UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Investigation into the factors affecting controlled struvite crystallization at the bench-scale Dastur, Mahazareen Behram 2001

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2002-0056.pdf [ 6.65MB ]
Metadata
JSON: 831-1.0063895.json
JSON-LD: 831-1.0063895-ld.json
RDF/XML (Pretty): 831-1.0063895-rdf.xml
RDF/JSON: 831-1.0063895-rdf.json
Turtle: 831-1.0063895-turtle.txt
N-Triples: 831-1.0063895-rdf-ntriples.txt
Original Record: 831-1.0063895-source.json
Full Text
831-1.0063895-fulltext.txt
Citation
831-1.0063895.ris

Full Text

INVESTIGATION INTO THE FACTORS AFFECTING CONTROLLED STRUVITE CRYSTALLIZATION A T THE B E N C H - S C A L E by M A H A Z A R E E N B E H R A M DASTUR B.E. (Civil Engineering), The University of Bombay, India, 1997 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M A S T E R OF APPLIED SCIENCE in THE F A C U L T Y OF G R A D U A T E STUDIES DEPARTMENT OF CIVIL ENGINEERING We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A December 2001 © Mahazareen Behram Dastur, 2001 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department of C/\/IL £hjG, tA/£eg/Aj<$-The University of British Columbia Vancouver, Canada bate 2>ece*f&eA os, zoo/ DE-6 (2/88) A B S T R A C T ABSTRACT This research was initiated with the purpose of understanding more about the process of struvite crystallization - a novel way of recovering phosphorus as phosphate from wastewater. A new reactor design was tested at the bench-scale, in terms of its ability to provide extra turbulence inside the reactor, in order to enhance the mechanical strength of the growing crystals and by that function, make for a more convenient harvesting operation. Selective withdrawal of hard crystals was possible after minor modifications to the original design, which was eventually adopted at the pilot-scale as well. An effort was made to arrive at some understanding, with respect to the use of the recycle stream as a tool to maintain the strength of feedwater in the reactor at some optimum concentration. Accordingly, three sets of runs, each with a different feedwater concentration (influent phosphate values of 40 mg 'L" 1 and 80 mg 'L" 1 ) and recycle ratio were conducted. Although the hypothesis justifying the use of a recycle stream as a diluent for the incoming phosphate stream seemed logical, the data collected from the three sets could not corroborate it. The restriction in the value of maximum allowable operational pH (pHijm) was identified as the reason for this shortcoming. A n increase in the operational pH beyond pH]jm was shown to be beneficial for P-removal and partly vindicated the logic for the requirement of a recycle stream. However, it was not possible to operate the bench-scale system at pH values exceeding pHii m without plugging the reactor within 24 hours after a run start-up. Limited data from the follow-up pilot-scale reactor established that it was possible to overcome the restriction of pHijm inherent at the bench-scale. The higher degree of turbulence inside the pilot-scale reactor and the increase in crystal loading over time (30 days - 40 days after run start-up), were identified as the factors affecting the limit in the operational pH at the bench-scale. Once the restriction afforded by pHijm was overcome, the recycle stream could function successfully as intended - as a diluent. A recycle stream of 6:1 and an operational pH of 8.2 were identified as the prime requirements to achieve phosphorus removals of 80%- 95% for an incoming phosphate concentration of 60mg'L" 1 - 70 mg ' L" 1 , at i i ABSTRACT pilot-scale. Runs with low influent phosphate concentrations (20 mg ' L" 1 - 30 mg ' L"1) were observed to exhibit signs of crystal growth at a relatively high pH value of 9.0, at bench-scale. This pointed to the possibility that P-recovery through crystallization from such concentrations, may not be economically feasible. The possibility of using struvite solubility criteria as a control parameter for a struvite crystallization experiment was also explored. A conditional solubility curve for struvite, generated from experiments conducted at the University of British Columbia (UBC), was used as a reference point for this purpose. However, a plot of the conditional solubility product values of the bench-scale runs against the curve, incorrectly implied that the conditions inside the reactor were undersaturated. It was hypothesized that this curve may not serve as a universal quantification of struvite solubility criteria for all struvite crystallization reactors, since the underlying reaction is highly dependent on the physical characteristics existing inside the reactor at any given time; this would include the degree of turbulence and the seed / crystal loading. i i i T A B L E OF C O N T E N T S TABLE OF CONTENTS Page Abstract i i Table Of Contents iv List Of Tables ix List Of Figures x Acknowledgements xi i Dedication xiv Chapter 1 INTRODUCTION 1 1.1 Phosphorus In Nature 1 1.1.1 Phosphorus and its status as an essential nutrient 1 1.1.2 A case of the right resource in the wrong place 1 1.2 The Driving Force For Phosphorus Recovery 4 1.2.1 The fertilizer industry and their future 4 1.2.2 Wastewater - an important phosphorus resource 5 1.2.3 Eutrophication and meeting more stringent discharge 5 regulations 1.3 Recovery Versus Removal 5 1.3.1 Forms and processes of phosphorus recovery as 6 phosphates Chapter 2 OBJECTIVES 10 i v T A B L E OF C O N T E N T S Page Chapter 3 B A C K G R O U N D A N D PRELIMINARY LITERATURE REVIEW 12 3.1 The Requirement For The Quantification Of The Solubility 12 Product Value For Struvite 3.1.1 Reported solubility product values for struvite 12 3.1.2 Experiments concerning the solubility product of struvite 16 3.2 Struvite Formation Potential 17 3.2.1 Agreements and disagreements 17 3.2.2 The concept of the conditional solubility product 18 3.2.3 The concept of the supersaturation ratio 21 3.2.4 Elaboration on the conditional solubility product curve 23 presented in Figure 3.1 3.3 Practical Experiences In Struvite Recovery 25 3.3.1 Experimental set-ups and crystallization reactor designs 25 3.3.2 Required supplementations 26 3.3.3 Seeding the crystallization reactor 26 3.3.4 pH control, recommended reaction times and P-removals 27 3.3.5 Ease of harvesting the final product 28 3.3.6 Process modeling 29 3.3.7 Effect of temperature on struvite formation 29 3.3.8 Experiences at the full-scale 29 3.4 The Concept Of The Recycle Ratio As A Control Parameter 30 3.5 The Rationale For The Reactor Design Tested / Used 30 Chapter 4 M E T H O D O L O G Y 32 4.1 Experimental Set-up 32 4.1.1 Constituents of the feedwater / influent 32 4.1.2 The reactor and the injection port 32 v T A B L E OF C O N T E N T S Page 4.1.3 Provision of a clarifier 34 4.1.4 Pump head configurations and requirements 35 4.1.5 Diameter of the tubing used 35 4.2 Experimental Technique 35 4.2.1 Terminology / definitions used in this work 35 4.2.2 Supplementations used 39 4.2.3 Circulation flow rate through the reactor and the degree of 40 generated turbulence 4.2.4 Seeding the crystallization reactor 41 4.2.5 Sampling methodology 41 4.2.6 Harvesting crystals 44 4.2.7 Determining crystal composition 44 Chapter 5 RESULTS A N D DISCUSSION 45 5.1 Testing The Reactor Design 45 5.2 Recommending Possible Modifications For Scale-up 45 5.3 Verified Crystal Composition 46 5.4 Factors Influencing Process Behaviour - General Observations 47 5.4.1 The effect of pH on P-removal 47 5.4.2 The effect of pH on the operational control of 48 crystallization 5.4.3 The effect of pH on the quality of harvested crystals 50 5.4.4 The effect of magnesium feed concentration on the quality 51 of harvested crystals 5.5 The Effect Of The Seeding Technique 51 5.6 Investigating The Influence Of The Recycle Ratio 53 5.6.1 The resulting PO4-P concentration inside the reactor 56 vi T A B L E OF C O N T E N T S Page 5.6.2 Indications given by the limited data from the pilot-scale 57 reactor 5.6.3 Possible causes for pHijm being a restriction at the 60 bench-scale operation 5.6.4 Effect of the recycle ratio on P-recovery 63 5.7 Investigating The Response Of The System To Lower P influent 64 Concentrations 5.8 Investigating The Feasibility Of Applying Solubility Criteria 67 As A Process Control Parameter 5.8.1 Conditional solubility product values 67 5.8.2 Reasons for runs Q and R differing from the general trend 70 5.8.3 The effect of temperature on struvite solubility 71 5.8.4 A n elaboration on the discrepancies between graphical 73 presentation of the data and actual observations 5.8.5 pH versus P-removal and pP s-mi 77 5.8.6 The utility of the pP s versus pH plot 77 Chapter 6 S U M M A R Y A N D CONCLUSIONS 82 6.1 The Broader Picture 82 6.2 Conclusions 84 Chapter 7 RECOMMENDATIONS 87 REFERENCES 91 APPENDIX A CALCULATIONS FOR R E C O M M E N D E D REACTION 100 TIME IN THE CRYSTALLIZATION V E S S E L T A B L E OF C O N T E N T S Page APPENDIX B CALCULATIONS FOR UPFLOW VELOCITIES IN THE 101 REACTOR LIMBS; CORRESPONDING TO A CONSTANT FLOW R A T E OF 400 mL • min 1 APPENDIX C CALCULATIONS FOR FLUID R E Y N O L D S N U M B E R S IN 102 REACTOR LIMBS; CORRESPONDING TO A CONSTANT FLOW R A T E OF 400 mL • min"1 A N D A N A M B I E N T T E M P E R A T U R E OF 20 ± 5 °C APPENDIX D SEEDING TECHNIQUE USED FOR THE 103 CRYSTALLIZATION REACTOR: SIZE A N D QUANTITY APPENDIX E INSTRUMENT OPERATIONAL P A R A M E T E R S 104 APPENDIX F SIZE DISTRIBUTIONS B Y WEIGHT OF THE SIEVED 106 H A R V E S T E D CRYSTALS APPENDIX G VERIFIED C R Y S T A L COMPOSITION OF THE 110 H A R V E S T E D C R Y S T A L S APPENDIX H D A I L Y RECORD FOR A L L RUNS 114 APPENDIX I ATTEMPTS TO M E A S U R E THE V E L O C I T Y OF THE 135 REACTION APPENDIX J C A L C U L A T E D IONIC STRENGTH FOR A L L RUNS 139 APPENDIX K ATTEMPTS TO PREDICT THE SEED SIZE B A S E D O N 141 STOKES' L A W / THE THEORY OF DISCRETE PARTICLE SETTLING L I S T OF T A B L E S LIST OF TABLES Page Table 3.1 Reported experimental values for the solubility product (K s p ) of 13 struvite (at 25 °C) Table 3.2 Methodologies and results of the struvite solubility studies 16 completed at U B C (at 25 °C) Table 3.3 Guidelines for pH control for struvite crystallization (collected 28 from the literature) Table 3.4 The crystallizer reactor - dimensions 31 Table 4.1 Calculated upflow velocities and fluid Reynolds numbers in the 40 reactor limbs corresponding to a constant flow rate of 400 mL • min"1 Table 5.1 Effects of the seeding technique at run start-up 52 Table 5.2 Investigating the influence of the recycle ratio 54 Table 5.3 Effect of pH on the PO4-P concentration inside the reactor 56 Table 5.4 Comparing the influence of the recycle ratio between the 58 bench-scale and pilot-scale reactors Table 5.5 Calculated Reynolds numbers for the bench-scale and pilot-scale 60 reactors (25 °C) Table 5.6 Effect of the recycle ratio on P-removal at the bench-scale 63 Table 5.7 Investigating the response of the system to lower PO4-P 65 concentrations Table 5.8 Size distributions by weight of crystals harvested from runs with 72 constant operational parameters at 15 °C and 25 °C ix L I S T OF F I G U R E S LIST OF FIGURES Page Figure 1.1 Nutrient distribution in an agrarian economy 2 Figure 1.2 Nutrient distribution in a simple urban economy 2 Figure 1.3 Nutrient distribution in a complex urban economy 3 Figure 3.1 Equilibrium conditional solubility product curve for struvite 20 resulting from experiments at U B C in August 2000 (25 °C) Figure 3.2 The concept of the metastable region 22 Figure 3.3 Equilibrium conditional solubility product curves for struvite from 23 experiments at U B C and an independent study Figure 3.4 The effect of the ionic strength on the positioning of the conditional 24 solubility product curve for struvite Figure 3.5 Design features of the crystallization reactor (not to scale) 31 Figure 4.1 Experimental set-up (not to scale) 33 Figure 4.2 A simple illustration of the injection port (front view, not to scale) 34 Figure 4.3 Pictorial presentation of a "good" crystal 38 Figure 4.4 Pictorial presentation of an "unacceptable" crystal 38 Figure 5.1 General response of a system to a constant pH 47 Figure 5.2 Effect of the pH on the PO4-P concentration 48 Figure 5.3 The management of pP s inside the pilot-scale reactor by the recycle 59 stream Figure 5.4 pP s versus pH plot - Set RCY1 68 Figure 5.5 pP s versus pH plot - Set RCY2 68 Figure 5.6 pP s versus pH plot - Set RCY3 69 Figure 5.7 pP s versus pH plot - A l l runs (20 °C - 25 °C) 70 Figure 5.8 The effect of temperature on struvite solubility 72 x L I S T OF F I G U R E S Page Figure 5.9 The relationship between pP s values from different runs 73 ( 2 0 ° C - 2 5 ° C ) Figure 5.10 Precipitation diagram for the formation of CaHP04 from the 75 CaCl2-Na 3 P04-H 2 0 system Figure 5.11 The relationship between pH, P-removal and pP s-mi (20 °C - 25 °C) 77 Figure 5.12 Tolerance window for pP s . m i values 79 xi A C K N O W L E D G E M E N T S ACKNOWLEDGEMENTS Acknowledgement and thanks are due to the following institutions and people, without whom, this research would not have been possible: • The University of British Columbia, for giving me the opportunity to continue my studies and gaining a true learning experience, • BC Hydro, for their generous funding of this research project, • Dr. D. S. Mavinic, my supervisor, for his exceptionally encouraging words throughout the course of my work, • Frederic Koch, the manager of the U B C Pilot Plant, for introducing me to this wonderfully novel and challenging project, and for helping me with my preliminary experimental set-up, • Dr. Ping Liao, "the chemist among us engineers", for being one of the greatest sounding boards of all time, • Dr. Noburu Yonemitsu, for his critical inputs concerning the design of the crystallization reactor tested in this work, • Paula Parkinson and Susan Harper, for their expert technical advice, and for teaching me the nuts and bolts of the analytical instrumentation, • Dr. K . J. Hall, the second reader of this thesis, for his important and much appreciated insights into this work, • A l i Adnan, the second student on this project, for graciously allowing me to utilize a part of the data from his work, and for all those brainstorming sessions, • Ahren Britton, the "quasi-second" student on this project, for always being so humorous, handy and helpful, • Akiko Yabuno, for her more than helpful translations of some of the literature, from Japanese to English, and xii A C K N O W L E D G E M E N T S • M y family and friends, for their positive outlook and their intense support, without which, I would not be here. To Mahaffeen, Marzban, Khurshid, Jamshed and Mehru xiv INTRODUCTION C H A P T E R 1 I N T R O D U C T I O N 1.1 Phosphorus In Nature 1.1.1 Phosphorus and its status as an essential nutrient Phosphorus is the eleventh most common element on earth, indispensable to all living organisms (1). In nature, phosphorus always occurs combined with oxygen and other elements, forming phosphates. It is a scant natural resource, for which there is intense competition between life forms, both on land and in the aquatic environment. It is believed that those organisms, which are most successful in foraging, storing and recycling this element, have a natural advantage over those life forms, which are not as successful in doing so. In earlier times, man used to employ a range of tactics to maintain the phosphate nutrient status of the soil. Some of these tactics date back to the Iron Age and have much in common with some of the waste disposal methods practiced today; e.g. sludge and manure spreading. For instance, contracts covering the lease of agricultural lands of English monasteries, had specific "dung clauses" which would provide for the tenant farmer to graze his sheep on the land, and to retain the meat and the wool for himself, but not the dung, which had to stay where it fell on the field. Thus, the nutrient status of agricultural soils was closely guarded. Instances have also been recorded where the dearth of such measures, has been responsible for the shifting of entire dependent populations to new lands (2). 