The continuous thickening of a uranium plant slurry was studied on a 7-ft diameter pilot plant. Flow patterns and
density profiles within the thickener were observed. It was found that there were two distinct stable modes of
operation which were named "settler" and "filter" modes. The "settler" mode coincided with the underloaded
operation while the "filter" mode corresponded to a fully loaded thickener. The flow patterns, internal circula-
tions, residence-time density functions, overflow turbidities, and visual appearances of these two states were
found to be very different. Maximum throughputs were 70-80% compared with those predicted by Coe and
Clevenger and flux theory methods. This is attributed mainly to a more rapid settling out of the settling zone of
the coarser particles, leaving the remaining pulp with a lower settling flux. Concentration zones in the thickener
were not found to agree with the flux theory predictions.
;IO A‘
2 1 I
u
10 30 50 70
4
0 02
Soilds concentrotion
0 04
gmslcc
I
Time minutes Figure 4. Reduced settling velocities for Coe and Clevenger and
Figure 2. Coe and Clevenger batch settling curves. Initial concen- Talmage and Fitch methods: A, Talmage and Fitch, initial concen-
trations (g/cm3): A, 0.0087; B, 0.0153; C, 0.0209; D, 0.0257; E, tration = 0.0087 g/cm3; B, Talmage and Fitch, initial concentra-
0.0342; F, 0.0340. tion = 0.0155 g/cm3; C, Talmage and Fitch, initial concentration =
0.0238 g/cm3; D, Talmage and Fitch, initial concentration = 0.0286
g/cm3; E, Coe and Clevenger
Y
H r
settler modes
0 01 0 02 0 03
Feed concentration gmrlcc
Figure 5 . The Internal circulation rate and settling zone concen-
tration as a function of feed concentration for an inlet flow rate of
12 l./min.
density
I 0.01 0.02 0.03
I I
0.04 profile
Concentration gmslcc
Figure 6. Velocity and density profile for the “settler” mode of‘
Figure 3. Test of the modified Richardson and Zaki equation for operation. Arrow lengths represent flow velocities.
feed and settling zone material: 0 , feed material 24.5OC; A, feed
material 22.8OC; .,settling zone material 22.8OC.
T h e transition from “settler” to “filter” mode was dra-
Table I. Constants for Modified Richardson and Zaki matic, occurring at the throughput where the thickener was
Equation a t maximum capacity. Any further increase in thickener
throughput resulted in a sliming thickener.
Temp, ’C u o , cmimin h , cm3/g T h e main difference between the two modes, besides
24.5 4.77 18.1 their throughputs, was in the flow patterns that existed in
22.6 3.39 17.7 the thickener. In “settler” mode large circulating flow pat-
21.0 2.72 17.9 terns existed in the settling and clear water zones, while in
20.0 2.16 17.0 “filter” mode circulating flow in these zones was almost
negligible as can be seen from Figure 5 . These differences
all probability this is due t o some segregation during set- in flow patterns had a significant effect on a number of sec-
tling (as found in the continuous settling). For this reason a ondary parameters, particularly in the settling zone where
Talmage and Fitch flux curve was not drawn. the depth and concentration of the settling zone increased
Continuous Thickening. Flow patterns and zones with- drastically as the circulation decreased. In fact the “filter”
in the thickener were very easily seen using the perspex mode derived its name from the very deep settling zone ex-
wedge and were very well defined. Under all conditions of isting under almost stagnant conditions and appearing to
t h e thickener in normal operation three zones could be eas- act as a filter bed for the feed material. This is also shown
ily distinguished, namely clear water zone, settling zone, in Figure 5.
and compaction zone. T h e steady-state runs showed the “Settler” Mode of Operation. T h e general condition of
thickener to operate under two distinct modes of operation, operation of “settler” mode is illustrated in Figure 6, show-
characterized by different types of flow patterns and densi- ing the zones, circulating flow patterns, and density profile.
ty profiles. Both modes, called “settler” and “filter” were T h e feed to the thickener entered the tank vertically
shown to be stable operating conditions, corresponding to downward through the inlet distributor and was diluted by
the thickener being underloaded and a t full capacity, re- the circulating water from the clear water zone. This feed
spectively. then turned horizontal and entered the thickener itself a t
Units
Figure 7. Velocity and density profiles for the “filter” mode of op-
eration. Arrow lengths represent flow velocities. I /
4
10 20 30 40 50 60 70
its compatible density level, forming the settling zone. This Time mlnutel
cascading underwater waterfall effect has been seen by a Figure 8. Liquid residence-time density function curves; settler
number of other authors. mode feed flow rate = 11.8 I./min; filter mode feed flow rate = 11.9
l./min.
