theory of backrjnixing by the irregular bouncing suggest = length of duct [m]

the following conclusions : = penetration of particles through cyclone Ikg/kg]

= radius of curvature of curved duct [m]
a) The penetration of coarse particles through a cyclone = dimensionless
Eq. (8c)
group denned by Eq.(8d) or
is much higher than expected. [-]
= residence time of particles [s]
b) Contrary to the performance in the fine dust region, = velocity of air at the inlet of cyclone [m/s]
the larger the particles, the lower the performance of a [m/s]
cyclone. = velocity of air in duct
= weight of fed particles [kg]
c) Larger cyclone is more effective than smaller one = weight of particles penetrated through cyclone [kg]
to collect coarse particles. = weight of particles collided against the inner
side-wall of curved duct [kg]

X
d) The mechanism of the penetration of coarse par- coefficient (Table 3) [m]
ticles through a cyclone is the back-mixing by irregular Vc modified penetration of particles through cyclone
bouncing of particles on the wall surface. defined by Eq.(9'c) [-]

e) The performance of cyclones as the coarse particle Vd - collision

by Eq.(9d)
ratio of particles in curved duct denned [-]
separators can be correlated with Eqs. (6c) ~ (9'c), which f* = viscosity of air [kg/m-s]
were derived from the similar equations correlating the p « =density of air [kg/m3]
collision of particles against the inner side-wall of curved Pp = density of particle
0i, 02 = coefficients in Eq.(6c), (Fig. 9)
[kg/m3]
[-]
ducts.
suffix
c-cyclone d=duct p-particle
Nomenclature

= coefficient in Eq.(ll) [-] Literature cited

= coefficients in Eq. (12) [-]
= coefficients in Eq.(l3) [-] 1) Iinoya, K.: Kagaku Kogaku Benran (Chemical Engineers'
= feed concentration of solid [kg-solid/kg-air] Handbook, Japan),Trans.p. JSME,
767 (1958), Maruzen Co. Ltd., Tokyo
[m] 2) Jotaki, T.: 23, 640 (1957)
= diameter
= collection efficiency of cyclone [kg/kg] 3) Mori, Y. & Suganuma, A.: Kagaku Kogaku, 29, 615 (1965)
= feed rate of particles tkg/s] 4) Mori, Y. & Suganuma, A.: ibid., 29, 1018 (1965)
= particle hold-up [kg] 5) Mori, Y. & Suganuma, A.: ibid., 30, 421 (1966)
r I
Ki, K2 = coefficients in Eq.(8d) 6) Ranz, W.E., Talandis, G.E. & Gutterman, B.: A. I. Ch. E.
Journal, 6, 124 (1960)
kpe -coefficient in Eq.(3)

PREDICTIONS OF CONSTANT PRESSURE AND CONSTANT RATE

FELTRATIONS BASED UPON AN APPROXIMATE CORRECTION
FOR SIDE WALL FRICTION IN COMPRESSION PERMEABILITY
CELL DATA"
MOMPEI SHIRATO, TSUTOMU ARAGAKI, R. MORI AND K. SAWAMOTO