1.1.2 A case of the right resource in the wrong place The growth of cities and relocation of centers of population remote from areas of agricultural production caused a break in the closed phosphorus loop. This led to a gradual deterioration 1 I N T R O D U C T I O N in the system of phosphorus reapplication to / retention in agricultural soil. As cities grew, sewage systems were built and later, sewage treatment works were added to provide some acceptable degree of treatment before discharging to receiving waters. The net result was (and in some ways, still is) a very competent system that conducts nutrients (phosphorus and nitrogen) away from the land and into the rivers and the sea. Figures 1.1 to 1.3 depict the changes in nutrient distribution as a function of the complexity of the economy. p + N Figure 1.1: Nutrient distribution in an agrarian economy (3). Figure 1.2: Nutrient distribution in a simple urban economy (3) 2 I N T R O D U C T I O N Figure 1.3: Nutrient distribution in a complex urban economy (3) Eutrophication It is estimated that it was close to a hundred years before man realized the impact of nutrient removal from land, and the resulting problems of excessive nutrient loading, causing what is now referred to as eutrophication, in the aquatic environment (2). In such environments, phosphorus is generally the limiting nutrient and therefore, its increased human-induced availability, leads to marked increases in the growth of algae. Algal blooms cause taste and odour problems in the water supply and can exert relatively high dissolved oxygen demands, which could lead to fish kills. Increased aquatic plant growth can also reduce accessibility for recreation (4). Oli gotrophication By the end of the 1980s, ways and means of reducing nutrient inputs from wastewater treatment plants (WTPs) into receiving water bodies had been found. While this resulted in marked improvements in receiving water bodies, there was a general failure in foreseeing its effects on other aspects of ecosystems, namely declining fish populations. Oligotrophic systems are described as "nutrient-deficient and of low biogenic production" (5). 3 I N T R O D U C T I O N Oligotrophication is the antithesis of eutrophication and has been linked directly to man-made imbalanced nutrient inputs in aquatic systems. The "over-efficient" removal of nutrients from lakes and rivers, through sewage treatment plants, has been earmarked as one of the causes of oligotrophication. It has been linked to some rather alarming declines in freshwater fisheries in large North American and European lakes, at times with severe economic results (5). 1.2 The Driving Force For Phosphorus Recovery As it stands today, logistical problems of increased demand for food from the growing population, and therefore increasing demand of replenishment of nutrients to land, has led to widespread use of man-made fertilizers (2, 5). In an effort to tip the scales from undesirable oligotrophic aquatic environments to optimal mesotrophic aquatic environments, serious thought is being given to replenishing nutrient deficient lakes with fertilizer. Indeed, fertilization of some lakes is already a reality in some parts of North America (5). Phosphorus lacks a gas phase. Unlike nitrogen and carbon, which can be replenished from atmospheric sources, the only natural source of phosphorus is that produced as a result of weathering of phosphate-containing rocks (2). Clearly, with the present non-sustainable phosphorus-related practices of the post-industrialized world, the demand for this nutrient far outweighs its supply. 1.2.1 The fertilizer industry and their future The phosphate fertilizer industry's main raw material comes from phosphate rock, the collective name given to natural calcium phosphates of various forms. The known reserves of phosphate rock are limited. It is estimated that the commercially exploitable resource base of phosphate rock could, at best, last for a little over a hundred years, and could be depleted in 4 I N T R O D U C T I O N as little as fifty years. Faced with the possibility of extinction of their crucial raw ingredient, the industry is keen to explore new avenues for production of unconventional sources of this material (2). 1.2.2 Wastewater - an important phosphorus resource Five million tonnes of rock phosphorus as P2O5 were consumed in North America in 1995 - 1996 (6). The expected phosphorus load factor from individual residences is estimated to be 3.28 g ' capita"1 'day"1 or 1.2 kg ' capita"1 " year"1 (7). This corresponds to a little more than 20% of phosphorus used as fertilizer in North America alone (8, 9). In Asia, the E U and Latin America, where the consumption of P2O5 exceeds its production, the corresponding number is higher (6). 1.2.3 Eutrophication and meeting more stringent discharge regulations The European Commission's Urban Wastewater Treatment Directive has been imposing increasingly stringent regulations on nutrient discharge to water bodies in the region since 1991. Phosphorus may be removed from wastewater by common salt precipitation methods and / or biological phosphorus removal. The former method generates large volumes of sludge, involves a fairly huge cost of chemicals and binds up the phosphorus in the sludge in such a manner, so as to make it impossible to retrieve it. The latter method gives rise to a wastewater, which will have significantly lower P concentrations, together with a phosphate-enriched sludge, which may or may not have a significant agricultural value. Issues concerning pathogenic and / or heavy metal contamination, odour and maximum permissible sludge loading to agricultural land lessens the attractiveness of biological phosphorus removal as the sole phosphorus removal technology (10, 11). 1.3 Recovery Versus Removal Recovery of phosphorus as opposed to its removal, could offer many potential benefits. Recovery technologies result in a significant decrease in sludge production and associated costs of disposal. The opportunity of recovering costs from sale of the phosphorus as 5 I N T R O D U C T I O N fertilizer adds to the logic of propagating the use of such technologies in the field of wastewater treatment (12). The fertilizer industry also perceives benefits from applying recovery technologies, since the availability of recycled raw material could help address some of its resource depletion issues. In fact, phosphates recovered from wastewater have been shown to be purer (with far fewer heavy metal impurities) than some natural phosphate rock (2, 13, 14). Furthermore, the disposal of manure is beginning to pose a crisis in many European countries. In comparison to municipal sewage, sewage from livestock operations is a more promising medium for phosphorus recovery; it is rich in nutrients and very concentrated (15). 1.3.1 Forms and processes ofphosphorus recovery as phosphates Phosphorus has been recovered from livestock manure (16, 17) and / or municipal wastewaters (13, 14, 18 - 22), with varying degrees of success, in the following crystalline forms. • Magnesium ammonium phosphate hexahydrate (also known as M A P or Mg-struvite or struvite), • Magnesium potassium phosphate (also known as K-struvite) and • Calcium phosphate (also known as hydroxyapatite) Crystallization is loosely defined as the controlled precipitation of selected substances in crystalline form from solution (23). Magnesium ammonium phosphate hexahydrate Magnesium ammonium phosphate hexahydrate (hereafter referred to as "struvite") is of primary interest to this project. Struvite formation under alkaline conditions is brought about by a reaction among magnesium, ammonium and phosphate ions and six water molecules. 6 I N T R O D U C T I O N base M g + 2 + N H 4 + + P0 4 " 3 + 6 H 2 0 • MgNH 4 P04.6H 2 0 (1) Struvite belongs to a range of salts called metal ammonium phosphates. It is an excellent plant fertilizer, due to the slow-release property which it exhibits, in response to plant nutrient demand. When the nutritional status of the plant becomes deficient, the plant generates citric acid, which leads to the dissolution and subsequent uptake of struvite. It is insoluble in water at neutral and alkaline pH values. In theory, its application to land should not adversely influence the nutrient level of the water table (24). There have been no published studies of struvite as a fertilizer for oligotrophic lakes. However, aspects of phosphorus release from a commercially manufactured slow-release fertilizer pellet, have been studied under simulated stream conditions (25). Despite such attractive agronomic properties, struvite is not widely manufactured by the fertilizer industry, the main reason appearing to be its high cost of production from raw chemicals - magnesium oxide, phosphoric acid and ammonia (16). Formation of struvite facilitates a simultaneous recovery of ammonia and phosphates from wastewaters, the former always being present in a higher ratio than the stoichiometric requirement. Untreated domestic wastewaters may contain 4 mg ' L" 1 - 15 mg ' L" 1 as total phosphorus, of which 3 mg " L" 1 - 10 mg ' L" 1 may exist as soluble orthophosphate (26). This relatively low concentration of phosphate in a large volume of sewage makes nutrient recovery in a useable form a significant challenge. However, the supernatant from an anaerobic digester of an enhanced biological nutrient removal (EBNR) WTP proves to be an excellent source of ammonia and phosphate for struvite production. Typical phosphate influent concentrations of 10 mg ' L" 1 can easily reach a concentration of 60 mg ' L" 1 , as a consequence of release during the anaerobic phase of the treatment. The bulk of the phosphate is thus concentrated in a relatively smaller volume or sidestream, which would 7 I N T R O D U C T I O N theoretically translate to smaller reactor volumes necessary for crystallization (21, 27). This information assumes importance in view of the overall economics of a crystallization process, which could go far in determining its viability. In sum, the "why" (reasons for phosphate recovery or P-recovery) and "where" (end use) of struvite as a P-recovery product have been sufficiently established. Required supplementations Supplementation of magnesium with an external source would be necessary, as fairly low levels of this cation would exist in the supernatant to fulfill stoichiometric requirements of the maximum possible struvite formation. Since struvite formation is greatly facilitated under alkaline conditions (pH = 8 - 10), addition of a base would also become essential (21, 26, 27). Processes for phosphate recovery To date, seven processes for crystalline P-recovery exist. The application of some of these processes has not advanced beyond the pilot-scale and in some cases, beyond the bench-scale. As of July 1999, three processes were reported as "fully operational", namely the D H V Crystallactor™ Pelletiser Process (The Netherlands), the Unitika Phosnix Process (Japan) and the Kurita Process (Japan) (15). Selecting a process as a benchmark for the U B C Phosphate Recovery Project Some intrinsic differences exist among these processes1. The D H V Crystallactor™ Pelletiser makes use of sand as a seeding material. The phosphate crystallizes on the surface of the sand, thus leading to a gradual increase in the size of the pellets. Once the pellets reach a desired size, they are removed and replaced with a smaller ' A general outline is given here, in order to provide a short background. More details may be found in Reference 10. Certain aspects (the need for seeding the reactor, the manner of harvesting / withdrawal of the struvite product from the reactor and so on), have been expanded on in later sections of this report. 8 INTRODUCTION diameter of seed / sand grains (10). The sand bed is fluidized by virtue of the upflow of the wastewater through the reactor. The Unitika Phosnix Process makes use of previously withdrawn / harvested struvite material, as seed grains. There is no reported need for replenishing the reactor with new seed material, once crystal withdrawal has been performed. A blower / aerator forces air through the crystallization reactor in the direction of the upflow of the wastewater. While the reason for / advantages of the use of air are not elaborated on, the process is reported to work satisfactorily in terms of the achieved P-recovery (10, 15). The Kurita Process employs phosphate rock as seed grains. Wastewater is fed into the base of the column. Unlike the Unitika Phosnix Process, this process does not employ air agitation through the reactor (10). The Kurita Process was chosen as a benchmark, since it was deemed to be the easiest in terms of operation and design (Frederic Koch, Research Associate, U B C Phosphate Recovery Project, Department of Civi l Engineering, U B C , pers. comm.). Using this process as a yardstick, does away with frequent and (perhaps wearisome) replacement of seed material. The possibility of using air agitation through the fluidized bed reactor was explored early on in the U B C Phosphate Recovery Project. However, its use was discounted, due to frequent and persistent struvite cementation problems associated with the air blower (Frederic Koch, pers. comm.). 9 O B J E C T I V E S CHAPTER 2 OBJECTIVES This project proposed to build on available knowledge from the pertinent process (the Kurita Process) and apply this summed knowledge to a new struvite crystallization reactor design (hereafter referred to as "the reactor"); this had come about as a result of extensive experimental studies conducted since August 1999 (Frederic Koch, pers. comm.). As mentioned in Section 1.3.1, the Kurita Process was considered to be the easiest, in terms of operation and design and hence, it was chosen as a benchmark (Frederic Koch, pers. comm.). No related work has been formally published at any time by the proprietors of the Kurita Process. As a result, the main knowledge base comprised of personal communications between some workers involved with the management of the process at a WTP in Japan and Frederic Koch at U B C . The stated aims of this project included • Testing the reactor design (particularly with respect to how well its behaviour conformed to general theory), • Recommending possible modifications if need be for eventual scale-ups, • Investigating the influence of recycle ratio on process control, • Investigating the behaviour of the system for different PO4-P concentrations, • Optimizing the operating conditions necessary for achieving the best possible yields for struvite recovery with the reactor and • Investigating the feasibility of applying solubility criteria as a process control parameter. 