KINETICS OFCONTINUOUS PRECIPITATION IN A DILUTE SYSTEM; MAGNESIUM HYDROXIDE ‘ Thesis for the Degree of M. S. M!CHIGAN STATE UNIVERSITY PHILIP B. HAR'MICK 1975 .1 unélé EX ABSTRACT KINETICS OF CONTINUOUS PRECIPITATION IN A DILUTE SYSTEM; MAGNESIUM HYDROXIDE By Philip B. Hartwick This study constitutes the second step of a pro- gram attempting to model the removal of dilute impurities from water streams using precipitation, agglomeration, and flocculation. The lime-soda ash process currently being used in many municipal water softening plants was the process under investigation. This process is also being looked at as a probable component of total recycle in industrial cooling water systems. In a previous study, it was shown that the water softening process can be par- tially described using crystal nucleation and growth kinetics. This research investigates the kinetics of pre- cipitation of magnesium hydroxide. Crystal size distri- bution (CSD) in a continuous backmix reactor at various residence times was determined as a function of super- saturation. McCabe's AL law (size independent crystal growth) was found to apply, so that population balance theory could be used to describe the CSD. Nucleation Philip B. Hartwick and growth rates were determined as a function of mag- nesium and hydroxide concentrations at 25°C. B° = nucleation rate (nuclei per minute per m1) 0 = 12 2 _ -]_O B 3.95 x 10 (COH CMg 6.95 x 10 ) G = growth rate (microns per minute) _ 5 -8 G - 3.83 x 10 (COHCMg - 2.97 x 10 ) In the light of these findings, several design changes to improve the separation process are recommended: 1. Larger rapid mix volume. 2. Diversion of some of the slaked lime feed to the flocculation basins. 3. Recycle of primary sludge in systems where this is not already practiced. The first two suggestions would decrease operating hydroxide concentrations in the rapid mix area, resulting in lower nucleation rates and, as an end result, a higher solid fraction in the sludge. Primary sludge recycle would have the same effect, while increasing the surface area for deposition to maintain the quality of the treated water. Further investigations with the mixed precipita- tions, characteristic of lime-soda systems, are necessary to verify the above recommendations. KINETICS OF CONTINUOUS PRECIPITATION IN A DILUTE SYSTEM; MAGNESIUM HYDROXIDE By Philip B. Hartwick A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Chemical Engineering 1975 This work is dedicated to all people who drink water. ii ACKNOWLEDGMENTS I would like to thank all those who in any way helped me further my studies so that I might now attempt to further the science of water purification. In par- ticular I would like to thank my Professor, P. M. Schierholz, for introducing me to the field of particulate processes and guiding my investigation; my wife Tracey for stabilizing my existence and sharpening my tone; and my assistant, G. R. Bailie, whose companionship and effort helped get the data in. iii TABLE OF CONTENTS Page LIST OF TABLES . . . . . . . . . . . . v LIST OF FIGURES . . . . . . . . . . . . vi NOMENCLATURE . . . . . . . . . . . . . viii INTRODUCTION 1 BACKGROUND 4 Kinetics Determination 7 Lime-Soda Water Softening . . . . . 8 The Problem . . . . . . . . . . . l4 EXPERIMENTAL . . . . . . . . . . . . . 15 RESULTS . . . . . . . . . . . . . . 25 DISCUSSION . . . . . . . . . . . . . 44 CONCLUSIONS . . . . . . . . . . . . . 57 RECOMMENDATIONS . . . . . . . . . . . . 59 Lime-Soda Ash Water Softening . . . . . 59 Lime-Soda Ash Modeling Study . . . . . 59 Appendices . . . . . . . . . . . . . 60 A. A PROGRAM FOR FITTING COUNTS TO THE SIZE INDEPENDENT GROWTH MODEL . . . . . 61 B. METHODS OF MAGNESIUM.AND HYDROXIDE ANALYSIS . . . . . . . . . . . . 65 C. DATA AVERAGING . . . . . . . . . . 69 BIBLIOGRAPHY . . . . . . . . . . . . . 76 iv LIST OF TABLES VTable 7 Page 1. Results from material balance . . . . . 27 2. Results from crystal Size distribution . . 28 LIST OF FIGURES Figure 1. Lansing water softening process 2. Mini-plant for precipitation studies 3. Crystallizer less feed and withdrawal tubes . . . . . . . 4. Bail and beaker sampler 5. Equipment list and manufacturers 6. Typical crystal size distribution . 7. Solid crystal with a clump of small flake aggregates . 8. Detail of Figure 7, edge shot of star aggregate . . . . . . 9. Crystal flakes and aggregates, thin edge 10. Unusual "sponge" crystal formed at 20 minute residence time, Run 6-3 . 11. Solid crystal, fairly common at longer residence times, Run 6-3 . 12. Unusual hollow "needle” crystal, Run 6-3 13. Detail of Figure 12, point of needle crystal . . . . . . . . 14. Large solid crystal, plate-like appearance, Run 5-15 . . . . 