Department of Chemical Engineering, Nagoya University, Nagoya

Variations of the internal flow rates of liquid and Recent developments of theoretical studies in flow
solids6'7) are important subjects in the field of the so-called through porous media and filtration operation emphasize
modern filtration theory13'10. In past investigations of the urgent need for more accurate data for porosity and
filtration theory and its practice, the concept of Ruth's local specific filtration resistance. In an actual filtration,
average resistance has been used almost exclusively. In both porosity and filtration resistance vary throughout the
Ruth's concept1'" it is assumed that the internal flow cake6). For purposes of numerical calculations, it is most
rates through a filter cake are constant. Recently it has important to obtain exactly correct values of local cake
been shown theoretically that the flow rates of liquid and properties through compression-permeability
16) data2'8'9'10'15'
solids through a filter cake are not constant and that the
new definition for the average filtration resistance is to Taylor12) and Grace2} mentioned that compression-per-
be the product of Ruth's resistance o,r and the correction meability data showed good reproducibilities if thickness
factor J-r7) which is a function of both pressure and slurry to diameter ratio Z/D of a compressed cake did not ex-
concentration. ceed 0.5 or 0.6. Actually there is appreciable side fric-
* Received on July 3, 1967 tion4'11>15) between the wall of the cell and the compressed
Presented at TokuyamaMeeting of the Society of Chem. cake, and all data obtained in compression-permeability
Engrs., Japan (1965)
cells must be corrected for the frictional effects.
86 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN
No general correlation for the side wall friction being
available, this analytical and experimental study repre-
sents an attempt to explore the approximate method of
frictional effects in compressed cakes. Based upon the
approximate correction of compression-permeability data,
it is reported in this paper that predicting calculations of
constant pressure and constant rate nitrations are com-
pared very favourably with the actual nitration experi-
ments in view of the internal flow variations in filter
cakes.

Fig. I Schematic diagram of vertical and horizontal

Approximate
sion-Permeability
Analysis
Cell
for Side Wall Friction in Compres- pressures in compression-permeability cell

In Fig. 1, the horizontal and vertical pressures in a cell

are illustrated. In the field of soil mechanics3), it is as-
sumed that the horizontal pressure ph is proportional to
the vertical pressure pv which can be expressed by
ph - kopv (l)
where ko is assumed to be a constant for each material.
Little information concerning the values of the coefficient
of friction /at the side wall is available. Utilizing the
concept of side friction and assuming that the vertical
load is uniformly distributed across the cell and that there
exists a constant cohesive force C at the wall, a force
balance over a differential height of cake can be written
as
nD2
4
-)Lpv~ (pv+dpv)1 = (kofpv+QnD-dz (2)
Rearranging and solving Eq. (2) for pv subject to pv-p
at z=0 yields

^~ i r hfp+c
£»/Lexp(4£o/ZAD) iJ {}
Fig. 2 Modified compression cell for measuring pr
where p is the applied pressure at the top. Substituting
limits pv-pr at z-Z leads to
_ 1 [ hfp+C
* 1 (,
pT~~ £o/Lexp(4£o/Z/D) CJ W Table I Experimental material used
Material Korean kaolin
where pr is the transmitted value of pv at the bottom of
the cake. The average compressive pressure ps may be True density of solids, 2.579
denned by ps [g-mass/cm3]
1 [z Stokes' dia. [//] residue [wt.%]
ps = -yr\ pv-dz (5)
Distribution of particle 20 2
Substituting Eq. (3) into Eq. (5) results the following size 10 14
equation. 5 30
2 51
*à"= IHSM1^ -«p<-"k^/D)] -if] (6) 1 64
Obviously, with variable stress within the compressed
I
cake, it is impossible to assume that the cake in the cell
has a uniform porosity and resistance. Therefore all data companyingwire strain gauges. As shown in the figure,
conventionally obtained in permeability cells must be the transmission mechanism is composed of a brass disk
modified by using Eq. (6). held between two rubber membranes of their thickness
0.4mm. The leading pipe to the strain gauge pressure
head is completely filled with water and the pressure
Side Wall Friction Tests and Results head preformance is calibrated by blank tests.
Apparatus and experimental procedure : Cell cylinders of stainless steel, gun metal, teflon and
The apparatus used in this work for studying the side acrylite, and Korean kaolin slurries (Table 1) are used
wall friction is based upon modification of the compres- for this investigation. In making the experiments, the
sion cell commonly used. Fig. 2 shows the schematic solid-liquid mixture is introduced into the cell and the
view of the assembled cell. Mechanical pressure is ap- transmitted pressure is measured at the equilibrium state
plied to the piston by weights up to 5200g-force/cm2. of compression. In addition to the side wall friction tests,
Transmitted pressure to the bottom of the compressed the normal compression permeability tests are also con-
cake can be measured by a force transducer, with ac- ducted.