10 O B J E C T I V E S Some of the more elaborate points have been expanded upon in Chapter 3 of this report. All testing was carried out at bench-scale. 11 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W CHAPTER 3 BACKGROUND AND PRELIMINARY LITERATURE REVIEW 3.1 The Requirement For The Quantification Of The Solubility Product Value For Struvite Environmental engineering chemistry frequently employs Le Chatelier's principle to its advantage. External stresses are applied to systems to engineer a shift in equilibria in the desired direction. A l l precipitation reactions are an example of manipulating equilibria. In these cases, the knowledge of the solubility product becomes important. Notable examples include the softening of hard waters by lime-soda ash treatment and the removal of metal ions from industrial wastes. 3.1.1 Reported Solubility Product Values For Struvite Extensive studies involving calculation of the solubility product (K s p ) value for struvite have been conducted in the past by various authors. Such an experiment involves formation and/or dissolution of pure precipitate in deionized or distilled water and under rigidly controlled conditions; such conditions include a constant temperature, carefully adjusted ionic strength and a constant degree of mixing energy imparted to the solution. Formation of pure precipitates is carried out using analytical grade reagents. Dissolution may be carried out using precipitate created during a formation experiment or naturally formed precipitate obtained from the field (WTP in this case). Every effort is made to exclude sources of any chemical species extraneous to the reactions in question (28 - 36). A simultaneous solution of all the relevant mass and charge balance relationships, along with the appropriate proton condition, gives the concentration of each species at equilibrium (37). 12 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W As Table 3.1 shows, the values reported by various investigators differ by as many as up to five orders of magnitude. Similar variations in magnitude have been recorded for several precipitation reactions of interest in water and wastewater engineering; for example gibbsite or Al(OH)3, metal sulphides and metal carbonates (32, 37). Table 3.1: Reported experimental values for the solubility product (K s p ) of struvite (at 25 °C) p K s p K s p Reference 9.41 3.89 X 10"'u 28 9.94 1.15 X 10"lu 29 11.84 1.44 X 10"12 30 12.60 2.51 X 10"1J 32 13.00 1.00 X 10"13 33 13.12 7.58 X 10"14 34 13.15 7.08 X 10"'4 31,35 13.26 5.50 X 10"14 36 14.10 7.94 X 10"15 (a) 13.32 - 14.29 4.79 X10"'4 -5.12X10"15 (b) Frederic Koch, pers. comm. Ping Liao, Research Associate, UBC Phosphate Recovery Project, Department of Chemical and Biological Engineering, UBC, pers. comm. Reasons for poor agreement between reported solubility product values for struvite Such poor agreement between values is thought to be caused by the following. • Uncertainty over whether real / true equilibrium has been attained: There is no agreement among authors as to how much time is needed for "true" equilibrium conditions to be reached in dissolution experiments for K s p determination. One study 13 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W suggests that equilibrium is reached 10 days after the start of the dissolution experiment under quiescent conditions (29). Yet another suggests 38 days as the correct time under similar quiescent conditions (38), while yet another advocates a time limit of 1 day, with "continuous and constant" mixing / stirring (34). • Lack of availability of accurate equilibrium constants for some of the reactions of interest: It has been noted that not all values of dissociation constants may be as precise as one would imagine them to be, especially at temperatures other than 25 °C (29, 31, 32, 38). Formation / stability constants of complex ions have also been shown to be of widely varying accuracy (32, 39). • Assumptions made in the degrees of chemical speciation: There is considerable uncertainty over the actual reactions that occur in the solution (31, 34, 35, 37). One of the latest works on struvite solubility suggests that the K s p value derived, considering three magnesium phosphate complexes, is more accurate than that derived considering one complex only (36). • The activity / stability of the solid phase being used in the dissolution study: Most precipitates undergo a phase transformation from an amorphous phase to a crystalline phase, after precipitating from solution. The term "amorphous" refers to a solid that does not have a well-organized crystalline structure and is almost always more soluble than the corresponding crystalline solid (32, 39). While no such disparity in phase has been reported specifically for struvite, not all the struvite solubility studies mentioned here provide confirmation about the purely crystalline phase of the struvite solid used in dissolution experiments (30, 31). A more thorough percent by weight analysis carried out in one study, showed the percent of magnesium in the precipitate from real liquors to vary from 60.7% to 107.2% from the theoretical, the corresponding values for phosphorus being 82.1% and 135.7%. For precipitates from synthetic liquors (reagent grade chemicals), the scatter was less pronounced - but it did exist. The percent of magnesium in the precipitate varied from 75% to 94.5% of 14 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W the theoretical and the corresponding values for phosphorus were 93.2% and 96.4% (40). There are also suggestions that the nature of the solid phase that precipitates first is governed by kinetic factors (32, 39). The degree of hydration of a particular solid can also have a pronounced effect on the rates of precipitation and dissolution (39). The Dana Mineral Classification System for hydrated phosphates lists two particular hydrated phosphates with different degrees of hydration, one of them being struvite and the other dittmarite or MgNH 4 P04 .H 2 0 (41). • Lack of a standard method for Redetermination: None of the studies referred to so far have used the same method or even a remotely similar laboratory set-up for such an experiment. While each study does define the methods, materials and parameters in use, the variations between them are striking. In fact, this lack of standardization in method has been referred to as one of the possible reasons for the disparity in K s p values for precipitation / dissolution experiments (42, 43). Since struvite is a major constituent of urinary stones, some of the K s p studies on hand were designed to cater to / mimic conditions of urinary calculi formation (29, 34, 44). • Thermodynamic considerations or the lack thereof: Older research on the subject (not referred to in Table 3.1) was thermodynamically irrelevant, since corrections for ionic strength were not included in the calculations. Other researchers assumed negligible ionic strength (32, 35, 39, 42). More recent studies included corrections for ion activities resulting from ionic strength effects, using either the Guntelberg or Davies approximation of the Debye-Huckel equation. The results have been deemed as "more appropriate" values. 15 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W The value of the ionic strength was fixed as 0.1 M in these studies and the solutions of equilibrium equations were performed using MINTEQA2, an aquatic chemistry equilibrium model (31, 38). 3.1.2 Experiments concerning the solubility product of struvite To date, two struvite solubility (dissolution) studies have been completed at U B C . Table 3.2 gives details of the differences in methodology and results of these two studies. Note that the solids used for dissolution were formed during separate and previous experiments with one of the older crystallization reactor designs. (Ping Liao, pers. comm.). Table 3.2: Methodologies and results of the struvite solubility studies completed at U B C (at 25 °C) Period of study Mode of study Dissolution technique p K s p August 2000" Dissolution only Distilled water solution 13.32 containing struvite precipitate stirred for 1 hour before sampling was commenced November 2000 Dissolution only Distilled water solution 14.29 containing struvite precipitate not stirred. Sampling was commenced after 4 days " The results from the August 2000 batch of experiments have been used throughout this work. There were indications that the results from the November 2000 batch depicted a system that had not reached equilibrium and hence the prerequisite of the exercise had not been truly met. 16 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 3.2 Struvite Formation Potential Struvite precipitation is a recognized problem in anaerobic sludge digesters, where it precipitates in digester supernatant recycle lines, especially at the elbows and the suction side of pumps. It is postulated that struvite precipitation occurs at these locations because they are regions of reduced pressure, where carbon dioxide is released from solution. Such decarbonation causes a rise in pH of the digester supernatant to values sufficient to favour precipitation. The extent of the precipitation depends on the concentrations of its constituent ions, magnesium (Mg + 2 ), ammonium ( N H / ) and soluble orthophosphate (depicted as PO4-P throughout this work). Struvite deposits are hard and once formed, are extremely difficult to remove. They cause damage to pumping equipment and almost totally block sludge pipes, resulting in costly maintenance and repairs, as well as disruptions to the operations of the WTP (28, 30, 32, 36, 39, 40, 42). Consequently, a large amount of research has been directed at trying to merely define conditions for struvite formation and pinpoint factors that may be manipulated, in order to alleviate or i f possible, to prevent the problem. 3.2.1 Agreements and disagreements Typical crippling blockages have also been observed in pumps used to transfer liquid from anaerobic animal waste treatment lagoons. However, the carbon dioxide evolution-pH rise theory was regarded questionable, since similar amounts of deposition were also observed in pipes where pressure drops would have been minimal, and also because of the lack of visible crystals at the bottom of the sludge. The authors concluded that the observed clogging, at least in that particular case, was due to the favoured precipitation of struvite on metallic components in the pumping-plumbing systems and showed that replacements of sewer pipes, vent pipes and pumps with plastic components considerably alleviated the problem (30). A different study on preferential struvite accumulation on selected construction materials supports this observation (45). 17 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 3.2.2 The concept of the conditional solubility product In spite of these disagreements, workers looking for a practical solution to the problem, recognize that some sort of simple quantification is required to ascertain the potential for the formation of such blockages. The conditional solubility product is thought to be able to provide just this sort of quantification. The involvement of large numbers of chemical species (noted as twelve in the case of struvite) taking part in / influencing the end reaction, leads to very complicated calculations as far as the K s p is concerned (31, 38). { M g + 2 } { N H 4 + } { P 0 4 - 3 } = K s p (2) Furthermore, equation (2) implies that the solubility product is only accurate for a single pH value, and it is common knowledge that the solubility of struvite is strongly influenced by the pH (40). The concept of conditional constants was introduced to overcome these factors. A conditional constant is defined as an equilibrium constant that holds only under a given set of experimental conditions (39). Such a constant incorporates the effect of the pH and does away with the limitation inherent in equation (2). The conditional constant gives a relationship between elements that are of direct interest. The case of struvite demonstrates a situation where more than one dissolving species is strongly governed by the solution pH - the ammonium ion N H 4 + and the phosphate ion P0 4 " 3 . Since an increase in pH would increase the soluble P0 4 " 3 ion concentration and decrease the soluble N H 4 + ion concentration, it follows that there must be some value of pH where the solubility of struvite or its concentration product - [Mg ] t o t a i • [NH 4 -N] t o ta i • [P0 4 -P] t o ta i - must be a minimum. 18 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W Given the solubility product equilibrium where M n A m ( s ) ^ mA- n + n M + m .' (3) And the solubility product being K s p = { M + m } n {A" n } m (4) The conditional solubility product would be defined as P s = Mtotal" Atotal"1 = K s p (5) OCM A A YM YA where { } depict molar concentrations corrected for activity, Mtotai, Atotai = total concentration of the anion and cation of the salt respectively, irrespective of the form they may be present in, C O M , aA = ionization fractions in solution for the cation and anion respectively; i f [ ] depicts the molar concentration, then a M = [M n] / M t o t a i n and ctA = [A m ] / A t o t a i m , YM, YA ~ activity coefficients in solution for the cation and anion respectively and P s = conditional solubility product. Applying this criteria to struvite gives the relationship 19 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W P s = [Mg + z ] t o t a l . [NH4-N] t0tal • [P04-P] totai = Ksp (6) a M g + 2 a N H 4 + aP04"3 y M g + 2 y N H 4 + ypo 4" 3 In theory, P s will be at a minimum when the product a M g + 2 a N H 4 + aP04"3 is at a maximum. Older aquatic chemistry textbooks provide a curve for pP s versus pH, with minimum solubility being shown to occur at a pH of 10.7 (32, 40), where pP s is the negative logarithm of P s . However, over time, due to the modifications and differences in the approach to finding the solubility product for struvite (Section 3.1.1), the pH for minimum solubility is thought to range from "approximately 9.0" (31) to 10.3 (38). Previous work at U B C suggests a value of 10.4 (Ping Liao, pers. comm.). Figure 3.1 shows the equilibrium conditional solubility product curve (pP s- e q versus pH) resulting from these experiments. 