15. Large clump of flake aggregates, most com- monly seen crystals in the laboratory samples, predominant at shorter resi- dente times, Run 5-15 16. Agglomerate taken after the rapid mix area of the Lansing drinking water treatment plant, representative formation . vi Page 10 16 18 21 23 26 30 30 32 32 34 34 36 36 38 38 Figure I Page 17. Layered finger, representative of sample taken after the rapid mix . . . . . . 40 18. Representative agglomerates from.the first flocculation basin . . . . . . . . 4O 19. Representative agglomerate from.the third flocculation basin . . . . . . . . 42 20. Representative agglomerate from the fifth and last flocculation basin . . . . . 42 21. Film theory for magnesium.hydroxide growth . . 45 22. Film theory for induction of high ionic strength near a growing crystal . . . . 47 23. Nucleation rate as a function of reactant concentrations . . . . . . . . . 50 24.‘ Growth rate as a function of reactant' concentrations . . . . . . . . . 51 2 25. Growth rate versus CMg COH . . . . . . 52 26. Nucleation rate versus CMg C0H . . . . . 53 vii NOMENCLATURE Description Nuclei birth rate, numbers per milliliter per minute. Magnesium concentration, equivalents per liter. Hydroxide concentration, equivalents per liter. Crystal growth rate, microns per minute. Growth constant. Nucleation constant. Shape factor, for a sphere n/6, this was the value used in this study. The characteristic dimension of a particle, determined by the increase in resistance across an aperture with an electrolyte moving through when a particle is present. Reported as an equivalent spherical diameter, microns. Suspension density calculated from material balance, grams of crystals per liter of solu— tion. Suspension density calculated from.the crystal size distribution, grams of crystals per liter of solution. Particle density, expressed as a function of L, numbers per milliliter per micron. Nuclei density, the particle density at L = 0, numbers per milliliter per micron. Rate of make of magnesium into crystals, cal- culated from.the material balance, equivalents per minute. Rate of make of hydroxide into crystals, cal- culated from.the material balance, equivalents per minute. viii Description Crystal density, taken as literature value for stable crystals. Supersaturation, solution concentration less equilibrium concentration. Growth supersaturation, an experimentally determined expression. Nucleation supersaturation, an experimentally determined expression. Mean residence time in the crystallizer, minutes. Free settling velocity (terminal velocity) of a particle moving through a fluid due to gravitational force. Average volume of crystallizer. ix INTRODUCTION The lime-soda ash system has been used extensively to improve the quality of municipal water supply. The process has been studied rather extensively; indeed, the Journal of the American Water Works Association regularly delves into its murky waters. A school of thought, of independent origin, with roots in Tucson and Ames, deal- ing in analysis and techniques of continuous crystalliza- tion, has presumed to challenge the aged knowledge of the AWWA. The proof, of course, is in the pudding. Can the system be improved upon? The primary tool of these upstarts is kinetic studies; growth and nucleation rates are related to each other and to supersaturation. P. M. Schierholz (25) kicked off the attack on lime-soda ash water softening by determining relationships between nucleation and growth for calcium carbonate precipitation. In this project magnesium hydroxide precipitation was considered. Between the two, the predominant constituents of water hardness were individually scrutinized. The necessity for a clearer picture of lime-soda ash water softening becomes apparent when considered in the context of the ecology movement in this country. Public outcry against pollution in the strident sixties resulted in the Federal Water Pollution Control Act of 1972, one result of which is an accelerating trend towards water recycle in a broad spectrum of industrial application. For example: cooling water is often recooled by evaporation for reuse, with blow down water being bled off and replaced with fresh water, to keep ionic levels out of the supersaturation range, preventing scaling of heat exchange equipment. Elimination of blow down necessitates an alternative method of hardness control. A current study by Hartline (12) indicates that the lime-soda ash process is preferred. Unfortunately, application to a recycle system.brings up problems not evident in a once-through system like municipal water supply. For instance, in present practice polyphosphates are added to the treated water prior to sand filtration to eliminate further deposition of crystals on the filter sand. This would likely have a ruinous