VOL1 NO.1 1968 87

 Experimental results : e* = 0.769^r6-d5?B 1
Results obtained with the setup shown in Fig. 2 are ax = 1.92.x 1O10 + 2.32 x lO10/>.0'538) ;
presented in Fig. 3, where the ratio of transmitted pres- It has become apparent that conventional neglect of
sure to applied pressure are plotted against final cake side wall friction will bring significant errors in results
thickness compressed at eachconstant load. On the basis for porosity and resistance. The approximate correction
of Eq.(4) and data obtained in runs similar to the one for the friction can be done by the method which is
illustrated in Fig.3, the values of kof and C can be mentioned in Eq. (6).
evaluated for a specified cylinder. Assuminga set of the
arbitrary values of fof and C yields a value of (/>zOcaie Theory for Prediction of Constant Pressure and Constant
calculated from Eq. (4) and, therefore, the root-mean- Rate Filtration
square deviation a of (pT)caic from its experimental values
can be calculated by Eq. (7) for each assumed set of hf In the past, it has been customary to assume that the
andc. rate of flow through a filter cake is constant at any in-
stant. Such an assumption is valid only for filtration of
° ~~ V~W^ TITS ~ x 10° W
dilute slurries. The variations of internal flow rate of
where N denotes the number of experimental runs. liquid qx and internal migration rate of solids rx through-
out the cake and their relationships to the rate of poros-
cr-kof-C diagram thus obtained from experiments by a ity change over a differential section of the cake are
stainless steel cylinder is illustrated in Fig. 4. In the
written in the following forms :7'14)
diagram, the minimuma-value indicates the best set of
0.29 for fe/ and 150g-force/cm2 for C. dqx _ dsx drx __ dsx
dx ~ d6' dx ~~ W
Assuming the porosity or specific resistance determined
under a given mechanical loading in a compression per- or qx + rx = qi (10)
meability cell is the. same as the porosity ex or resistance The fundamental differential equation relating the pres-
ax at a vertical distance x measured from the medium sure gradient to flow rate can be written as7:>
in a filter cake where the cake compressive pressure is f uxxiqx~exrx) = " =.- , (ll)
the same as the solid pressure in the compression per- r dvux dwx
meability cell, £x and ax data obtained from experiments where px is the hydraulic pressure, ps the cake compres-
by a gun metal cylinder are plotted against corrected cake sive pressure, qx the apparent liquid velocity, rx the ap-
compressive pressure ps and uncorrected applied pressure
p in Figs.5 and 6. sx and ax agree fairly with the
parent solid velocity at a distance x from the bottom of
the cake, and ex denotes the local void ratio defined by
power function relations as following : ex-Sx/d-ex), pi the viscosity of filtrate, wx the mass o
ex = 0.820p dry solids deposited per unit area in the distance x from
ax=1.92 x 1010+1.12 x lO10^1
,10 I 1 1OvIAlO.0.615 [ (8) the medium. Eqs.(10) and (ll) form the basis of modern

Fig. 3 pr/p vs. Z Frg. 4 0-kof-C diagram

Fig. 5 Sx vs. ps &> p. Fig. 6 ax vs. pp &> p.