10.0 -I , 1 , , , 1 5.0 6.0 7.0 8.0 9.0 10.0 11.0 pH Figure 3.1: Equilibrium conditional solubility product curve for struvite resulting from experiments at U B C in August 2000 (25 °C) 20 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 3.2.3 The concept of the supersaturation ratio The definition in terms of aquatic chemistry An increase in the concentration of any one of the constituents of struvite would increase the conditional solubility product and hence, the potential for precipitation. In order to quantify this potential, a term called supersaturation ratio (SSR) is used (32, 39), where SSR= _Ps_ (7) Ps-eq And Ps-eq = conditional solubility product at equilibrium. Theoretically, SSR > 1.0 would imply that precipitation is possible and that the system is supersaturated, SSR =1.0 would imply that the system is at equilibrium and SSR < 1.0 would imply that precipitation is not possible and that the system is undersaturated. The effect of a particular pH on the solution chemistry would be taken into account by reading the corresponding values for P s and P s - e q off the graph in Figure 3.1 for that pH. The concept of the SSR as an indicator for struvite formation potential has been corroborated against available data from case histories of WTPs (32, 39, 40) and animal waste effluent lagoons (30, 42) plagued by clogging due to struvite formation. The definition in terms of crystallization Textbooks and journal papers on crystallization (without reference to any particular salt / precipitate) offer some refinements to the general idea of the SSR. They submit that crystallization comprises of two consecutive stages - nucleation and 21 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W growth. Nucleation is believed to be controlled by the solubility chemistry, while growth rate is believed to be limited by low turbulence or low mixing energy in the crystallization reactor (46, 47). Figure 3.2 shows the solubility of a crystal as a function of pH, where solubility decreases with an increase in pH up to a certain value (pH of minimum solubility). Figure 3.2: The concept of the metastable region The region between the curve A B and CD is the metastable zone, which represents that limited area where the rate of nucleation is very small compared to the rate of growth and is generally regarded as a region of controlled crystallization. The area above the curve CD may be classified as the region where the rate of nucleation greatly exceeds the rate of growth and is generally referred to as the region of spontaneous nucleation (48). It therefore follows that the desired process control can be achieved only in the metastable 22 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W region and consequently, the efforts of this work were directed towards defining the boundaries of that region. It also follows that there must be some value of the SSR which represents the boundaries of the metastable region. It is important to note here that no previous investigations have considered defining crystallization process control, in terms of the metastable zone boundaries for full-scale P-recovery applications from wastewater. The idea has been quantified to a limited extent, with only one study conducted at bench-scale (46). 3.2.4 Elaboration on the conditional solubility product curve presented in Figure 3.1 Once the solubility product curve in Figure 3.1 was plotted, there was naturally some effort made to ascertain the extent of its agreement / disagreement with the solitary conditional solubility product curve available in the literature (46). Figure 3.3 shows both curves on a common plot. For the sake of simplicity, the latter curve has been referred to as "Ohlinger's curve", after the author of the study. 0.0 2.0 4.0 -O -CO %, 6.0-8.0 A 10.0 A 4.0 6.0 8.0 pH 10.0 12.0 14.0 Ohlinger's curve Conditonal solubility product curve from UBC experiments Figure 3.3: Equilibrium conditional solubility product curves for struvite from experiments at U B C and an independent study 23 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W While there appears to be remarkable agreement between both curves, a very important difference exists between them. Within the pH range of interest (7 - 9), Ohlinger's curve represents struvite solubility at an ionic strength of 0.1 M (46). The corresponding value for the U B C curve is 0.01 M (Ping Liao, pers. comm.). This translates to a variation in ionic strength of about 10 times between the two curves. This difference in values becomes very significant in light of the fact that the ionic strength plays a deciding role in the solubility of any precipitation-crystallization reaction. An increase in ionic strength would lead to a resultant decrease in the effective concentration of the component ions of struvite (32, 37, 39, 46). This would result in an increase in struvite solubility and therefore less crystallization out of solution. A n increase in the ionic strength (I) would cause the solubility curve to shift in the upward direction, as shown in Figure 3.4 (32). pPs -eq I, pH Figure 3.4: The effect of the ionic strength on the positioning of the conditional solubility product curve for struvite 24 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W Figure 3.3 implies that the calculated variation in the ionic strength did not have any bearing on the solubility of struvite, which was certainly quite unusual. Adequate replicate tests have been carried out to verify the reproducibility of the experimental data pertaining to the U B C curve and there is a very good agreement between replicate sets of data (Ping Liao, pers. comm.). Since the possibility of a mistake in the experimental data was disqualified, the possible explanations for such an anomaly could have been the difference in the activity of the crystals used in the different dissolution studies and / or the mode of crystal dissolution (Section 3.1.1). The U B C curve was generated using a distilled water solution containing struvite precipitate, which was stirred vigorously for 1 hour before sampling was commenced. Ohlinger's curve was generated using a quiescent distilled water solution containing struvite precipitate, sampled after 38 days (38). Bearing that in mind, it was decided to use the U B C curve as a background for investigating the feasibility of using the solubility criteria ratio as a process control parameter. 3.3 Practical Experiences In Struvite Recovery As mentioned earlier, a fair amount of research has been directed at trying to merely define conditions for struvite formation. Sections 3.3.1 to 3.3.6 present published literature on the reported experiences with struvite recovery at the bench-scale and pilot-scale specifically. 3.3.1 Experimental set-ups and crystallization reactor designs Crystallization reactor designs have ranged from simple glass beakers at the bench-scale (49 - 51) to more complicated reactor designs, such as fluidized beds (19, 20, 52, 53) and what are generally referred to as "agitated crystallizers" (54, 55) in crystallization terminology (56). The more complicated designs have been used at the bench-scale, as well as the full-scale. 25 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 3.3.2 Required supplementations As mentioned previously, supplementation of magnesium and base with an external source is necessary. External magnesium sources previously experimented with include magnesium chloride (13, 22), magnesium oxide (16, 54), seawater (22, 55) and bittern'" (22). Since struvite forms in a theoretical Mg:N:P molar ratio of 1:1:1 (Equation 1), it follows that a magnesium ion supplementation of at least the stoichiometric requirement would be a must for the reaction to be complete. Sources of base previously experimented with include magnesium oxide (16, 54) and sodium hydroxide (13, 46, 52, 53, 55, 57). 3.3.3 Seeding the crystallization reactor Nucleation is primarily a reaction-controlled process. It has an inherent time lag period, which is a function of the saturation level of struvite. As the saturation is increased, higher concentrations of the constituent ions become available for crystallization and the nucleation lag period is reduced. By selecting struvite crystals as the seeding media, the inherent nucleation lag period is shortened considerably (46). It is possible that seed crystals must be above some minimum size before they contribute effectively to the crystallization process. This size range is reported to vary from 192 u - 225 u or 0.192 mm - 0.225 mm (58). Evidently, as the crystal particles grow to some specific size, they would have to be harvested continuously. Some studies in the literature consider additional reseeding as crucial, in order to "maintain the efficiency" of the crystallization process (46). However, there are also suggestions that once the process is underway, it becomes self-seeding (51, 54) and that reseeding is unnecessary. "' Bittern is the bitter solution of salts, which remains as a waste product after sodium chloride has crystallized out of brine. It is reportedly used for making bean curd. The total magnesium concentration of bittern can be 20 times - 25 times higher than that of seawater (22). 26 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W Previous, preliminary, bench-scale studies at the U B C Pilot Plant show that, while seeding is required at start-up, the process eventually becomes self-seeding (Frederic Koch, pers. comm.). The kinetics of struvite precipitation, and hence soluble PO4-P removal (henceforth referred to as P-removal), have also been linked to the amount of seed provided in the reactor at the time of start-up. For the given experimental set-up, the reaction rate constant so defined was shown to increase from 0.35 L ' (mmol' min)"1 - 1.4 L " (mmol ' min)"1, with corresponding struvite crystal seed concentrations of 500 mg ' L" 1 - 7000 mg ' L" 1 . The size of the struvite seed crystals used ranged from 0.074 mm - 0.105 mm (51). Some studies made use of sand (20, 21) due to its easy availability. In such cases, "frequent replacement" of seeding material is advised (21), presumably on account of the seed surfaces becoming "spent" and losing their adsorptive power (59). 3.3.4pH control, recommended reaction times and P-removals Guidelines for pH control are usually generalized based on the authors' findings and vary from study to study. A few examples are given in Table 3.3. While the general trend seems to indicate that pH control between 8.0 and 9.5 would provide a good percent of P-removal, the fact remains that the data within different rows is just that - a general trend. Significant magnitudes of difference (from a few minutes to an hour) exist between some of the reported values, even at similar influent soluble PO4-P concentrations. Previous preliminary bench-scale studies at the U B C Pilot Plant had shown that P-removals of over 95% are attainable for influent soluble P0 4 -P concentrations of 50 mg "L" 1 -400 mg ' L " ! , at a pH value of approximately 8.5. However, the form of the resultant struvite harvest (particle sizes equal to and less than 0.125 mm) was too small for convenient harvesting. It also precluded all attempts at smooth reactor operation, presumably 27 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W because the process was being operated in the zone of spontaneous nucleation (Frederic Koch, pers. comm.). Table 3.3: Guidelines for pH control for struvite crystallization (collected from the literature) (a ) Influent Scale of Recommended Recommended Reported Reference soluble operation operational time in the P-removal PO4-P pH value reaction vessel (%) (mg L"1) 50 90 - 100 Full-scale 8.4 w 8.2 ( c ) ~3 -10 minutes w 63-85 13 150-200 Bench-scale 8.0 - 9.5 ( c ) Not provided 93-95 21 50 - 600 Bench-scale 9.5 - 10.5 ( c ) 10-20 minutes 95 22 90 - 120 Bench-scale 7.8 - 8.0 ( d ) Oyer 1 hour Over 80 46 100 100 Bench-scale 8.0W 9.0 ( c ) 1 hour 30 minutes 50 80 51 60 Pilot-scale 8.5 ( c ) 1 hour 94 54 Constituents provided / present in values higher than the required stoichiometric ratio; molar ratios N:P = 6 to 8:l,Mg:P = 1 to 1.3:1, except Reference (13) where N:P = 25.1:1 and Mg:P = 0.76:1 ^ Calculated from given operational parameters (Appendix A); process runs air through the reactor Represents effluent pH value ^ pH value adjusted outside the reaction vessel 3.3.5 Ease of harvesting the final product There is no information available about the ease of the harvesting operation. In operations where crushed struvite crystals were used as seeding material, the harvested crystal sizes range from 0.002 mm (22, 54) to 0.11 mm (52, 53). While struvite has a density of 28 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 1.7 g 'cm" (60), and hence settles quite well, personal experience has shown that harvesting crystals in the size range mentioned above is extremely tedious at bench-scale, and would be much more so at a larger scale of operation. A full-scale study conducted at a Japanese WTP, provides reference to a more realistic harvested crystal size of 2 mm (13). 3.3.6Process modeling Information concerning process modeling of the crystallization reactor / experimental set-up performance is generally lacking. One study did seem to develop a fairly successful model with respect to its experimental set-up. The model incorporated operational pH values along with the authors' definition of contact time. However, the authors' claim, that the model can be universally applied, is suspect, since its concept of contact time employs four different definitions for hydraulic retention time; its application with respect to the present experimental set-up, was invalid (20). 3.3.7 Effect of temperature on struvite formation Struvite is less soluble at lower temperatures. A p K s p value of 13.12 and 12.84 has been reported at 25 °C and 45 °C, respectively (34). 3.3.8 Experiences at the full-scale There is little published research concerning practical experiences in P-recovery through struvite crystallization at the full-scale. Some valuable insights have been provided by papers detailing the use of the Unitika Phosnix Process at full-scale WTPs in Japan. As mentioned earlier (Section 1.3.1), the feasibility of adopting the Unitika Phosnix Process as a benchmark process was tested previously (before the reactor which this research tested was designed), and discounted. Nevertheless, material related to this process provides the best information for struvite formation from domestic wastewaters to date (13, 55). 29 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W 3.4 The Concept Of Recycle Ratio As A Control Parameter While none of the material referred to in Section 3.3.8 alludes explicitly to the concept of a supersaturation ratio, it does provide the recurring concept that each crystallization reactor can treat a limited strength of phosphate-containing wastewater under a given set of conditions. As a result, one of the proposed objectives of this research was to arrive at some understanding as to the use of the recycle ratio as a parameter to maintain the strength of feedwater in the reactor at some optimum concentration. 