effect on the precipitation reaction in a recycle system, Polyphos- phates are also added to cooling water to reduce scaling, compounding the problem. If by improving the efficiency of the precipitation process, operating hardness levels could be economically reduced to a point where scaling was no longer a problem, then polyphosphates could be eliminated from the system. The goal of this series of investigations is economical and operational improvement in the lime—soda ash water softening process. Under the conditions of low suspension density, short residence times and small crystal size distribution, the phenomena of crystal growth and nucleation were isolated and modeled for magnesium hydroxide precipitation in the absence of other precipitating compounds. Techniques employed.with success by Randolph and Larson (22) on concentrated systems and adapted by Schierholz and Stevens (26) to dilute systems were used in evaluation of crystal size distribution. BACKGROUND The problem of characterizing a population of particles produced in a system.is not a simple one. If the system involves the formation of particles of a 5: single habit (characteristic shape) then a particle of ii a given size can be formed by three mechanisms: growth --a smaller particle can grow by solute deposition; E attrition--a larger particle can break due to mechanical action; and agglomeration--smaller particles can come together forming a larger one. A fourth mechanism, flocculation, occurs when a flocculating agent tends to sweep other particles out of suspension, forming a much larger aggregate. Magnesium hydroxide is considered to be a flocculating agent in lime-soda water softening (31). In a flow system with no crystals in the feed, yet another mechanism may occur, nucleation--the birth of very small crystals. The picture is simplified con- siderably if the crystals do not interact with one another or with the crystallization vessel, in which case the crystals don't break (attrition) or stick together (agglomeration and flocculation). Under this assump- tion, nucleation and growth are the only means by which a particular size may be reached. The no interaction condition is approached in dilute slurries with short 'residence times. The driving force for precipitation, super- saturation, has been defined by Mullin (19) as the difference between solute concentration and equilibrium concentration. unfortunately, most systems investigated in the past have involved immeasurably low levels of supersaturation. Temperature difference in cooling crystallizers has been used as a measure of super- saturation. Ideally, the kinetic parameters (nucleation and growth) should be modeled in terms of the concen- trations of the reacting species for isothermal crystal- lization (22). In dilute systems relative supersatura- tions are much higher than in most concentrated systems studied in the past, raising the hope that results from this investigation can correlate supersaturation with nucleation and growth rates. The traditional definition of supersaturation lacks generality when there are more than one reacting species. A general model cannot assume stoichiometric ratios or constant concentration of one of the reactants. The definition of supersaturation should be modified to apply to ionic precipitation reactions. It has been shown that new crystal formation can result from homogeneous nucleation, heterogeneous nucle- ation, and secondary nucleation. Homogeneous nucleation is the formation of a new crystal from the liquid phase as a result of supersaturation alone. Heterogeneous nucleation normally involves new crystals formed around minute foreign material. The foreign particles provide sites for deposition with reduced energy requirements. Secondary nucleation is throught to involve crystals bumping into one another creating nuclei. Randolph and Larson (22) model secondary nucleation rate as propor- tional to the mass of solids in suspension. In light suspensions secondary nucleation should not be an important factor. Homogeneous nucleation has been reported for magnesium hydroxide in batch studies (16) with nucleation rate dependent on (CMg x COH2)33 at supersaturation ratios as low as four. In other batch studies, Gunn and Murthy (11) reported induction times in terms of concentrations: At (sec) = 1.389 x 10'15 (C 1C 2)-l.85 . . . Mg OH When the ionic reactants are mixed together in a supersaturated condition nuclei may even- tually appear. The time interval between mixing and the appearance of crystals is known as the induction period. The relation between induction period in batch studies and nucleation rate in continuous systems is rather obtuse, but some correlation should exist. Growth rates for magnesium hydroxide