig. 4 for the case when the feed is relatively wet. It is seen that, at 0 = 2, there is over a threefold variation in moisture content should C vary from 0 to 1. Even when t: is only 0.5 (which corresponds to normal working conditions for drying of wool fleeces), there is still a twofold variation. Feed-depth irregularities thus lead to substantial crosswise variations in moisture content besides the streamwise differences that follow with through-circulation. Such differences are never entirely ironed out in the drying process. In the limit, the ratio of the overall difference in Cpto the mean value

(11)
to introduce dimen1.6

by writing dt = dz/u,. It is convenient sionless parameters: -the humidity potential l-I= YWY, Yo Y,,

(12a)
a

12 -.-.-._ 0.6

where Y, is the wet-bulb humidity (taken to be constant everywhere in adiabatic drying), and YcO is the inlet-air humidity at the band; -the characteristic moisture content: @=A x-x X,. - x, where X,, is the critical moisture equilibrium value; content, and X, (12W is the

0.4

4
A,

e
Fig. 4. Variation of local moisture content CD as a function of contact time 0 and streamwise extensiveness c.

-the relative drying time, in terms of the time needed to remove band moisture from the solids at the initial drying

Shorter Communications is 2 no matter how small the residual moisture content is. The calculations underscore the importance of ensuring that the feed material is spread as uniformly as possible in the first place. As Keey and Wu (1989) point out the method can be extended to consider variations in bulk density or freeness. Essentially, one can consider the equivalent case with a reference porosity but a variable thickness which has the same extensiveness or NTU. Y YL. YO K.30 &V solids depth, m least solids depth, m air humidity, kg kg-i inlet-air humidity, kg kg- wet-bulb humidity, kg kg-

1935

Acknowledgement--The

original work, from which this communication stems, was supported by the Wool Research Organisation of New Zealand. R. B. KEEY

Department of Chemical and Process Engineering

University of Canterbury

Christchurch

1, New Zealand
NOTATION

f G G, H k L P P, s
d % : X,, X,

interfacial area per unit volume, m-l band breadth, m relative drying rate [eq. (15)] gas rate, kgme2 s- standard gas rate, kgm-zs- hold-up, kg unevenness factor [eq. (l)] band length, m cumulative probability probability of least solids depth maximum solids depth set by feeder, m local velocity at solids depth, m s- 1 local velocity at maximum solids depth d, ms- band width, m moisture content, kg kg- critical moisture content, kg kg- t equilibrium moisture content, kg kg- t

Greek letters mass-transfer coefBcient, m s- 1 mass-transfer coefficient at standard gas rate, kgm-fs- relative solids depth [eq. (l)] total pressure drop, Pa porosity number of transfer units in streamwise direction relative contact time relative humidity potential bulk density, kgm- gas density, kgm- solids density, kgm- humidity-potential coefficient relative free moisture content [(X - X,)/(X,, - X,)] relative effect width [eq. (2)]

REFERENCES

Keey, R. B., 1978, Introduction to Industrial Drying Operations. Pergamon Press, Oxford. Keey, R. B. and Wu, Y., 1989, The influence of unevenness of the wool mat on the drying of loose wool. Chent. Engng Process. 26, 127-137. Kunii, D. and Suzuki, M. 1967, Particle-to-fluid heat and mass transfer in packed beds of fine particles. Int. J. Heat

Mass Transfer SO, 84$-852.

Schliinder, E. II., 1977, On the mechanism of mass transfer in heterogeneous systems-in particular in fixed beds, fluidiid beds and on bubble trays. Chem. Engng Sci. 32, 845-851.

Chemicrrl Engineering Science, Vol. Printed in Great Britain.

45,No. 7, pp. 1935-1938.1990.

om!-2509/w 53.00 + o.al 0 1990 Pergamon Press plc

Correlation
(Received

effects in counterdiffusion

in zeolites
25 January 1990)

2 May 1988; accepted for publication

Diffusion in zeolites has been a subject of extensive investigation over the last three decades. The diffusion behaviour in zeolites has been commonly investigated by gravimetry and volumentry. These methods permit study of unidirectional transport of sorbate molecules in the zeolite cavities. Studies devoted to counterdiffusion in zeolites, however, have been rather limited [Pitale et al. (1988), Choudhary et al. (1989) and references therein] in spite of its relevance to practical application of zeolites as catalysts and adsorbents. While most of the experimental data obtained by gravimetry and volumetry can be anaiysed using Fick s law of diffusion, interestingly a large number of reported counterdiffusion data cannot be correlated by a constant value of diffusivity. Sattertield and co-workers (Satterfield et al., 1971; Satterfield and Katzer, 1971) studied counterdiffusion of several aro-

matic hydrocarbons in mordenite and several cationic forms of zeolite Y. The diffusivities were observed to decrease with the extent of counterdiffusion (time). The effect was more significant in the case of mordenite. Satterfield and Chiu (1974) studied counterdiffusion of aromatic and naphthenic hydrocarbons and observed that diffusivities decreased after 50% of equilibrium was reached. Moore (1971) and Moore and Katzer (1972) also observed a similar decrease for counterdiffusion of hydrocarbons in zeolite Y. Pitale (1987) and Pitale et al. (1988) recently studied tracer diffusion of carbon dioxide and methanol and counterdiffusion of methanolethanol and methanol-hexane in NaY and ZSM-5 and observed that diffusivities decrease with the extent of counterdiffusion. In several experiments the counterdiffusion came to a halt before the expected equilibrium was estab-