Bimodality has been observed in the PSD from batch ... crystal dissolution during PSD measurements. .... that small uncertainties in the PSD can make a large.
0263±8762/97/$07.00+ 0.00 Institution of Chemical Engineers
EXPERIMENTAL INVESTIGATION INTO DYNAMICS AND STABILITY OF CONTINUOUS MSMPR AGGLOMERATIVE PRECIPITATION OF CaCO3 CRYSTALS J. A. WOÂJCIK and A. G. JONES (FELLOW) Department of Chemical and Biochemical Engineering, University College London, UK
E
xperiments were carried out to study the dynamic and kinetic behaviour during the agglomerative precipitation of CaCO3 in an MSMPR crystallizer. The results demonstrate that for certain conditions steady state can be achieved earlier for such a process than for simple crystallization. Bimodality of the PSD was once more con® rmed. Kinetic constants for nucleation, growth, aggregation and disruption were estimated by non-linear least squares ® tted to a discretized MSMPR population balance model. Preliminary results indicate a dependence of both crystal aggregation and agglomerate disruption on crystal growth rate, but to reverse extents. Keywords: agglomerative precipitation; crystal growth; nucleation; aggregation; disruption; dynamics; CaCO3
and Randolph14 quantitatively studied aggregation and disruption of calcium oxalate dihydrate crystals in urinelike mother liquor. To separate the in¯ uences of growth and nucleation from aggregation and disruption, they used a mininucleator (MSMPR crystallizer) and aggregator (Couette-¯ ow crystallizer) operated in series. As a part of a continuing study, a dynamic and kinetic analysis of agglomerative precipitation is reported using an aggregation-disruption model of crystal agglomeration. Preliminary results indicate for the ® rst time that both processes are dependent on crystal growth rate.
INTRODUCTION The MSMPR crystallizer exhibiting simultaneous nucleation and growth is generally stable, taking 10±15 residence times to come to steady-state (Randolph and Larson1 ). Secondary particle formation phenomena such as crystal aggregation and particle breakage, however, can have a substantial e ect on the product PSD. This is especially the case for precipitation processes, but the e ects of these processes on crystallizer dynamics and stability is not yet fully understood. For example, Beckman and Farmer2 reported that steady state was achieved in a MSMPR precipitation in only 6±8 residence times. Chakraborty et al.3 obtained steady state for speci® c conditions even after 3±4 residence times. Bimodality has been observed in the PSD from batch (SoÈhnel et al.4 , Wachi and Jones5, SoÈhnel and Mullin6,) semi-batch (Tavare and Garside7 ) and continuous MSMPR (Beckman and Farmer2 , HostomskyÂand Jones8 , Tai and Chen9 ) crystallizers. Such bimodality might be caused either by a size-dependent or dispersed aggregation rate, agglomerate disruption, Ostwald ripening, or a combination of these, but there is as yet no general agreement as to the mechanism. In previous kinetic studies, mentioned above, agglomeration via aggregation alone has been reported. Halfon and Kaliaguine10 found breakage in their system (batch alumina trihydroxide crystallization) but did not account for it in their model. Remillard et al.11 studied breakage only for the same system. Synowiec et al.12 investigated theoretically and experimentally particle attrition in dilute turbulently stirred suspensions and found a considerable e ect of solution turbulence on PSD. Hartel et al.13 and Hartel
EXPERIMENTATION Continuous MSMPR CaCO3 agglomerative precipitation was investigated experimentally in a 3 ´ 10- 4 m3 crystallizer similar to that described by Zacek et al.15 in a laboratory installation like that used by Budz et al.16 (Figure 1). Initially, the assumption of ideal mixing was tested in order to justify use of the MSMPR model. The impeller rotation rate and applied torque were read from the display of a Heidolph RZR 2101 stirrer. The power curve was determined in the range of measured torque values. Then several mixing criteria were calculated: tM
Ö
= 12
-----
q
g
2
(SoÈhnel and Garside17 ) (Po.Re)3/ 8 Re3/k 4 = 1.4 >> 1 H 1/ 8 Dt 1/ 4 L L (Pohorecki and Baldyga18 )
( )( )
113
WOÂJCIK and JONES
114
Table 2. Range of changes of experimental conditions. Parameter
min
residence time/min stirrer speed/min- 1 solution concentration/M pH
tms =
( )( ) z L
2
Dt L
2
1/ 3 Dt H 1/ 3 1 L L (Pohorecki and Baldyga18 ) ln Sc - 1.27 tmd = 1/ 2 0.324 2
Po1/ 3
(
3.248 600 0.025 8.84
13.146 2000 0.20 11.45
observed. The PSD was monitored by very frequent sampling (1¸ 10 min) and evaluated by a Coulter Counter Multisizer II with the 100 l m aperture together with a Malvern Zetasizer 3. An especially clean electrolyte saturated with CaCO3 was prepared to avoid crystal dissolution during PSD measurements. [Ca2+ ] was continuously determined using an EDT Calcium QSE 310 electrode covered by ® lter paper to avoid encrustation of the membrane surface. The solids holdup was estimated by an Analite Portable Nephelometer 156. The pH values were read from an a lfa 500 pH meter. For some of experiments, drops of suspension were taken as a sample, dried for ca two hours on the surface of a ® lter paper and ® nally prepared for SEM analysis.
Figure 1. Schematic diagram of the apparatus (adapted from Budz et al.16). A-Na2CO3 stock solution tank; B-Ca(NO3)2 stock solution tank; C-crystallizer; D-level control; E-discharge valve; F-constant pressure head device; G-peristaltic pump; H-pH electrode; I-ion selective electrode; J-reference electrode; L-inlet ¯ ux control valve.
2.07
max
)( )
( ) v
(Pohorecki and Baldyga18 ) VR 19 s (Baldyga et al. ) nd3 1, tms + tmd