3.5 Rationale For The Reactor Design Tested / Used As mentioned earlier, there is evidence that the growth rate is limited by low mixing energy (46). Furthermore, based on experiences with previous crystallization reactor designs, the volume of the reactor was always too small to support the growing crystals from successful runs, for an adequate time before harvesting them. As a result, the harvested crystalline material was quite brittle (presumably because it had to be withdrawn before it had had enough time to attain sufficient hardness) and, at times, disintegrated to what can only be described as "mush" at the time of harvesting (Frederic Koch, pers. comm.). Therefore, some attention had to be directed towards ensuring that the growing crystals had enough time and enough space to grow inside the reactor. The design of the reactor tested / used in this study was intended to address these concerns. The reactor design follows the concept of a fluidized bed reactor. It is depicted in Figure 3.5 and Table 3.4 provides the reactor dimensions. It had four varying areas of cross-section (increasing from the bottom to the top) and hence, a different value of upflow velocity in each cylindrical section / limb. The upflow velocity was highest in limb A and decreased for limbs B, C and D, as cross-sectional areas increased. It follows then that, as the crystals grow, a natural particle size gradation would occur as a result of differential settling velocities. Larger particles would tend to settle in the lower portion of the reactor while the smaller particles would tend to occupy either limbs B or C of the reactor, depending on their 30 B A C K G R O U N D A N D P R E L I M I N A R Y L I T E R A T U R E R E V I E W size. Since the larger particles in the reactor would have the opportunity to spend time in that limb of the reactor with the highest value of turbulence, it was anticipated that they would acquire enough mechanical strength / hardness, in order to be harvested, with minimum breakage. D C B Table 3.4: The crystallization reactor - dimensions Limb Length Nominal Volume (cm) diameter (cm) (mL) A 45.7 1.54 58 B 45.7 2.06 130 C 45.7 2.62 232 D 45.7 7.81 2090 Total reactor volume = 2.5 L Feed injection port / withdrawal port Figure 3.5: Design features of the crystallization reactor (not to scale) Of course, the extent of the maximum attainable mechanical strength would be limited. On a Moh's hardness scale from 0 (description = liquid) to 10 (description = diamond), struvite stands at 1.5 - 2.0 (description = talc - gypsum) (60, 61). 31 M E T H O D O L O G Y CHAPTER 4 METHODOLOGY 4.1 Experimental Set-up 4.1.1 Constituents of the feedwater / influent Since sufficient volumes of supernatant were not available for continuous operation of the reactor and since mil control over feed characteristics was imperative at this stage, artificial feedwater, containing the constituent ions of struvite, was used as influent. The reagent salts used to supply the constituent ions for struvite were commercial grade magnesium chloride (Mg feed), diammonium hydrogen phosphate (P feed) and ammonium chloride (N feed). Commercial grade sodium hydroxide (NaOH feed, 0.2 N) was used to increase the reaction pH to the desired value. Depending on the ease of calculations for feed volume and stoichiometric requirements for the run, the same dosing tank was used for either the P and N feeds or the M g and N feeds. Separate dosing tanks were required, since using a common dosing tank for all three constituent ions of struvite could / would have resulted in some struvite precipitation inside the dosing tank itself (in spite of an approximately neutral pH value). This was borne out by past experiences at U B C (Frederic Koch, pers. comm.). 4.1.2 The reactor and the injection port Figure 4.1 illustrates the experimental set-up. The reactor was made of transparent polyvinyl chloride (PVC) plastic. Three non-transparent P V C fittings, each approximately 6 cm in height, served as transitions between limbs A - B , B-C and C-D. The pH was monitored continuously in limb D of the reactor, using a Cole-Parmer® 32 M E T H O D O L O G Y bench-top pH meter and an Oakton gel-filled epoxy body pH probe. The pH probe was calibrated daily, in line with the recommended quality assurance procedures for monitoring equipment (62). The temperature of the feedwater in the reactor was monitored once daily. Since the reactor was not maintained at a constant temperature, the recorded temperature reflected the ambient temperature at the time of its recording only. Down-pipe Clarifier Recycle line pH meter D Pumping assembly B g) Feed injection port P (+ N) feeds Mg (+ N) feeds NaOH feed Figure 4.1: Experimental set-up (not to scale) The flow characteristics of the crystallization reactor closely resembled those of a plug flow reactor. However, the constituents of the feedwater were being fed from three different tanks and were required to be adequately mixed, before their entry into the reactor. In order to facilitate this, a injection port comprising of four entry points, one for each input (including 33 M E T H O D O L O G Y the recycle flow) to the reactor, was provided at the base of limb A of the reactor. The diameter of the entry points was 3 / 3 2 " for the recycle line input and 3/^j" for the other three feedwater inputs. A slightly larger diameter was provided for the recycle line input as an added precaution, since it was more prone to plugging due to fines formation. Figure 4.2 provides a simple illustration of the injection port. It was made of stainless steel. Combined flow from all four entry points to limb A of reactor Entry point Entry point Valve / port for crystal withdrawal Figure 4.2: A simple illustration of the injection port (front view, not to scale) 4.1.3 Provision of a clarifier A clarifier was provided in order to allow settling of any fine struvite particles (hereafter referred to as "fines"), which may flow out of the crystallization reactor. The clarifier was made of transparent P V C plastic. It had an inner diameter of 9.9 cm and a height of 34 M E T H O D O L O G Y 33.1 cm i v. 4.1.4 Pump head configurations and requirements ® The process fluids were fed into the injection port using Cole-Parmer MasterFlex peristaltic pumps, in conjunction with an adjustable console drive. The number and configuration of the pump heads used depended on the planned recycle ratio requirement for the particular run and ready availability of the pump heads. The number and configuration of the pump heads used, established the ratio of the flow-rates from the different feed tanks into the reactor. This ratio was predetermined (since the number and configuration of the pump heads was known) and it was checked at the start of every run. The pump speed was controlled / set, based on the required flow rate of the feedwater, through the reactor (Section 4.2.3). 4.1.5 Diameter of the tubing used GreenLine polyethylene tubing with an internal diameter of VA" was used to supply the different constituents of the feedwater, from the pumping assembly to each entry point of the injection port. 4.2 Experimental Technique 4.2.1 Terminology /definitions used in this work A run The start of a separate run was marked by a change in any one of the following: • P / N / Mg feeds • Recycle ratio for a constant set of feeds Note that the clarifier was not specifically designed for the purpose of this research. 35 M E T H O D O L O G Y A separate run would need to be started as a consequence of one or more of the following reasons. • Plugging at the injection port due to prolonged or instantaneous struvite cementation; this was analogous to the plugging / clogging problems associated with struvite in some WTPs (Section 3.2), and / or • Partial loss of fluidization of the crystal bed, which generally occurred after 7 days - 12 days in some runs (due to a lack of enough reactor volume to support its growth), or • The diameter of the growing crystals exceeding 4 mm, thus making their harvesting through the relatively constricted withdrawal port impossible, without large amounts of crystal breakage. The time frame for prolonged struvite cementation at the bench-scale was generally 7 days - 10 days, while that for instantaneous struvite cementation was 24 hours or less. The term "partial loss of fluidization of the crystal bed" refers to a state where the measured circulation flow rate of the mother liquor through the reactor would be equal to the usual predetermined / set value, but there would be no movement of grown crystals in either or both limbs B and C. Given the nature of struvite, some fouling of the reactor and clarifier walls, transitions between limbs, pH probe and transmission tubing was inevitable. Accordingly, the system (includes the reactor, clarifier, recycle line transmission tubing and the injection port) was given an acid wash (2.5% v/v hydrochloric acid) prior to the start of each run. This was done so that a clear demarcation of the effects between one run and the next, could be made. 36 M E T H O D O L O G Y The recycle ratio The recycle ratio was set with respect to the flow of the P feed through the reactor. Therefore, a recycle ratio of 1:1 implied equal flow rates of recycled effluent and the P feed through the reactor. P-removal P-removal was calculated on the basis of the following equation: P-removal (%) = P i n f l u e n t - PefflUent * 100 (8) Pinfluent where Pinfluent - Concentration of PO4-P entering the injection port, after taking into account dilution by Mg and N feed streams (mg ' L"1) and Peffiuent- Concentration of PO4-P in the effluent collected from the clarifier (mg" L" 1). Quality of the grown crystals The quality of the grown crystals has been referred to as either "good" or "unacceptable", and generally refers to their brittleness rather than their size. While the hardness of the crystals was not specifically quantified for this work, "good" crystals were generally characterized by their superior mechanical strength. Such crystals were less brittle than the "unacceptable" crystals and always exhibited a somewhat rounded shape (Figure 4.3). 37 M E T H O D O L O G Y At the other end of the spectrum were, crystals that were exceptionally dendritic and very brittle in form (Figure 4.4). These crystals were considered as "unacceptable" on account of their not being able to withstand a harvesting operation without considerable breakage. Figure 4.4: Pictorial presentation of an "unacceptable" crystal 38 M E T H O D O L O G Y 4.2.2 Supplementations used The Mg:P ratios used ranged from approximately 0.3:1 to 1.5:1. The N:P ratios used ranged from approximately 2.7:1 to 6:1. A Mg:P ratio of 1.5:1 is slightly higher than the stoichiometric requirement of 1:1. It was used nonetheless, since over time, the magnesium chloride feed (MgCl2.6H20) was highly prone to lose its waters of hydration. This, in turn, would cause the Mg:P molar ratio to fall below 1:1 and make Mg the limiting ingredient for crystallization, even when it was not intended to be so. N:P ratios greater than 6:1 were not experimented with, since it was expected that P-recovery by this method would be utilized to serve anaerobically digested sludge dewatering sidestreams from enhanced biological phosphorus removal (EBPR) WTPs. Use of a Mg:P ratio below the normal stoichiometric requirement of 1:1, was the result of an error rather than design, but the error came in handy for the investigation pertaining to the use of the solubility criteria as a process control parameter, as did the use of N:P ratios below 6:1. pH control As shown in Figure 4.1, the pH was monitored continuously inside the reactor. Increasing / decreasing the pH entailed adding more base to that feed tank / diluting the solution already present in the tank. The amount of base addition or the degree of dilution was governed by the required pH value and the response of the pH probev. v While this may seem self-explanatory, an effort has been made to emphasize the matter of pH control specifically, for the following reason. Most runs could be operated at values of x + 0.2 only. However, +0.1 pH unit is thought to generally represent the limit of accuracy under "normal conditions, especially for measurement of water..." (63). This may call into question, the discussion on the effect of pH, on the operational control of the reactor. However, this question can be laid to rest, because the observations so made (outlined in Section 5.4) always accompanied the planned increase in pH, no matter how slight this increase. 39 M E T H O D O L O G Y 4.2.3 Circulation flow rate through the reactor and the degree of generated turbulence After considering the requirements for feed volumes, the circulation flow rate through the crystallization reactor was set to 400 mL ' min"1. In order to facilitate a fair comparison between results from different runs, the flow rate (and hence the overall effect due to the generated turbulence) was not altered throughout the entire exercise. It was maintained at a value of 400 + 20 mL " min"1 for all runs. The set upflow velocities corresponding to this flow rate, in each limb of the reactor are provided in Table 4.1. Table 4.1: Calculated upflow velocities and fluid Reynolds numbers in the reactor limbs corresponding to a constant flow rate of 400 m L ' min"1 Reactor limb Upflow velocity (cm' min" 1 ) ( a ) Fluid Reynolds number ( b ) A 217 542 ± 66 B 120 419 ± 51 C 74 321 + 39 D 8 109 + 13 Calculations provided in Appendix B Values calculated for ambient temperatures of 20 + 5 °C, calculations provided in Appendix C The Kurita Process was reported to be operating at an upflow velocity of 100 cm " min" (Frederic Koch, pers. comm.), which is quite close to that in limb B of the reactor. The concept of the fluid Reynolds number was used as a measure of the degree of turbulence in each limb of the reactor (Appendix C). However, this calculation would only serve as a guideline, since once the limbs would start to fill up with growing crystals, the degree of turbulence offered by the liquid upflow would be altered to some extent. Nevertheless, the 40 M E T H O D O L O G Y fluid Reynolds number can provide a general idea of the differences between the degrees of turbulence generated by different scales of the crystallization reactor. 4.2.4 Seeding the crystallization reactor The reactor was seeded at the start of every run, in order to overcome the time lag of the nucleation period (46, 47). Struvite crystals formed during previous experiments were used as seedvl. The size and quantity of the seed used generally depended on its availability and behaviour; i.e. whether it was retained in the reactor or washed out of i t v n . The crystal seed sizes ranged from 0.25 mm - 1.00 mm, while the quantities ranged from approximately 4 g - 7 g (precise details have been provided in Appendix D). Since there has been some speculation as to the importance of the specific surface area of the seed in the process of crystallization (46), the same seeding technique (size and quantity) was used for those runs, the results of which were to serve as comparisons between recycle ratios or differences in temperature. This was merely an added precaution, because there has been no concrete proof to date, that a smaller specific surface area of struvite seed does actually lead to better removals. 4.2.5 Sampling methodology Constituents sampled for / analyzed The constituents sampled for / analyzed included total magnesium or [ M g + 2 ] t o t a i , total ammonia-nitrogen or [NH4-N] t otai and total orthophosphate or [P04-P] t otai- Sample containers V1 Since it was probable that the recovered struvite could be used as a plant fertilizer and / or a lake fertilizer, the use of sand as seeding material was not considered prudent. v" An effort was made to predict the size of seed that could withstand the upflow velocities in each limb of the reactor, using the Stokes' law - theory of discrete particle settling (Appendix K); these concepts being the only available tools, which could used in such a case. However, the predicted / calculated values for critical particle diameter always exceeded the observed values. This discrepancy could have been a result of an incorrect assumption for the value of the shape factor constant, and the added fact that the crystals do not really behave as discrete particles (26). 