The settling zone was relatively narrow, and the feed
into the settling zone covered the complete depth of the
settling zone. All three zones were very well defined with to the horizontal flow and the first tracer appeared in the
definite interfaces between them. overflow after 4 min. This was the result of approximately
Under “settler” mode the circulation of clear water from two equal time delays, namely in the settling zone from the
the clear water zone was very high, being as high as 2.6 center to the perimeter and the flow a t the perimeter from
times the volumetric feed flow. This was calculated from the settling zone t o the overflow. T h e peak in the residence
the dilution of the feed as it entered the thickener by si- time density function curve appears after about 8 min. By
phoning samples into the wedge. As this diluted feed en- this time the fastest circulating flow in the thickener had
tered the settling zone it spread very quickly toward the completed two cycles, giving the appearance of tracer com-
perimeter, with a decrease in horizontal velocity as the pe- pletely mixed throughout the settling and clear water
rimeter was approached compatible with the increased area zones. The liquid residence time distribution is thus not
available for flow. At the perimeter the water flowed up- dissimilar t o that of a perfectly mixed vessel with an initial
ward into the clear water zone leaving most of the solids delay.
behind in the settling zone. T h e circulation back t o the “Filter” Mode of Operation. T h e operation of the
center originated from the perimeter, with a n increase in thickener under “filter” mode is illustrated in Figure 7. As
horizontal velocity as the center was approached. At the in “settler” mode the feed entered the thickener vertically
perimeter part of the upflow from the settling zone to clear from the inlet distributor, and then entered the settling
water zone did not recirculate to the center. This consti- zone horizontally at its compatible density level. Under fil-
tuted the overflow. ter mode the settling zone was very deep and the feed en-
Vertical velocity profiles existed in both the settling and tered over almost the entire depth. T h e settling zone ex-
clear water zones, as illustrated in Figure 6. In addition a tended from the compaction zone almost to the surface,
horizontal velocity profile existed in both these zones and leaving a very narrow clear water zone.
could be seen by viewing the thickener in plan. A liquid T h e rate of circulation of clear water was very low, and
tracer added to the feed showed random maximum veloci- thus very little dilution of the incoming feed occurred. T h e
ties as the tracer front moved out non-uniformly from the rate of circulating water t o fresh feed was approximately
center of the thickener. This uneven tracer front was visi- 0.2, thus resulting in a settling zone concentration of much
ble in both settling and clear water zones. higher concentration than under settler mode. Similarly,
T h e flow through the settling and clear water zones ap- the flow from the center out radially was only to replace
peared to be laminar. No mixing or transfer of material be- solids that settled into the compaction zone and water
tween or within the two streams occurred except a t the flowing to the overflow. Thus the horizontal flow in the set-
feed distributor and at the perimeter. tling zone was slow in comparison of t h a t of settler mode.
In its passage from the center t o the perimeter the feed T h e flow out radially was laminar with a vertical velocity
appeared to undergo batch settling with the interface be- profile as shown in Figure 7. Close to the interface with the
tween the settling zone and clear water zone becoming bet- clear water zone flow was small and random. An additional
ter defined as the concentration of the settling zone in- horizontal velocity profile occurred similar to the one for
creased. The upflow a t the perimeter carried particles from “settler” mode.
the cloudy interface to the overflow, giving the “settler” In “filter” mode the settling zone was both much greater
mode a relatively turbid overflow. T h e turbidity of the ov- in depth and much higher in concentration than in “set-
erflow decreased as the concentration of the settling zone tler” mode. The higher concentration was due in part to
increased. These are phenomena that had parallels in the the lower circulation rate of clear water, and in part to a
batch settling experiments (Turner, 1972). higher feed concentration as the throughput increased.
Generally the settling zone was of uniform depth with ra- T h e settling zone was of uniform concentration and
dius, but under conditions where the settling zone was very structure throughout and appeared to consist of a coherent
narrow a slight wedge shape was noticed, with the settling mass of floc particles, all interconnected and showing elas-
zone thinning toward the perimeter. This would be expect- tic properties. These acted as a filter bed for the liquid
ed if there was batch settling from this zone. which percolated out of it, showing a sharply defined inter-
The liquid residence time density function is shown in face and a crystal-clear clear-water zone. T h e sedimenta-
Figure 8 illustrating clearly the circulating flow. These were tion from the settling zone appeared t o have the same sur-
obtained from dye and radioactive tracers in the liquid. face and flow structure as those found in batch settling
Due t o the large circulating flow within the thickener and tests.