crystals are not reported in the literature. Schierholz (25) in his investigation of calcium carbonate reported growth rates but found no correlation to reactant concentra- tion. It is well known that for small crystals (less than five microns) growth rate depends on size (19). Randolph and Larson (22) mention that for most highly hydrated crystals growth is a function of size, with alum as an exception. Magnesium hydroxide is known to be a highly hydrated crystal (19). The McCabe AL law states that growth rate is independent of size. For crystals larger than a few microns this has been found to often be the case (22, 25). Kinetics Determination A mixed suspension mixed product removal (MSMPR) crystallizer was used to determine the kinetics of mag- nesium.hydroxide precipitation. Population balance theory (Randolph and Larson, 22) was applied to express the crystal size distribution in terms of nuclei density, growth rate and mean residence time. Assuming steady state, no crystals in the feed, no attrition, agglomera- tion or flocculation, no classification of withdrawal, perfect mixing, and growth rate independent of size, application of a population balance results in, dn 11 _ dL'+ G? - O . Solution of the differential equation yields, n = no exp (-L/GT) , where n is the particle density of crystals with a characteristic dimension L, G is the crystal growth rate and T is the mean residence time. The nucleation rate may be expressed as, Lime-Soda Water Softening Water taken from wells or streams is often soft- ened before distribution. Hard water is water containing objectionable amounts of dissolved salts of calcium.and magnesium" In large municipal systems, the raw water is softened via a lime-soda ash or lime process to a finished product containing 1.6-2.4 meg/1 of magnesium and calcium. The two types of water hardness are carbonate (temporary) and noncarbonate (permanent). Temporary hardness, calcium.and magnesium bicarbonate, is removed by the addition of lime, resulting in the precipitation of calcium carbonate and magnesium hydroxide. The reac- tions in lime treatment are: Ca(HC03)2 + Ca(OH)2--+ 2 CaC03+ + 2 H20 Mg(HC03)2 + Ca(OH)2-——+ MgC03 + CaCO3¢ + 2 H20 MgCo3 + Ca(OH)2--+ Mg(OH)2+ + CaC03+ Noncarbonate hardness is removed by addition of lime and soda ash. The additional reactions for lime-soda ash tre atment are : MgClz + Ca(OH)2--* Mg(OH2)+ + CaClZ Mg804 + Ca(OH)2-——» Mg(OH)2+ + CaSO4 CaCl2 + Na2C03-——+ CaCO3+ + 2 NaCl CaSO4 + Na2CO3--—+ CaCO3+ + NaZSO4 Two schemes of lime-soda water softening are currently in use. Split treatment is represented by Lansing's water softening facilities while Dayton often uses only lime softening in a one-shot system. The following description of split treatment is based on the Lansing and Dayton systems. Lansing's sys- tem is diagrammed schematically in Figure l. Dayton's system follows a similar flow chart with a different physical configuration. Rapid Mix Quick lime (CaO) is slaked with water in the chemical feeder to form calcium.hydroxide. The slaked lime is then added to about 80 percent of the raw well water in the rapid mixer. One equivalent of lime is required for each equivalent of calcium while two are lO nonwoaooou maaoma no aflom 0H onu ou umson mwvaam mumvcooom machomm “one; was scam mwvsflm umuaau seem flLl weapon woflauumm Hmawm \\ \ \\. heapaooom 5mm moom .mmmUOHm wnficmuMOm HoumB wawmamqun.a ouswwh mas scam humafium uoflwuamomu no vaom OH “mum: - mweeam swoon mawauuom \ \. \\.\. \. .IIIIIII. canon was comma «and . vexeam sow ewe mN NON “mums HHm3 3mm 11 required for each equivalent of magnesium. Excess lime is added to drive the reactions. Flocculation In the series of flocculation chambers the reactants are given time to take the precipitation toward equilibrium. Nucleation, growth, agglomeration and flocculation are thought to take place here. The mag- nesium.hydroxide acts as a flocculating agent facili- tating the removal of fine calcium carbonate crystals (31). Primary Settling The purpose of the settling basis is to separate the fine crystalline materials from the water. No agi- tation is used and the crystals sink to the bottom where scrapers take them to a sludge sump. Carbonation If a once-through system is used the excess lime can be neutralized by bubbling carbon dioxide gas through the water: Ca(OH)2 + C02-——+ CaCO3+ + H20 Any excess carbon dioxide is likely to dissolve calcium carbonate: H _ 12 The carbon dioxide is normally available from.the recal- ciner where the sludge is fired to reform the quick lime used in treatment: CaCO3 -Z* Ca0 + C02+ At Dayton, one-pass primary treatment with carbonation is used when part of the system is down for maintenance. Back-up underwater burners are in place to supply C02 should