41 M E T H O D O L O G Y were chosen and prepared for sampling as per the methodology provided by the United States Environmental Protection Agency (USEPA) (62). Given that tap water was used to dissolve the different constituents of the feedwater and that the amount of copper in tap water around Vancouver is relatively high, a one-time analysis was also made for the total copper or [Cu + 2 ] t o ta i in the tap water. [Mg + 2 ] t o t a i and [Cu + 2 ] t o ta i were analyzed using flame atomic absorption spectrophotometry (model Varian Inc. SpectrAA220 Fast Sequential Atomic Absorption Spectrophotometer). [NH4-N] t otai and [PC>4-P]totai were analyzed using flow injection analysis (model LaChat QuikChem 8000). Instrument operational parameters can be found in Appendix E. A test solution of a known concentration was used for each constituent of interest and with every set of samples analyzed, as a check against the prepared calibration standards. The precision or reproducibility of each set of the analyses was monitored / checkedvl". A l l samples for [Mg + 2 ] t o t a i were analyzed in triplicate. One in every five samples for [NH 4 -N] t o ta i and [P04-P]totai were analyzed in duplicate1". Sample collection Influent and effluent samples were collected at least once every 24 hours, by which time at least 100 system hydraulic residence times had elapsed and the system had assuredly reached steady state (Appendix I). Influent samples were collected from the feed tanks in proportion to the ratio of the flows, in V1" The measure of the degree of agreement among the replicate analyses of a sample is provided by the percent relative standard deviation (i.e. the standard deviation expressed as a percentage of the total). The precision was acceptable - it rarely exceeded 2% and never exceeded 5%. This data has not been provided here, since it would be too voluminous. 1X Analyzing all samples for [NH 4-N] t o t ai and [P04-P ] t o l a i in triplicate, would have led to quicker consumption and more frequent replenishing of six different chemicals involved in the analyses processes. 42 M E T H O D O L O G Y order to obtain a representative value of the feedwater concentration entering the reactor. This ratio depended on the number and configuration of the pump heads used and was also checked from time to time. Consequently, the influent samples were flow-apportioned influent samples, excluding the NaOH input stream". Since the values the concentrations of the feedwater constituents were predetermined, and the ratio of inflow between the constituents was known, the flow-apportioned influent sample served as a check / verification of the required theoretical concentrations. Effluent samples were collected from the clarifier. Each effluent sample was filtered through a 1.5 p, Whatman glass micro fibre filter, as a precaution against plugging the fine ® micro-tubing in the LaChat QuickChem instrument. Sample preservation and storage Samples for [PCVPJtotai (influent and effluent) were preserved using phenyl mercuric acetate and 3% v/v sulphuric acid. The acid preservative was added so as to decrease the sample pH to a value of less than 2, in order to prevent any further precipitation / crystallization in the sample container. A l l samples for [NH4-N]totai were preserved using 3% v/v sulphuric acid (62). A l l samples for [Mg + 2] t otai were preserved using concentrated nitric acid (62). Samples for [P04-P]totai and [NH 4-N] t otai were stored at 4 °C until analysis. Although the USEPA recommends 48 hours as the maximum holding time for [P04-P]totai samples (62), it was not feasible to complete sample analysis within this short time. Prior tests were conducted to ensure that an extended holding time of a week would not compromise the integrity of the sample, which it did not. The analyses for [NH4-N]totai and [Mg + 2] t otai were completed within USEPA recommended holding times (62). x Excluding the NaOH input stream from the flow-apportioned influent samples became necessary, because it was noticed that it promotes crystallization of struvite in the flow-apportioned influent sample. 43 M E T H O D O L O G Y 4.2.6 Harvesting crystals The crystals were harvested / withdrawn at the end of each run. The harvested crystals were ® sieved through a series of W. S. Tyler sieves, conforming to the manufacturing requirements of the specifications A S T M E - l l . The sieve size openings used were 2.362 mm, 2.0 mm, 1.0 mm, 0.5 mm, 0.25 mm and 0.125 mm. The sieved crystals were left to dry on the sieves, before being weighed (Appendix F) and stored in glass jars. Material of the size below 0.25 mm was extremely fine material and has been referred to as fines for the remainder of this report. 4.2.7 Determining crystal composition The composition of the harvested crystals was ascertained with an analysis by weight. A known weight of crystals was placed in a 250 mL volumetric flask. 1 mL of concentrated acid was added to the flask and the volume was made up to 250 mL using deionized water. The flasks were then placed in an ultrasound bath maintained at room temperature for 2 hours, so as to speed up the process of dissolution. The resulting solution was analyzed in the same manner as described in Section 4.2.5 of this report. 44 R E S U L T S A N D D I S C U S S I O N CHAPTER 5 RESULTS AND DISCUSSION 5.1 Testing The Reactor Design The response of the reactor design was very encouraging, in terms of its behaviour conforming to the predicted theory (Section 3.5). It was able to demonstrate a distinct classification in size for the growing crystals. Under certain conditions (runs A and B - Appendix H), it also demonstrated that crystal withdrawal could be performed successfully in as little as 5 days - 6 days, with minimal amounts of breakage. Past experience with older reactor designs (all of them incorporating a uniform cross-sectional area), necessitated a minimum of crystal retention time of 10 days - 14 days (Frederic Koch, pers. comm.). It should be stressed here that, at this stage, the prime concern was to assess the response of the reactor design with respect to the general characteristics of crystal growth alone. 5.2 Recommending Possible Modifications For Scale-up Preliminary testing confirmed the positive aspects of the reactor design and it was eventually decided to adopt the same design to a follow-up larger scale, i.e. a pilot-scale system. However, some modifications were advisable. It was logical to expect that the larger crystals had more time to spend in the reactor than a smaller crystal or even a crystallite that would have just come out of solution. The growth of such large crystals would occur mainly in limb A of the reactor, where the degree of the turbulence was greatest and therefore, the mechanical strength or hardness of these crystals 45 R E S U L T S A N D D I S C U S S I O N would be superior to those of smaller crystals. Any crystal withdrawal translated to emptying the entire reactor, the result being that the smaller crystals did not have as much time to grow and acquire the mechanical strength of their larger counterparts. There was no control for selective crystal withdrawal at the bench-scale and as such, an important advantage of this design was not being capitalized on x l. The pilot-scale reactor did incorporate this modification and quite successfully too. A series of valves was placed at the transitions A-B and B-C, to allow for withdrawal of crystals only from limb A or B or C, as the case might be (Ahren Britton, M.A.Sc. student, U B C Phosphate Recovery Project, pers. comm.). Other modifications suggested at the time, included the provision of a larger volume for what served as limb C. This suggestion was made in light of the observation that the majority of the smaller crystals tended to reside in limb C for quite some time, before dropping into sections B and A . This suggestion was not implemented in this study, as its usefulness was not apparent at the time x n. The design of the injection port remained unchanged in the follow-up pilot-scale system. 5.3 Verified Crystal Composition The crystals harvested from all the runs presented in this work, were very close to being pure struvite (Appendix G), with the exception of some traces of copper (sourced by the tap water). The crystals may be regarded as pure struvite for all practical purposes. The concentration of copper present in the tap water averaged about 0.41 mg ' L" 1 . *' A second bench-scale reactor was eventually built, with an attempt to include selective crystal withdrawal in the design. Valves were placed at transitions A-B and B-C. Unfortunately, the modification was not as effective as desired, because the level of control was still not as sophisticated as that of the follow-up pilot-scale reactor. x" The volume of the said section was eventually increased for the follow-up pilot-scale reactor. 46 R E S U L T S A N D D I S C U S S I O N 5.4 Factors Influencing Process Behaviour - General Observations 5.4.1 The effect of pH on P-removal The response of a given set of parameters (P, Mg and N feed concentrations) to a particular pH was quite rapid. Initial testing at a more or less constant pH (a pH unit difference of ±0.1) showed that the total P-removal possible at the given parameters had occurred. Such behaviour has also been reported in the literature (21). Figure 5.1 shows one such example. The operational pH in this case (run C, Appendix H) was 8.2 + 0.1. Figure 5.1: General response of the system to a constant pH (8.2 + 0.1) Increasing the operational pH led to a further decrease in the PO4-P concentration. The same set of parameters (P, Mg and N feed concentrations) as those for run C, were applied for run D (Appendix H), the only difference being the operational pH. Figure 5.2 shows the effect of an increasing pH on the PO4-P concentration (runs C and D combined). 47 R E S U L T S A N D D I S C U S S I O N 8.0 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 pH Figure 5.2: Effect of pH on the PO4-P concentration P-removal increased by almost 50% between a pH of 8.1 and 8.8. It follows then that i f the P-removal is to be increased, operation of the reactor at higher pH values becomes imperative. 5.4.2 The effect ofpH on the operational control of crystallization With further testing, it became apparent that operational control (smooth or chaotic) of the reactor at the bench-scale, could be defined qualitatively in terms of "fines" production at points outside the reactor. A prolonged and persistent appearance of fines in the down-pipe (Figure 4.1) always marked a turnaround for the system functioning, from smooth to chaotic, within 1 day to 3 days. The appearance of fines in the down-pipe was also accompanied by fines formation at other points outside the reactor, such as the clarifier and the transmission 48 RESULTS AND DISCUSSION tubing that served as the recycle line. The fines would eventually build up inside the recycle line and be carried towards the injection port, where plugging would occur, necessitating a system shutdown"111. It was observed that smooth operational control required that a minimal amount of fines be produced in the systemxlv. For any given set of experimental conditions being tested, there was a certain point beyond which the boundary of operational control shifted from smooth to chaotic. This shift in control, in all cases, was quite drastic - there appeared to be very little middle ground between these two defined modes of reactor behaviour - and it was brought about by a rise in the operational pH. Consequently, it was recognized that for any given set of conditions, there was a definite working limit with respect to the pH. This limit can be referred to as "the maximum allowable pH" (hereafter referred to as "pH| j m") and it varied from one set of conditions to another. From the given theory concerning the pH of minimum solubility (Section 3.2.2 and Figure 3.1) and the observations pertaining to Figure 5.2XV, it follows that a value of pHijm would restrict adequate / maximum possible struvite recovery. The appearance of fines at points outside the reactor when the operational pH exceeds pHijm, underscores the point that there is still potential for P-removal, but that the potential cannot be exploited in a manner conducive to a smooth operation. The highest measured P-removal for this project was X1" Another (perhaps less important) point in the case against fines production in the down-pipe, would be that the material forms outside the crystallization reactor; this, to some extent, defeats the purpose of providing a reactor in the first place. X1V The word "minimal" is employed here, since theoretically, it is not possible to have nil fines production. Fines can be argued to be a result of nucleation; i.e. the birth of crystallites or crystal nuclei out of solution, and crystallization cannot proceed without nucleation. Therefore, a parallel may be drawn between the terms "amount of fines" and "rate of nucleation". Thus, cases with minimal amounts of fines production, can be taken as cases where the rate of crystallite nucleation is quite limited, compared to the rate of crystal growth (47). While it may seem that limiting the rate of nucleation would also limit the rate of growth, the fact remains that both rates have to be managed in a controlled manner inside the reactor, so as to ensure that the theoretical metastable zone width is not breached and by that process, obtain an easily harvestable end product. x v pHitm for run D in Figure 5.2 was less than 8.4 pH units. 49 R E S U L T S A N D D I S C U S S I O N 74.5% (ran E - May 17, 2001, Appendix H). However, the recycle line and the injection port were plugged within a matter of 20 hours after start-up. Since smooth operational control was imperative to obtain any kind of consistent data for an overall insight of the crystallization process / reaction, the reactor was almost always operated at a p H value as close as possible x v l to or equal to pHi j m , unless otherwise specifically stated. This value was determined based on experimental observations, i.e. the first signs of appearance of fines at points outside the reactor. Further experiences with the reactor clearly demonstrated that it was not prudent to exceed this value. It naturally follows that the focal point of this project became not so much of increasing the P-removal, but rather to investigate the factors that affect P-removal. 