88 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN

filtration theory. Recently based upon these equations filtration at exactly constant rate. Fig. 7 shows the
under constant pressure condition, the equations for the schematic picture of the experimental apparatus. It con-
flow variations, the hydraulic pressure distribution px/p, sists mainly of a bomb filter equipped with a stuffing box,
and the correction factor JR for the conventional Ruth's a long piston, and a device for driving the piston at
resistance aR have been derived as following :7) constant velocity. Filtration pressure is applied to the
qx rx {ex-£avx){m-l) x ,.^
slurry in the bomb through the piston which is forced at
q\ q\ SaviX ~ ms) L a constant speed into the filter by the tool post of a
machine lathe and the filtration is carried out exactly at
px _ Jo L\q\/ J a constant rate. The filtration pressure at any time d
p \ [("ff)(i+ex) ~Aax(i~sx)'d(-x/L) can be measured by a force transducer, with accompany-
ing strain gauges.
(13) Discussions for prediction :
a = Jrolr Based upon thedatacorrected and uncorrected compression-
_fpf {ex-eavx)(m-l) fx\\ fwx\] permeability for the side wall friction, numerical
~ Uol1 {l-ex)saV{l-ms) *5VL/F\w)\ calculations involving Eqs. (12), (13), (14) and (15) are
P~.pl i carried out. The results of m, JR and a are illustrated
rp-Pi
in Figs. 8 and 9, respectively. Fig. 10 illustrates the
Jo CCx ^
(14)
predicted values and the experimental values of Ruth's
where L is the total thickness of the cake, eaM the
average value of ex between zero and x, and £av is the
filtration coefficient at constant pressure. The data shown
in these figures clearly indicate that conventional neglect
average porosity of the total cake. The ratio m of total for frictional effects in the compression cell leads to
mass of wet cake to total mass of dry cake is given by significant errors in estimating filtration characteristics.
m=i+-j
ps \1"~Sav)
- Q5) In Fig. ll, v and p vs. 6 obtained from constant rate
filtration experiments are illustrated. In the same figure,
Using numerical procedures based upon Eqs. (12), (13), the predicted data calculated from Eq. (18) are also plotted.
(14) and (15), the filtration characteristic values of a, m It is apparent that the experimental runs are of exactly
and Jr at constant pressure can be predicted form com- constant rate and that the predicted values are much
pression-permeability data.
more accurate (the maximumdeviation of 20%in p-0
Constant pressure : relataion) than the values (the maximumdeviation 45%)
In view of the flow variation throughout a cake, the
basic differential
dv
equationgcp may be placed2 , in the -vformf .

^ ad {jt{JRaRW + Km) K
and Ruth's coefficient K at constant pressure maybe
represented as
K =~2gcp(l - ms)
{JtpSJR ' aR
where Rm is the medium resistance, and v and vo are
the actual and fictitious volume of filtrate per unit area,
respectively. Jr, a*.and m being known, the predicted
value of K can be obtained from Eq. (17).
Constant rate filtration :
Substituting v=qim0 into Eq. (16) and solving it with
respect to time 6 gives p vs. 6 relation at the constant
rate filtration.
=.^l-ms) _ } (lg)
fiJRaRpsqi r r
where p\ is the hydraulic pressure at the interface of Fig 7 Schematic view of experimental apparatus
medium and cake. On the assumption that the filtration for constant rate filtration
characteristics will have approximately the same values
at the samepressure drop across the cake in both constant
pressure and constant rate operation, the p vs. 0 relation
at the constant rate filtration can be calculated from Eq.
(18).

Filtration Experiments and Discussions for Prediction

Experimental apparatus :
For studying the filtration characteristics at constant
pressure filtration, a bombfilter9) is used. The apparatus
used in constant rate filtration is based upon modification Fig.8 Jrand m vs. p
of a bomb filter commonly used with several devices for
VOL.1 NO.1 1968 89
 Acknowledgment

The authors are grateful to the Science Research Foundation

of Educational Ministry, Japan for support which contributed to
this article. They also acknowledge with gratitude the gifts of
teflon pipes by Nippon Asbestos Co., Ltd. and of slurry materials
by Kyoritsu Yogyo Genryo Co., Ltd.
Notation