the recalcining kiln be down. Secondary Rapid Mix Here water from the primary settling basis, by-passed raw water from the wells, soda ash and recircu- lated secondary sludge are mixed. The soda ash precipi- tates noncarbonate hardness and the recycled secondary sludge facilitates formation of large crystals as well as providing surface area for deposition. Following secondary rapid mix are flocculation and settling chambers much the same as in primary treat- ment with an additional settling chamber to reduce fine crystals in the water prior to filtration and distribu- tion to the customer. The sludge is either dewatered, recalcined and used as quick lime or diverted to a disposal pond where it slowly dries and compacts. 13 Design consideratons are multifaceted. In Lansing the primary flocculating chambers have recently been modified to facilitate the movement of crystalline floc from one to another. This was accomplished by moving the connecting slots from midway up the walls dividing the flocculation basins to the bottom of the walls. The motive was to reduce crystalline buildup on the agitators by reducing suspension density. This move did reduce maintenance costs. However, reducing suspen- sion density is counter to literature recommendations (14, 22) which suggest sludge recycle to improve sludge settling characteristics tut increasing suspension density. Sludge in the Lansing plant has taken a turn for the worse since the modifications. Previously, 15 percent solids was not uncommon; presently, 10 percent solids is normal. The result is a 20 ton per day cal— cining plant can only produce 16 tons of quick lime daily. The economic trade-off is a net loss. Sludge recycle at Dayton is profitably being used in the primary rapid mix area resulting in high suspension density and higher sludge solid compositions (30). At Dayton the soda as is added in the primary rapid mix rather than the secondary. This practice seems reasonable because the pH is higher in the primary system: high enough to keep all of the C02 in the carbonate form. In Dayton the secondary rapid mix basin mixes the water from the 14 primary with by-pass water, secondary sludge, and potato starch; a flocculant. The Problem The experimental phase of this investigation focused on the kinetics of precipitation of magnesium hydroxide in a (MSMPR) crystallizer. The mathematical tool used is the population balance. This is in addi- tion to the traditional chemical engineering practice of using material and energy balances to characterize separation processes. The purpose of the experiments was to determine the nucleation and growth rates at various residence times and reactant concentrations, the ultimate objective being to determine nucleation and growth rates as a function of reactant concentra- tions. Temperature is also recognized as an important parameter; however, all experiments were conducted at 25°C. The use of various residence times served as an indication of the generality of the kinetic data obtained. EXPERIMENTAL A series of experiments were conducted in a mixed suspension mixed product removal (MSMPR) crystal- lizer system, assembled by the author. Previous systems described by Schierholz (25) and Randolph and Larson (22) were used as guides. Experience also became a guide. The system described in the following discussion is diagrammed in Figure 2. Feed solutions were made using distilled water. A solution of magnesium chloride was held in the Mg reservoir, and a solution of sodium hydroxide was held in the OH reservoir. Crystallizer concentrations were intended to range around those encountered in lime softening. Without knowing the kinetics of precipitation, the concentrations in the crystallizer could not be determined from the feed con- centrations, making crystallizer concentration predic- tion an iterative process from one run to the next. A styrofoam.float in each reservoir suspended a cooling coil consisting of tygon tubing wrapped around a stainless steel frame with tap water as a coolant. Also on the float was a Haake temperature controller set at 25°C. Two rotary pumps supplied water to con- stant head tanks, by way of two wrapped string 5 micron 15 .mmwpaum nounumuwmwooum How unmamnwawz:n.m shaman piano comm uwobummmm mo Houunoo 3oam Hwoo wafiHoou cowhH umoam 16 Houuaoo .mamH ooo L emu aoum choppy 36amuo>o mason room mason vooammeAMHum> amen uaoumuoo 17 filters. Since the filters were rated for flow rates of 50 gallons per minute and used at rates around .04 gallons per minute the author presumes that only very small particles could be swept through the filter. The feed pumps consisted of two tandem peristaltic pumps driven by an 1800 rpm.motor. Flow rates were measured at the beginning and end of