5.4.3 The effect ofpH on the quality of harvested crystals The mode of p H operation played a major role in the quality of the harvested crystals. It was observed that runs that were operated at a p H value as close as possible to or equal to pHn m , always produced crystals of good quality. Runs that were operated at a p H value exceeding pHnm, always produced dendritic crystals; i.e. crystals of an unacceptable quality. Another observation made with respect to dendritic crystals, was their markedly increased and persistent affinity to stick to each other and the walls of the reactor. It is inherent in the nature of struvite to cause some fouling along reactor walls and this was observed in every ran. However, crystals of good quality never exhibited any attraction amongst themselves, unlike the dendritic crystals. The appearance of the latter was always undesirable from the point of view of harvesting, since they were so brittle. X V 1 Obtaining an exact value of a required pH from day-to-day was not possible, as a pH controller was not used in this work. 50 R E S U L T S A N D D I S C U S S I O N The quality of the growing / harvested crystals was decidedly independent of the shape of the seed used at the start-up of the run. The use of dendritic seed crystals would give rise to crystals of a good quality, as long as the restriction afforded by p H i j m was observed (runs A , F, G and H - Appendix H). Conversely, the use of rounded or non-dendritic seed crystals would give rise to dendritic crystals of poor quality, only when the restriction afforded by pHiim was not observed (runs D, E, I, J and K - Appendix H). 5.4.4 The effect of magnesium feed concentration on the quality of harvested crystals There has been a suggestion that "higher" magnesium feed concentrations may improve the quality of the crystals by favouring aggregation (49). However, the suggestion did not ring true in this work. Runs C and I, which were operated at a relatively high Mg:P molar ratio of 4.59:1 and 1.43:1 respectively (Appendix H), produced unacceptable crystals. Conversely, runs A, B , F and L, operated at a relatively low Mg:P ratio of 0.34:1 (Appendix H), managed to produce crystals of a good quality. As a result, it was clear that the magnesium feed concentration did not affect the quality of the crystals grown inside the reactor in this work. 5.5 The Effect Of The Seeding Technique Section 4.2.4 mentioned the speculation that the specific surface area of the seed could be one of the factors influencing P-removal. A seed with a smaller specific surface area and therefore a smaller size, could promote P-removal. This implies that differences in P-removal may exist for different seeding techniques (with respect to the seed size and quantity), where one seeding technique may offer an advantage in P-removal over the other. Since differences did exist in the seeding technique between some of the runs (depending on the seeding material on hand at the time), it was necessary to be able to judge i f and when that logic might apply. In this connection, a comparison in the seeding technique alone was made between two separate runs (M and N), keeping all other factors (P, M g and N feed concentrations, 51 R E S U L T S A N D D I S C U S S I O N circulation flow rate, NaOH addition, temperature, number of days of the run) constant (Appendix H). Table 5.1 shows this comparison. In order to overcome the uncertainty of a +0.2 pH unit reading inherent in pH probes, and also because the pH probe used so far needed replacing, the NaOH addition for both runs was governed strictly by the amount of base added to the NaOH feed tank. The runs were operated in succession, ensuring as small a difference as possible in the daily temperaturexvn. Run M was seeded with 3.88 g of 0.5 mm -1.0 mm seed size. Run N was seeded with 8.44 g of 0.25 mm - 0.5 mm sized seed, in addition to 4.80 g of 1.0 mm - 2.0 mm sized seed. Table 5.1: Effects of the seeding technique at run start-up Run Total weight of seed Pinfluent (mg - L"') P-removal (%) ( a ) identifier added at start-up (g) M 3.88 39.2 20.8 ±6 .9 N 13.24 39.9 23.2 ±7 .1 Average + standard deviation, operational pH was not increased beyond pH];, From Table 5.1, it becomes clear that the difference in P-removal between the two runs was not significant. These observations and the result in Table 5.1 become important in discounting the effect of different seeding techniques on the P-removal. The most commonly used seed size for the other runs was 0.5 mm - 1.0 mm, while the amounts of seed used varied from 3.88 g - 6.99 g (Appendix D). x v" This becomes significant in light of the fact that the difference between the recorded temperatures of runs operated in March and September was at least 10 °C. 52 R E S U L T S A N D D I S C U S S I O N Further, i f seed with a larger specific surface area (or a larger size) were to somehow degrade the P-removal efficiency, then it would not be unreasonable to expect no further growth or slower growth at such sizes. Both the largest and the smallest seed used for this exercise showed signs of growth two days after run start-up. There was also no perceivable difference in the rate of growth among the seed sizes tested. In fact, the largest seed size (1.0 mm - 2.0 mm) had grown to a diameter of about 5 mm before it was harvested. 5.6 Investigating The Influence Of The Recycle Ratio The influence of the recycle ratio was investigated with three different sets of feedwater constituents. A summary of the operating conditions pertinent to the discussion in this section is provided in Table 5.2. A complete daily record of the data appears in Appendix H . 53 o o CD J—i CD CD o G CD =3 _g -*—» cl .HP co CD > CN CD -4—» CD C/5 O cd -o PH <U w 00 Hi PQ .3 5 + 3 3 < C O - - H rt .s CD CD 00 I-I Cl, 2 1 JH o CD i _ rt CD c « rt G " CD T3 ON ro ro CN b ro ro ro VO ro ro ON oo CN ro CN ON CN 00 ON ON p od o p I—, od od 00 o o o -+—» + - » -HH vq r--. oo r-^  r-^  o CN , o CN vd vd ro ro ON CN CN * d O ON b ON ON CO ro O PH C N u rt CD 00 °r o <w> Cd - O PH cu " IH ^ 00 CD CQ :a ^ + C O CD 23 rt .3 <+H O c o • » d •c cl o O C D - ^ e o CD o PH Cl ^ O PH CD 60 C H PH ,3 £ 8 CD rt PH S-I CD * g T3 CN oo ON od * o 00 b b CN CN od od CN ro ro <r-i ro o o » -*-» i—1 CN * o ON ON uo ro CN i - H ON CN oo CN co oo oo a u o C N + 1 cj CH O U •s •B -a u CJ CH 0> CO . H I n o 60 o i o .3 cd > K a-•a o 2; | o o o l-l o tD o d <u (D .2? > CN ro u i P o PH bo u + CO ° S g s s £ * -J3 o d U PH g • C PH 42 o O <t> U PH C£ l-i P. PH .2 ^ rt o 8 2 rt rt g in co in ro CN oo i/-> vd r - H VO o vd Ov CN oo 00 V ro 00 00 oo o o -4—> o 00 oo r-ON i CN ro oo o CN ro ON 00 iri U u o <N + 1 JS o a <u C 3 C JS 60 t3 l-i <U > a JS o TS £ .g u = 1) O a o C3 > I) T3 •a s o < 2 l-l > R E S U L T S A N D D I S C U S S I O N 5.6.1 The resulting PO4-P concentration inside the reactor It can be seen from sets RCY1 and RCY2 in Table 5.2, that as the recycle ratio increases, the contribution of PO4-P concentration from the feed tank decreases, while that from the recycle line increases. The resulting PO4-P concentration inside the reactor eventually ends up being virtually the same, regardless of the recycle ratio. An effort was made to operate the reactor beyond pHijm at a higher recycle ratio of 7:1 (Table 5.2, set RCY3), but this required a daily cleaning out of plugged recycle lines and a plugged injection port. There was a slight decrease in the average resulting PO4-P concentration inside the reactor (25.5 mg ' L" 1), but it was not much lower than its counterpart at a recycle ratio of 1:1 (33.4 mg " L" 1). Detailed observation of a part of the daily record (Table 5.3) shows that a higher pH did manage to reduce the PO4-P concentration inside the reactor, compared to a lower pH. Table 5.3: Effect of pH on the PO4-P concentration inside the reactor Run PH Contribution of Contribution of Resulting PO4-P identifier PO4-P from the PO4-P from the inside the reactor Feed tank (A) recycle line (B) (A + B) D 8.4 9.0 23.2 32.2 D 8.8 9.0 9.6 18.6 All concentrations in mg' L" According to the concept of the recycle ratio as a control parameter (Section 3.4), a recycle stream would be expected to dilute the PO4-P laden-stream from the feed tank, so that a 56 R E S U L T S A N D D I S C U S S I O N smaller PO4-P concentration would result inside the reactorxvl". Also, when one considers the SSR (Section 3.2.3), for a particular pH, a smoother operation control would be achieved for a smaller conditional solubility product (Ps), than for a larger one. Therefore, although the need to employ the recycle stream to obtain a smaller P s inside the reactor seems to be logical, it cannot be corroborated by the results in Table 5.2, the reason being the inadequate crystallization of PO4-P to struvite, which in turn is the result of the restriction provided by pHnm. 5.6.2 Indications given by the limited data from the pilot-scale reactor Fortunately, limited results from the pilot-scale reactor, that can substantiate this theory are available (Ali Adnan, M.A.Sc. student, U B C Phosphate Recovery Project, pers. comm.). The operational conditions of the follow-up pilot-scale reactor are similar to some of those at the bench-scale, with respect to the Mg:P and N:P molar ratios, and the ambient temperature (Table 5.2, set RCY2). The Pinfluent concentrations between the two scales differ by 10 mg ' L ' 1 to 20 mg ' L" 1 . The association between the two scales is shown in Table 5.4. The results in set R C Y 2 do not contain a run at a recycle ratio of 6:1. However, it is evident that the resulting PO4-P concentration inside the reactor stays largely unchanged among different recycle ratios, for each set of Table 5.2 and as such, it would be highly improbable to suppose that the results for a recycle ratio of 6:1 for set RCY2 would differ from this norm. It can be estimated with a reasonable amount of confidence, that the results for a bench-scale run with a 6:1 recycle ratio and P influent value of 80 mg'L" 1 , would have a contribution of 17.8 mg ' L" 1 of PO4-P from the feed tank and a contribution of about 52.2 mg ' L" 1 PO4-P from the recycle line. Since the resulting PO4-P concentration inside the reactor remains unchanged among runs with different recycle ratios, the value of the operational pH would also remain unchanged (i.e. 7.3). The above estimation has been shown in Table 5.4, in italics. X V 1" This analogy, of course, also applies to the other feedwater constituent streams required for struvite crystallization - M g and N . In order to keep the explanation short and also because the main focus of the experiments was the P-removal, only PO4-P concentrations have been referred to in this report. 57 d cu cu cu cd « I-I cu o cu I—l (L) •B M - l o cu o c cu fl CD -fl 00 fl "S OH 3 o U i r i CD H GO O -4—» o cd CD cd o oo C U cd o 00 . f l O fl CD JO CD HA * o O cd " l-i 00 CD .S £ •3 u fl -o 00 CD co rt .3 <4H o c fl -D -4—» fl o U PH <+H CD O .fl C C D --fl -~ B ^ o 2 ot i .3 PH J D O ^ 2 B < j o • a PH cd fl T 3 O O CD U PH O£ CH pLn cd O PH & o '"fl S 1 e CD PH1 O cd CD Pi "cd o 00 o o N O T f O N CN © IT) i r i i r i CN CN TI- r o O N O N b CN CO CO od 00 00 r o r--CO O *-; i r i N O i r i O o o ^q N O in r—1 T—i *—M CN CN vd N O vd IT) CN CN b CN od N O N O N O CH fl CD "cd CD on CN O N N O 00 b CN r o Ti-r o CN OO 00 00 N O N O b TT CN od od TT CN r o N O ^H O N O N i r i i r i CN O N C N i—i 00 CO oo CN lo 0O <N~I K O N fo *-1 0O o PS « fl - K cu ry -5 CQ & t? cu U o l O + 1 g iS HS o C3 CJ c iS 2 nt IH CJ CJ "3 J 3 o a CJ X> CJ so c3 IH CJ > ca CJ t+H CJ I H -a J" X • S CJ w o ' f l <S M IH O a CJ o a o o CJ B a CJ R E S U L T S A N D D I S C U S S I O N The data from the pilot-scale reactor (Table 5.4) indicates that the recycle stream and the on-going crystallization reaction, together, managed to reduce the overall PO4-P concentration inside the reactor. This, in turn, provides a smaller and more manageable P s inside the reactor (Figure 5.3). 5.5 6.5 9.5 7.0 . pPs of the feed to be treated (at the injection port) pPs (inside the reactor) 7.5 8.0 8.5 9.0 pH 9.5 10.0 10.5 11.0 • Limited data from the scaled-up reactor Conditional solubility product curve from UBC experiments Figure 5.3: The management of pP s inside the pilot-scale reactor by the recycle stream It is worth mentioning here that the P-removal under these operational conditions was 89% - 96% and that the quality of crystals obtained during this period was good (Aii Adnan, pers. comm.). The recorded ambient temperature in this case was 20 °C. The data from the pilot-scale reactor also indicates that the operational pH could be taken to a value of 8.3, which is greater than a value of 7.3 at the bench-scale. This bears out the theory of pHnm being responsible for the failure of the recycle stream to function as intended, at the bench-scale. 59 R E S U L T S A N D D I S C U S S I O N 5.6.3 Possible causes for pHu,n being a restriction at the bench-scale operation Table 5.4 shows that there is not much of a difference in the feedwater concentration being treated at the bench-scale and the pilot-scale. Yet, a rigorous restriction of pHijm exists at the bench-scale. The pilot-scale reactor seems to be subjected to this restriction as well, but to a milder degree in the initial stages of a run and apparently none in its later stages. There are two possible factors that could account for this difference in behaviour between the two scales. The first (and perhaps more obvious) factor is the difference in the degree of turbulence offered at equivalent points inside the reactor between the two scales. The use of the fluid Reynolds number as some sort of a measure for the degree of turbulence, has already been explained in Section 4.2.3. Table 5.5 shows the calculated fluid Reynolds number for the two scales of the reactor design. While the bench-scale reactor was operated at a circulation flow rate of 400mL'min" 1 , the fluid Reynolds numbers for a circulation flow rate of 550 mL 'min"1 have also been included, since the pilot-scale reactor was operated at upflow velocities equivalent to a circulation flow rate of 550 mL ' min"1 at the bench-scale (Ahren Britton, pers. comm.). Table 5.5: Calculated Reynolds numbers for the bench-scale and pilot-scale reactors (25 °C) Reactor limb Fluid Reynolds number Bench-scale Bench-scale Pilot-scale (400 m L ' min"1) (550 m L ' min'1) (equivalent to an upflow velocity of 550 m L ' min"1 at the bench-scale) A 609 837 2072 B 471 648 1608 C 360 496 1083 60 R E S U L T S A N D D I S C U S S I O N The degree of turbulence generated in the pilot-scale reactor exceeds that in the bench-scale reactor by a factor of 2.5 - 3.5, notwithstanding a higher circulation flow rate of 550 mL ' min"1 at the bench-scale. There have been suggestions that increased turbulence can "overcome limitations for transport of constituent ions to the incorporation sites in the growing crystal lattice configuration" (46). A more simple (but equally ambiguous) association may be made between the kinetics of the crystallization reaction and the provided parameters, which are more physical than chemical in nature (turbulence in this case, assuming all other physical parameters are constant). This phenomenon has been used as a possible explanation as to why there is still so much confusion in the literature with respect to the measured kinetics and mechanisms of struvite crystallization (29, 50, 64, 65), as well as crystallization in general (56, 66 - 74). Therefore, it is conceivable that the increased turbulence in the pilot-scale reactor speeds up the kinetics of the crystallization reaction, so as to make it complete within the reactor and prevent its formation at points outside the reactor. Therefore, although the factors pertaining to the chemistry of the solution (P, Mg, N feed concentrations and the recycle ratio) are quite similar between the two scales, the turbulence inherent at the bigger scale creates conditions beneficial to the system's overall performance. An effort was made to measure the velocity of the reaction with the bench-scale set-up, by duplicating all of the conditions for three particular runs (Appendix I). These runs were chosen based on the first signs of observed growth in the seed (within 2 days - 3 days of run start-up); these were the earliest of all the runs experimented with and also of course, the existing ambient temperature at the time. However, the rate of change in the constituents was too small for the analytical instrument to record accurately. This may be taken as another indication of the kinetics of the reaction being particularly slow at the bench-scale, with the provided conditions, even for the runs exhibiting what was considered to be the best observed pace of crystal growth. 