CD = cohesive force per unit area

==diameter of cell cylinder
[g-force/cm2]
[cm]
ex - local value of voidage [-]
/ = coefficient of internal friction [-]
gc = conversion factor [dyne/g-force]
Fig 9 Jr = correction factor denned by Eq. (14)
a vs. p
kQ = a constant [-]
k = Ruth's nitration coefficient [cm2/sec]
m = ratio of wet to dry cake mass [-]
N - total number of runs [-]
p - applied pressure [g-force/cm2]
ph = horizontal compressive pressure [g-force/cm2]
Pt - transmitted at z=Z [g-force/cm2]
pv - vertical compressive pressure [g-force/cm2]
ps and px = local compressive and hydraulic pressures[g-force/cm2]
p\ = value of px at x=0 [g-force/cm2]
qx and rx - apparent velocities of liquid and solids [cm/sec]
q\ = dv/dd, nitration rate [cm/sec]
Rm - gcpi/ftqi, medium resistance [I/cm]
s = mass fraction of solids in slurry [-]
v and vn = actual and fictitious volume of filtrate per unit
area [cm]
w - dry cake mass per unit area [g-mass/cm2]
wx = mass of cake solid per unit area in distance x
from the medium [g-mass/cm2]
x = distance from the filter medium [cm]
Fia. IO
z = distance from the cake surface in compression
permeability cell [cm]
Z = thickness of the compressed cake [cm]
Greek letters
a and ax - average and local specific resistances [cm/g-mass]
aR - Ruth's average specific resistance defined by
Eq. (14) t [cm/g-mass]
£x = local porosity at a distance*x [-]
£avx - average porosityx for cake lying between medium
and distance [-]
a = defined by Eq. (7) [-]
0 = time [sec]
fi = viscosity of filtrate [g-mass/cm- sec]
p and ps = densities of liquid and solids [g-mass/cm3]
Literature cited

1) Carman, P. C.: Trans. Instn. Chem. Engrs. (London)i 16,

Fig. II pandvvs. 6 168 (1936)
2) Grace, H.P.: Chem. Eng. Prog., 49, 303, 367 (1953)
3) Janssen, H.A.: V. D. I.-Z., 39, 1045'(1895)
which do not account for the side wall friction. 4)
5)
Lu,
Ruth,
W.M.:
B.F.:
Thesis,
Ind.
Univ. Chem.,
Eng.
of Houston (1964)
49, 564 (1946)
6) Shirato, M. and S. Okamura: Kagaku Kogaku, 19, 104(1955)
Conclusion 7) 31,Shirato,
359 (1967)
M., M. Sambuichi, H. Kato and T. Aragaki: ibid.,
8) Shirato,M. and S. Okamura: ibid., 20, 678(1956): 23, 8
In the compression-permeability cell, there is an ap- (1959)
preciable side friction between the wall and the solids.
9) Shirato, M. and S. Okamura: ibid.,.2S: ll (1959)
This friction leads to significant errors in filtration chara- 10) Shirato, M. and M. Sambuichi: ibid., 27, 470 (1963)
cteristic values estimated in accord with the conventional ll) Shirato, M., M. Sambuichi, H. Hirate, R. Matsuyama and
K. Sawamoto : Paper presented at Tokuyama Meeting, Society
method.
Based upon the approximate analytical equation for the
of
12) Chem.
Taylor, Engrs.,
D.W.: Japan (1965) I. T. (1942)
Thesis,M.
13) Tiller, F.M.: Chem. Eng. Prog., 49, 467 (1953)
side wall friction, compression-permeability data are cor- 14) Tiller, F.M. and M. Shirato: A. I. Ch. E.Journal, 10, 61
rected and nitration characteristics are evaluated in view (1964) : Kagaku Kogaku, 26, 925 (1962)
15) Tiller, F.M.: A. I. Ch. E. Journal, 4, 170 (1958) : C. E. P.,
of the so-called modern filtration theory. 51, 282 (1955)
Reasonable coincidence between theories and experi-
16) (1962)
Tiller, F.M. and H.R. Cooper: A. I. Ch. E. Journal, 8, 445
ments are assured for both constant pressure and constant
rate nitration. 17) Welch, R.C.: Thesis, Univ. of Texas (1964)

90 JOURNAL OF CHEMICAL ENGINEERING OF JAPAN