each run with a 250 ml volumetric flask and a timer. Often no measurable dif- ference was found. Residence time was varied between 8, 12 and 20 minutes by use of an adjustable gear reducer between the motor and the feed pumps. The crystallizer is diagrammed in Figure 3. To prevent crystal build-up on the reactor surfaces all inside parts were scraped every five minutes with a rubber spatula. In runs where the crystallizer was not scraped, steady state in the crystal size distribution and the reactant concentrations was not obtainable. Baffles and draft tube are considered essential for complete mixing in high suspension density systems. They were not used in this system to minimize area for deposition, and to facilitate scraping. In this low suspension density system they seemed to be dispensable. To obtain isothermal conditions the room temperature was maintained at 25°C. Crystallizer temperature varied between 24.9 and 25.1°C. An agitator speed of 500 rpm was used in all cases. Withdrawal was accomplished 18 Time delay Mercury switch 17.3cxn l6fi7cm *I T? L k Figure 3.--Crysta11izer less feed and withdrawal tubes. l9 intermittently using a float and a mercury switch to activate a pump. Each withdrawal pulse removed about 325 ml or 8 percent of the 4.1 liter reactor volume. Intermittent withdrawal has been found to be an effective tool for reducing classification in systems of a difficult nature (ut - free settling velocity > 3 cm.s-1) (1). Although this system is classified as an easy problem (ut < .5 cm 3'1) with maximum calculated values of ut equal to .075 cm 3’1, intermittent with- drawal was used to satisfy prejudices built up over the years. One unfortunate facet of intermittent withdrawal is that it imposes a cycling residence time on the tank, accentuating any tendency for instability in the crystal size distribution. Aeschback (1) developed a crystallizer design that, although not used in this investigation, might insure complete mixing in unbaffled operation. The bottom is convoluted to follow flow streamlines and the impeller blades are reversed so center flow is upward allowing higher stirring rates without vortexing. Previous investigators reported that steady state in a (MSMPR) crystallizer is reached after from 8 to 15 residence times (2, 4, 21). In this investiga- tion, sampling to determine crystal size distributions and reactant concentrations was commenced ten residence times after startup. 20 Samples for analysis were removed from the crystallizer with a bail and beaker to avoid any inaccuracy that might arise due to nonideal mixing of the feed streams at the surface. Crystal size distribu- tions (CSD) were determined on a size range from 8 to 25 microns, using a Coulter counter. Due to the rather primitive nature of the counter a separate analysis had to be made for each point on the crystal size dis- tribution curve. Two points were obtained from a single sample, within 30 seconds of its removal from the crys- tallizer. Points were taken in pairs starting with the lowest size countable and terminating when the count fell below 15. Each count or point consisted of all particles in a 2 ml sample exhibiting an equivalent diameter greater than a lower threshold and less than an upper threshold. The raw data was fit, by least squares, using the model [n = n° exp (-L/GT)], to deter- mine the growth rate G and the nuclei density n°. The computer program used for this fit can be found in Appendix A . Analysis of reactant concentrations started with the bail-beaker system. About 150 ml of slurry was then filtered using fine filter paper in a buchner funnel. Hydroxide concentrations were determined by titrations of a 50 ml sample against 0.10 normal hydrochloric acid. A pH meter was used to determine the end points for 21 L- LLLLU. LL. MAMA ' J Beaker up--Empty Figure 4.--Bail and beaker sampler. total and phenolpthaline alkalinities. The iterative technique used to back hydroxide concentrations out of alkalinity measurements, its assumptions and defense, can be found in Appendix B. The same sample was then titrated against EDTA (ethylene diamine tetra acetic acid), to determine magnesium as described by Diehl (2). By lowering the pH to 4.9 in the first titration, the problem of an indistinct EDTA endpoint described by Schierholz and encountered in this work was avoided. The indistinct nature of the EDTA endpoint for a fresh 22 sample whose pH hadn't first been lowered to 4.9 might be attributed to some kind of a prenucleation quasi- stable association between hydroxide ions and magnesium ions. Alternative methods to determine reactant concen- trations were considered, such as direct pH measurement of OH concentration, specific ion electrodes for the magnesium concentration, and