the narrow settling zone the feed spread very quickly to the Thus during the radial flow in the settling zone, batch
perimeter. Under the actual experimental conditions the settling occurred. With a uniform settling zone concentra-
feed reached the perimeter in 1.5-2 min, and the first trac- tion, settling was proportional to area, and this governed
er appeared back a t the center in 3.5-4 min. T h e upflow the flow velocities. T h e horizontal velocity decreased very
from the perimeter to the overflow was slow in comparison quickly with radius, and solid tracer material in the settling
9 -
12 -
A
. I5 -
E, 0.4 0.8 1.2 1.6 2.0
-I8 settiin> Solids volume fraction kc underflow
concent rotions concentrations
0,
=8 21 - Figure 12. Operating lines of runs no. 5, 6, and 7 (inlet flow rate =
12.2 l./min): - - -, run no. 5: settler mode; run no. 6: settler e.,
g 24-
mode; -, run no. 7: filter mode.
-a
Q
27 -
f 3ob c \ [ teed
concentrat ions A feed flux curve
(Ij
lll
lll
lll
ll B settling zone flux curve
0 04 08 12 16 20 24
Reduced solids concentration kc
Figure 10. Density profiles (or “settler” and “filter” modes (runs 5 F 0.04
and 7): 0 , filter mode half way t o the perimeter; 0,filter mode at
the centre of the tank; A , settler mode half way to the perimeter; 0
CL
A. settler mode at the centre of the tank.
1 I i I i h
-
0.4 0.8 1.2 1.6
settlinb ion; Solids volume traction kc zzx
teed concentrations concentrations
toncenlratlons
0.08 Figure 13. Operating lines of runs no. 8, 9, and 10 (inlet flow rate
A teed flux curve = 11.1 l,/min): -, run no. 8: filter mode; - - -, run no. 9: between
B settling zone flux curve settler and filter mode;. . . run No. 10: settler modes.
a,
p 0.04
V
teed
concentrations
d 0 08
A feed tlux curve
:I=
i D
bulk of displaced water percolated up, with a n occasional Figure 14. Operating lines of runs no. 11, 12, and 13 (inlet flow
rate = 9.6 l./min): -, run no. 12: filter mode; - - -, run no. 13: filter
channel formed to assist this upflow. These channels were
mode,. . .-,run no. 14: settler mode.
destroyed by the disturbances due t o the rake.
Density Profiles. T h e density profile and flow patterns
were taken simultaneously in all the experiments. Two re- centrations, and the underflow concentrations are also
sults are shown in Figure 10 for the thickener under “set- shown on these graphs.
tler” and “filter” modes, respectively. As noted previously,
under all conditions the feed entered the thickener at its Discussion
compatible density level. Density profiles taken a t differ- Under normal operating conditions the sedimentation
ent points in the tank showed no significant differences. tank consisted of three zones: clear liquid zone on the top,
Operating Conditions Relative to the Flux Curves. the settling zone in the center, and the compaction zone
Operating lines for all the runs are shown relative to the re- thickening the discharge.
duced flux curves of the feed and the settling zone pulps re- Sedimentation out of the settling zone appeared as batch
spectively in Figures 11-14. I t can clearly be seen t h a t the settling with segregation. T h e coarser particles settled out
“filter” modes correspond t o the highest throughputs. Also first in both “filter” and “settler” modes, leaving a settling
we may note t h a t the operating lines for all the “filter” zone consisting of feed material minus the coarser fraction
modes lie below the tangent line to the feed flux curve and of particle sizes. T h e settling zone pulp thus displayed
above the tangent line for the settling zone flux curve. This lower batch settling characteristics from t h a t of the feed
is t o be expected as the upper flux curve is not constructed pulp, but settling out of this zone nevertheless appeared as
for the true settling zone material while the operating line batch settling. Flow velocities were sufficiently slow not to
is not drawn for the correct settling zone flux (as some ma- really affect the settling process. The batch settling had su-
terial has bypassed the settling zone). There was insuffi- perimposed on it the effect of the underflow which in-
cient information to draw the correct operating line, due t o creased the batch settling flux by a certain amount.