conductivity measurements coupled with material balances for both concentrations. All were discarded for inaccuracy, lack of funds or lack of laboratory time for calibration. See Appendix B for a discussion of methods of hydroxide and magnesium concentration determination. When the operator felt steady state had been reached the hydroxide reservoir was sampled and titrated to determine feed hydroxide concentration. This was necessary because the feed solution was never at equilibrium with the carbon dioxide in the air. 002 from the air continually dissolved in the feed, converted to carbonate and used two hydroxides in the process. - -2 co2 + 2 OH -——+ H20 + CO3 At the pH (ll-11.5) where the experiments took place the driving force for this reaction was large, resulting in substantial hydroxide concentration drops overnight (on the order of 2 x 10'4 eq/l). a Temperature controllers Pumps feeding constant head tanks Filtersc Filter linersc Feed pump drived Feed pumps Mercury switch for level controle Stirrer drivef Withdrawal pumpb Particle counterg Filter paper pH Meter Air conditioner 23 Haake Model E52 Eastern Industries Pump Model D-ll Filterite Model moms Five Micron, Filterite Wrapped String ~C5AlOS Gill Electric Motor, Type C3/SYNCH with GEAR REDUCER Cole Parmer Master Flex Model 7017 Mercoid Corp. Switch Model 9-51H Lightnin Mixer Model E Eastern Rotary Pump Model VT-S Coulter Counter Model B No. 42 Whatman Corning Model 10 12,000 BTU Comfort-Aire Figure 5.--Equipment list and manufacturers. aHaake Inc. P.O. Box 610 244 Saddle River Road Saddle Brook, N.J. 07662 bEastern Eng. New Haven, Conn. cFilterite Corp. eMercoid Corp. Chicago 41, Ill. fMixing Equipment Co. Rochester, New York gCoulter Electronics 580 W. 20th St. Hialeah, Florida Timonium, Maryland 21093 dGill Electric Brighton, England 24 Between ten residence times and the end of the run, a sample of solution was filtered through a .46 micron filter and the crystals stored over Dryrite in a desiccator for later viewing under a scanning electron microscope. The electron microscope investigation was somewhat of a fishing trip; however, proper interpre- tation could lead to better understanding of crystalliza- tion. Photomicrographs were also taken of crystal samples from.the Lansing lime-soda water treatment plant for qualitative comparison. RESULTS Data available for the analysis of each run con- sisted of the following: --4 to 8 sets of numbers vs. characteristic length data (crystal size distributions). --Feed and crystallizer concentrations for Mg, OH, CO3, HCO3, Cl and Na. --Feed rates and reactor volume. The feed rates and reactor volume give residence times. When coupled with feed and crystallizer concentrations in a material balance, rate of make (eq/min) and sus- pension density (gm/l) can be calculated. If the crystal size distributions for each run fit the model, as they did (see Figure 6 for a representative plot), the CSD data can be reduced to growth and nuclei density versus time plots. The third moment of the CSD also gives a suspension density (22). All of these values and plots were calculated, and plotted in Appendix C. The growth and nucleation rates were averaged for each run, primarily to smooth out oscillations in the nucleation values. Results for the nine runs are tabulated in Tables 1 and 2. 25 26 \‘—n° 10,000 F‘ E; \ '3 \ .H \ a \ x \ ,4 \ é \ \, \ =5 1,000 \ z T‘ Slope = -L/GT e. / 3.. ‘5 Z v-l (U .U ‘9. 5 100 —- {:1 1, 1 l 1 J J 0 ' 4 8 12 16 20 24 L Characteristic Length, microns Figure 6.--Typica1 crystal size distribution. Run 5-24, Series 13:20. 27 oHo. mm.H mo.N No.m N¢.N mH.mN no.m wlo Mao. mq.a qw.a mm.~ Nm.q wH.om mo.w m no 0N0. mn.H Ho.a mm.H no.H n<.Hm mm.om m to owe. mm.H w¢.H Hw.N Hm.H m~.om Ho.ma swim mac. mm.H ¢¢.H ¢H.~ qa.¢ NN.¢H ow.NH «mum wHo. mm.H mo.a mN.N nm.q wH.©H mm.NH NNIm mmo. NH.H Hm.H mo.H Ne.m em.NH em.om mHIm mHo. oo.H om.N mo.m Nu.m qq.qa Nu.n malm NNo. wN.H OH.N mo.m Hm.o mm.ma oq.NH oalm H\aw w «oaxcfla\co coaxafla\vm qoaxa\cm Oaxa\vo mousafis kuamamn oaufiflmmxmz Bowmmcwmz mowxoupzm wZo moo «EH8 puma Gowmammmsm monopwmmm 03m: mo puma .moamamn Hmwumuma aoum muaamoMIl.H mqmzwme mH mm sew .xxm sumequHm .wex sum xsm .me» 22m wm.mex 22m xm .mazHom may mH mzo mmeszz.zH mzHH mozmmHmmm may mH .o mezsoo aHm ow mmm CT—_—_C>-.C}_ e413 .2 “ [5 I I I L l J 12:30 13:00 13:30 14:00 14:30 15:00 Time HourszMinutes Figure C.4.--Run 5-22, started 10:20. A Nuclei density X 10-5 Nu/ (ml micron) A Nuclei density X10.5 Nu/ (m1 micron) OGrowth rate, Microns/min 0 Growth rate, Microns/min 73 11:00 11:30 12:00 12:30 13:00 13:30 14:00 Time Hours:Minutes Figure C.5.--Run 5-24, started 9:00. 1 l l l l 12:00 12:30 13:00 13:30 14:00 15:00 Time Hours:Minutes Figure C.6.--Run 5—27, started 10:00. ANuclei density X 10-5 Nu/ (m1 micron) ANuclei density X 10-5 Nu/(ml micron) OGrowth rate, Microns/min OGrowth rate, Microns/min .5“- 4 _. O .3 " 25 .2 " .1 _' l l 1 J l 74 12:30 13:00 13:30 14:00 14:30 15:00 15:30 Time Hours:Minutes Figure C.7.--Run 6-3, started 9:15. 11:15 11:30 11:45 12:00 12:15 12:30 12:45 Time Hours:Minutes Figure C.8.--Run 6-5, started 10:07. lkNuclei density><10-5 Nu/(ml micron) ()Growth rate, Microns/min 75 l l l l l 16:00 16:15 16:30 16:45 17:00 17:15 Time Hours:Minutes Figure C.9.--Run 6-8, started 14:25. 17:30 BIBLIOGRAPHY 76 10. ll. BIBLIOGRAPHY Aeschback, S., Attainment of homogeneous suspension in a continuous stirred tank. Chem. Eng. J., V4 N3, Dec. 1972. Anshus, Byron E., On the stability of a well stirred 1 isothermal crystallizer. Chem. Eng. Sci. V28 N2, " Feb. 1973. Baker, C. G. J. and Bergongnou, M..A., Precipitation of sparingly soluble salts: A model of agglomera- tion controlled growth. The Canadian Journal of Chemical Engineering, V52, April 1974. was... i— Bauer, L. G., Contact nucleation of magnesium sul- fate heptahydrate in a continuous MSMPR crystallizer. Unpublished Ph.D. thesis. Ames, Iowa, Library, Iowa State University, 1972. Bendig, L. L., Contact nucleation and small crystal growth. Unpublished M.S. thesis. Ames, Iowa, Library, Iowa State University, 1974. Black, Shuley, Fleming, From lime-soda softening sludges. Recovery of calcium and magnesium values. J. American Water Works Ass., V63 N10, Oct. 1971. Brulhart, P. Process and plant layout considera- tions in the treatment of drinking water. Sulgen Tech. Rev. V52 N70, p. 45. Chabereck, 8, Courtney, R. C. and Martell, A. E., Magnesium Hydroxide dissociation. J. American Chemical Soc., V 74, p. 5057, 1952. Diehl, H. C. Quantitative analysis. Ames, Iowa, Oakland St. Science Press, 1970. Fair, G. M., Geyer, J. C. and D. A. Okum. Water and wastewater engineering. New York, N.Y., John Wiley and Sons, Inc., 1968. Gunn, D. J. and Murthy, M. S. Kinetics and mechan- isms of precipitations, Chemical Engineering Science, V 27, p. 1293, 1972. 77 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 78 Hartline, Ralph A., Computer analysis of cooling water blowdown reduction. Unpublished M.S. thesis, Engineering Library, Michigan State University, East Lansing, Michigan, 1975. Hudson, H. E., Evaluation of plant operating and Jar-Test Data. .I.American Water Works Ass., V65 N6, May 1973. Judkins, J. F. and Wynne, R. H., Crystal-seed condi- tioning of lime-softening sludge. J. American Water Works Ass., V64, May 1972. Kester, Dana and Pytkowicz, R. M., Magnesium sulfate association at 25°C in synthetic sea water. Limnology and oceanography, V13 N4, Oct. 1968. Klein, David H., Homogeneous nucleation of magnesium hydroxide. Talania Rev. Int. Ed. V14 N8, 1967. Loveland, J. W., Conductivity and Oscillometry, in Treatise on Analytical Chem., I.M. Kolthoff and P. J. Elving, eds. P + I, Vol. 4, Interscience Pub. div. John Wiley and Sons, New York, N.Y., 1963. Moore, W. J. Physical Chemistry. Prentice-Hall, Inc., Englewood Cliffs, N.J., 1964. Mullin, J. W. Crystallisation. Butterworths, London, 1972. Nyvlt, J. Evaluation and characteristics of the particle size composition of products made in agi- tated crystallizers. Int. Chem. Eng., V9 N3, July 1969. Nyvlt, J. and Mullin, M. W. Periodic behavior of continuous crystallizers. Chem. Eng. Sci. V25 Nl, Jan. 1970. Randolph, A. D. and Larson, M. A. Theory of Particu- late Processes. Academic Press, New.York, N.Y., 1971. Rossum, John R., Checking the accuracy of water analysis through the use of conductivity. J. Ameri- can Water Works Ass., April 1975. Sawyer, C. N. and McCarty, P. L. Chemistry for Sanitary Engineers. McGraw-Hill Book Co., New York, N.Y., 1967. 25. 26. 27. 28. 29. 30. 31. 32. 79 Schierholz, Paul M., Kinetics of crystallization in a dilute system. Unpublished Ph.D. thesis, Ames, Iowa, Library, Iowa State University, 1974. Schierholz, Paul M. and Stevens, John D. Determina- tion of the kinetics of precipitation in a dilute system. Paper presented to 67th AIChE Annual Meeting, Washington, D.C., December 1-5, 1974. Standard Methods for the examination of water and wastewater. American Public Health Association, Inc., New York, N.Y., 1967. Strickland, Constable, R. J. Kinetics and mechanism of crystallization. Academic Press, London, 1968. Thompson, Mary E., Magnesium in sea wat r: An electrode measurement. Science, V153, aug. 1966. Water, City of Dayton Water Department, 101 West Third Street, Dayton, Ohio 45402. Water supply system, Board of Water and Light, Lansing, Michigan 48895. Wood, W. W., Rapid reaction rates between water and a calcareous clay as observed by specific-ion electrodes. J. Research U.S. Geol. Survey, V1 N2, Mar.—April 1973.