the feed material containing a distribution of particle sizes. Results in Relation to the Flux Theory. T h e flux
The amount of material bypassing the settling zone is thus theory has been well presented by many workers (Robins,
not known. T h e feed concentrations, the settling zone con- 1964, Jernqvist, 1965). T h e theory is criticized for the fact
t h a t its basic assumptions are not valid for industrial thick- tion of “filter” mode of operation, also give consistent re-
eners and hence the predicted concentration zones in the sults. Actual throughputs are 70-80% of those predicted by
thickener did not occur experimentally. Capacities did Coe and Clevenger. T h e formula of Coe and Clevenger may
show agreement. he readily transposed to read (Tory, 1961)
I t is necessary to clarify what is meant by the flux theo-
ry. The drawing of a flux curve from hatch settling tests
and the associated mass balance line are not fundamental
only to the flux theory. I t is the relationship between the
flux curve and this line and the association of various zones This relationship of Coe and Clevenger can be derived
in the thickener with this graph which is unique to the flux from the basic assumptions of the flux theory, which helps
theory. to understand the agreement between throughputs and
T h e basic assumption of the flux theory is the postulate throughputs calculated by both methods. Thus eq 2, a t a
of the feed being distributed evenly and continuously specific concentration, defines the capacity for both meth-
across the entire area of the thickener. Then due to the set- ods. T h e flux theory, however, defines the capacity of the
tling flux a t the feed concentration being greater than the thickener as limited by the minimum value of eq 2, which
feed flux, the concentration immediately drops to a lower occurs a t the lower conjugate concentration. Coe and Clev-
concentration (upper conjugate concentration) where the ener define thickener capacity as the minimum value of eq
settling flux equals the feed flux. This is contrary to the 2 between the feed concentration and the highest concen-
concept of the feed entering a t its compatible density tration of free settling pulps. These two minimum values
(Fitch and Lutz, 1960; Sawyer, 1956), as seen in the present are the same, unless the minimum value of the flux curve is
work. With the upper conjugate concentration lower than a t a concentration either lower than the feed concentration
the feed concentration we would expect the feed to cascade or higher than the limiting concentration of free settling
t o a point below this zone and then enter a t its compatible pulps. Thus generally we would expect equivalent through-
density level. T h e floc bed a t the upper conjugate concen- puts from both methods.
tration, if it exists, must then exist above the feed position In the present work the capacities for both methods are
and not below it. T h e present work also showed the neces- the same for runs 2 and 7 . In these two runs the lower con-
sity for a separate flux curve for the settling zone, due to jugate concentration falls between the feed concentration
the segregation of the feed. and the highest concentration of free settling pulps. In runs
In the compaction zone limitations of predictions by the 8, 12, and 13, however, the lower conjugate concentration is
flux theory have been expressed (Robins, 1964). The theory a t a concentration below the feed concentration, and thus
predicts three concentrations; upper and lower conjugate the predicted flux theory capacity is lower than the Coe
concentrations and the discharge concentration. The na- and Clevenger capacity.
ture of the compaction zone is not predicted, and the theo- The zones and their concentrations described by Coe and
ry considers the thickening as independent of height. The Clevenger (1916) to exist in a thickener closely approxi-
effect of overlying solids (or depth of compaction zone) is mate the conditions obtained under “filter” mode of the
not considered unless we use the flux envelope of Tory present work, with a settling zone concentration fairly close
(1968). to the feed concentration. Two distinct differences oc-
Capacities. The predicted capacities of the thickener curred.
operating a t maximum capacity, as postulated by the flux (i) The coarse material effectively bypassed the settling
theory, and the predicted concentration zones are shown in zone resulting in a settling zone with settling rate lower
Table 111. These are compared with the actual conditions than that expected from hatch tests on the feed material.
for the thickener operating under “filter” mode in runs 2, 7 , Although not all the material entered the settling zone it
8, 12, and 13. For runs 8, 12, and 13, where the feed concen- was nevertheless rate limiting with regard to capacity
tration is greater than the lower conjugate concentration, under “filter” mode.
the upper conjugate concentration is determined by the (ii) Coe and Clevenger batch tests are expected to show
normal method. enhanced settling rates a t intermediate concentrations due
Although flux theory capacities are in fairly consistent to the induction period which allows a settling mechanism
agreement with actual capacities, the upper and lower con- to develop. Thus with concentrations in the intermediate
jugate concentrations were not found to exist. This can be range, up to the highest concentration of free settling
seen to he true even for an upper conjugate concentration pulps, settling rates are expected to he lower than Coe and
determined by a fictitious settling zone operating line Clevenger predictions.
drawn on the settling zone flux curve. There does not ap- The combined result of (i) and (ii) is actual throughputs
pear to he any direct correlation between the lower conju- consistently lower than Coe and Clevenger capacities.
gate concentration and the settling zone concentration. Irrespective of the design method used, the problem of
Coe and Clevenger capacities, calculated for each condi- measuring the flux under actual operating conditions ex-