61 R E S U L T S A N D D I S C U S S I O N The second factor that could account for the difference in behaviour between the two reactor scales is the amount of material (crystals) present in the reactor at any time during the run. Prolonged runs lasting 30 days - 40 days at the pilot-scale have shown that the pH can be increased gradually, as the amount of material builds up inside the reactor. For the specific case referred to in Table 5.4, the pH was increased from a value of 7.4 to the reported value of 8.3, over a period of 33 days (Ali Adnan, pers. comm.). Some loss of fines has been reported to occur in the down-pipe of the scaled-up system as well, but this did not cause as severe a result as that associated with the bench-scale. Therefore, the concept of pHum also extends to the pilot-scale, but to a lesser degree. Interestingly, once the reactor is close to being fully loaded with crystals (which happens over time as crystal growth progresses), there is no fines loss in the down-pipe, regardless of the operational pH (Ahren Britton and A l i Adnan, pers. comm.). It would then appear that the restriction created by pHijm is removed, at such a stage, and that the operator of the system is free to increase the pH to such a value that would provide the best possible system performance. It is probable that a fully loaded reactorxlx has enough material to bind with the tiny crystallites as they nucleate from the solution and thus acts as a retainingxx sieve, which prevents the tinier material from escaping from inside the reactor. A partially loaded reactor would be a less effective retaining sieve. This behaviour was noticed at the bench-scale as well (runs A , B, G - Appendix H), but to a lesser extent, since complete withdrawal of crystals usually became inevitable 7 days - 12 days after the start of a run, and the reactor was never completely loaded with crystals within that time. Another point worth considering with respect to a more completely loaded reactor would be X1X No precise quantitative definition exists for the term "fully loaded reactor" as yet. In general, for the follow-up pilot-scale reactor, a fully loaded reactor may be looked on as one where frequent (daily to once in three days) crystal withdrawal from limb A becomes necessary, in order to provide space for the growing crystals in limb C (Ali Adnan, pers. comm.). x x Acknowledgement and thanks are due to Ali , for supplying the very appropriate word "retaining" to my analogy of the "sieve". 62 R E S U L T S A N D D I S C U S S I O N its ability to speed up the kinetics of the crystallization reaction (51). Contact between crystals and crystal-crystallizer components increases the rate of production of potential nuclei, as well as the rate of growth (47, 56, 58, 75). It has also been shown that the required time to attain "steady state" can be shortened, i f an appropriate seeding regimen is adopted at start-up (76). In this light, a re-examination of the material in Section 5.5 implies that at the bench-scale, the seed loading by weight would have to be greater than 13.24 g, in order to increase the kinetics of the crystallization reaction. 5.6.4 Effect of the recycle ratio on P-recovery Table 5.6 shows a summary of the effect of the recycle ratio on P-recovery. Table 5.6: Effect of the recycle ratio on P-removal at the bench-scale Run identifier Recycle ratio pH range ( a ) P-removal (%) ( b ) Set 0 1:1 7.6 to 8.0 25.2 + 6.5 RCY1 M 2.9:1 7.7 to 8.0 20.8 ± 6.9 P 6.2:1 7.8 to 8.1 27.9 ± 11.1 Set Q 1:1 7.1 to 7.5 16.5 ±3 .2 RCY2 R 2.9:1 7.2 to 7.3 17.9 ±2 .3 Set H 1:1 8.0 to 8.3 37.9 ±8 .2 RCY3 D 7:1 8.4 to 8.8 47.6±21.7 Operational pH was not increased beyond pH^ , unless otherwise stated Average + standard deviation The observed spurt in the standard deviation for run P, is the result of the operational pH 63 R E S U L T S A N D D I S C U S S I O N exceeding pHijm for two of the days of the run (Appendix H). In sets RCY1 and RCY2, where the pH was maintained at a value close to or equal to pHijm, an increase in the recycle ratio could not and did not lead to a statistically different P-removal. This is not surprising, as there was also no reported decrease in the eventual PO4-P concentration inside the reactor (Table 5.2). Run D did see a slightly greater P-removal than run H , as the pH was increased to a value beyond the allowable pHijm. The higher the pH overshoot beyond pHijm, the greater the P-removal (Appendix H). The quality of crystals harvested from all the runs in sets RCY1 and R C Y 2 was good. Although the quality of crystals harvested from run H was also good, run D produced unacceptable crystals on account of its being operated at a pH value exceeding pHijm. 5.7 Investigating The Response O f The System To Lower Pjnnuent Concentrat ions The response of the crystallization system was also tested at lower Pjnfiuent concentrations of 20 mg ' L" 1 - 30 mg ' L" 1 . The summary of results is provided in Table 5.7. Tables 5.2 and 5.6 have already provided some information concerning the behaviour of Pinfluent concentrations ranging from 40 mg ' L" 1 (set RCY1) to 80 mg ' L" 1 (set RCY2). In spite of the reactor being operated at a ceiling value of pHijm (7.9 and 7.3 pH units respectively), the first signs of crystal growth were normally visible within 3 days - 4 days of the run start-up. 64 CO o 'fa o o o o PH 1-o S m -*-> CO >. CO CO G O P . CO <D l-i <D -*-» 60 c • t-H -4-> cd • i—i -4—» CO > a (U •§ H O rt " l-i .S •£ co .SS <U co rt .3 l l PH 3 •c +-> . — o O >> U PH g l-l 'a PH i2 o O <P O rt 60 - O 43 £ CO <D o co .1 CO -*-» CO l-l p_, cd O *11 R-i & o (D 8 2 rt i—i g £ rt g T3 00 CO © CN i r i i—i o CN vq O N CN i r i i r i ON 14.6 ON ON 11.2 ON O o 00 oi oi o CN r -o\ o 1*0 c n CO © CO co ON o *—1 r—1 T-H r—< CN CN ON 00 CN 00 O CO 00 H u o m +1 a i3 rt > s 2 o o ft 3 ^ ,a 01 l-l rt > to &o a> u rt "rt > > R E S U L T S A N D D I S C U S S I O N A lower Pmfiuent concentration generally required a comparatively higher operating pH to start exhibiting signs of growth. The volume of NaOH addition required to achieve such a pH was 4.5 times - 5 times more than that required at higher PO4-P concentrations. Bearing this point in mind, it would appear that treating Pmfiuent concentrations at or below a value of 30 mg ' L" 1 may be much less cost-effective than treating higher Pmfiuent concentrations'"". Runs S and U (Table 5.7) show that doubling the recycle ratio did not affect the overall process of crystallization, for the same reasons as mentioned in Section 5.6.1. Runs S, T and U displayed the first signs of crystal growth at a pH value of about 9.0. Any operation at a pH below this value led to a gradual dissolution of the seed until it eventually disappeared, implying that the conditions inside the reactor were undersaturated. Once a pH of 9.0 was eventually applied, the pace of growth of the seeded crystals was noticed to be extremely slow; roughly 2 times - 3 times slower than that observed for sets RCY1 to R C Y 3 . Run T was operated at a significantly higher N:P molar ratio of about 5:1, a Mg:P molar ratio of about 1.4:1 and a lower PO4-P concentration of about 20 mg ' L" 1 . It also required a high operational pH (9.0), to ensure that the seed did not dissolve. At a pH of 8.9 - 9.0, the P-removals for these runs were calculated to be about 40% (Appendix H). The amount of fines production in the down-pipe at a pH of 8.9 - 9.0 was "alarming", enough not to warrant a further increase in the operational pH. Although three of the runs were eventually operated at a comparatively high pH of about 9.0, the harvested crystals were confirmed to be struvite (Appendix G). However, the quality of crystals harvested from each of these runs was unacceptable. X X 1 The same pump head configuration was used for high PO4-P and low PO4-P concentrations (Appendix H). Therefore, the noted differences in the required volume of NaOH are truly attributable to the differences in the PO4-P concentrations alone, and not to any differences in dilution, which may arise if different pump head configurations were used. 66 R E S U L T S A N D D I S C U S S I O N 5.8 Investigating The Feasibility Of Applying Solubility Criteria As A Process Control Parameter 5.8.1 Conditional solubility product values Figures 5.4 to 5.6 represent the negative log of the conditional solubility products (pPs) for sets R C Y 1 , RCY2 and RCY3, plotted on the pP s versus pH curve from Figure 3.1. Three values of the pP s have been plotted for each set, namely • pP s of the flow-apportioned influent sample or pPs-inf, • pP s of the mother liquor inside the reactor or pP s-mi, and • pP s of the effluent or pPs-eff-Reference to pP s-mi assumes significance because of the role played by seed-crystal loading in the overall kinetics of the reaction (Section 5.6.3). Theoretically, the pPs.inf would indicate the potential for struvite formation, but it cannot and does not provide a reflection of the important parameter of reaction kinetics. Theoretically, the pPs-eff should give an idea as to how much of the struvite formation potential has been spent in the crystallization reaction. Each data point in Figures 5.4 to 5.6 represents the average value of the pP s under consideration, throughout the run. Since Figure 3.1 is applicable for a temperature of 25 °C, using it as a comparison against data points generated at temperatures other than 25 °C would not be correct. The majority of the runs were operated in a recorded temperature range of 20 °C - 25 °C. Since the rates of the dissolution reaction were unknown, it was not possible to ascertain what (if any) change the curve in Figure 3.1 would undergo at a temperature of 20 °C, by means of a simple 67 R E S U L T S A N D D I S C U S S I O N Arrhenius-temperature dependence relationship (26). The runs in sets R C Y 1 , RCY2 and RCY3 were carried out within an ambient recorded temperature range of 20 °C - 25 °C. In plotting Figures 5.4 to 5.6, it has been assumed that no appreciable differences exist in struvite solubility between 20 °C and 25 °C. The validity of this assumption will be borne out in Section 5.8.4. 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 pH a pPs-inf A pPs-ml • pPs-eff Conditional solubility product curve from UBC experiments Figure 5.4: pP s versus pH plot - Set RCY1 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 pH a pPs-inf A pPs-ml • pPs-eff Conditional solubility product curve from UBC experiments Figure 5.5: pP s versus pH plot - Set RCY2 68 R E S U L T S A N D D I S C U S S I O N 5.0 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 pH • pPs-inf A pPs-ml • pPs-eff Conditional solubility product curve from UBC experiments Figure 5.6: pP s versus pH plot - Set RCY3 The data point representing pP s-mi lies below the solubility curve in Figures 5.4 and 5.6. Theoretically, this would mean that the conditions inside the reactor are not supersaturated and no struvite formation would be possible. However, the observations and results show that struvite was formed to a certain extent and therefore, in reality, the conditions inside the reactor were supersaturated. This inconsistency between the graphical implication of the data and the actual observations was noticed for almost all the runs (except runs Q and R) operated in the temperature range 20 °C - 25 °C, and is depicted in Figure 5.7. Each data point represents the average value of the pP s under consideration throughout the run. The reason for this inconsistency / behaviour has been elaborated on in a later section (Section 5.8.4). 69 R E S U L T S A N D D I S C U S S I O N 6.5 9.5 -I 1 1 1 1 1 1 1 1 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 pH • pPs-inf * pPs-ml • pPs-eff Conditional solubility product curve from UBC experiments Figure 5.7: pP s versus pH plot - A l l runs (20 °C - 25 °C) The data points representing pP s-mi and pPs-eff almost overlap one another in Figures 5.4, 5.5 and for run H in Figure 5.6. This would suggest that not enough P-removal is being achieved for a distinction between the two data points to be possible. From Table 5.6, it appears that this limit is set by a P-removal of approximately 40% (run H). Run D in set RCY3 managed to achieve a P-removal of 69.7% (Appendix H) at a pH value of 8.8 (> pHijm). In this case, the data points representing pP s-mi and pPs-eff have been able to record such a difference in terms of the achieved P-removal. Figure 5.7 also shows three data points for pP s-inf (corresponding to runs S, T and U), which lie below the curve. This would imply that the influent is undersaturated and that no struvite formation is possible. However, crystalline growth was observed, contrary to this implication. 5.8.2 Reasons for runs Q and R differing from the general trend Runs Q and R were not conducted at a temperature any higher than that for the other runs 70 R E S U L T S A N D D I S C U S S I O N presented in Figure 5.7. Therefore, the factor of a differing temperature may be safely ruled out as a cause for these runs differing from the general trend. The runs presented in Figure 5.7 have a calculated average ionic strength of 0.01 M for the mother liquor inside the reactor (Appendix J). However, the corresponding value for runs Q and R is 0.03 M . As Figure 3.4 demonstrates, an increase in the ionic strength would cause the solubility curve to shift in an upward direction. This would imply that the pP s-mi data points for runs with an ionic strength exceeding 0.01 M would also shift in an upward direction, which is what we see in Figures 5.5 and 5.7. Therefore, differences in ionic strength account for these runs varying from the general observed trend. This explanation underscores the importance of ionic strength where struvite solubility is concerned. 5.8.3 The effect of temperature on struvite solubility Since this project was started in March 2001 and concluded in September 2001, it was possible to make a short study of the effect of temperature on struvite solubility. Run A was operated at a measured ambient temperature of 15 °C, while run L was operated at a measured ambient temperature of 25 °C. A l l other influencing parameters such as the Pinfluent, the M g and N supplementations, the pH, seeding technique and number of days of the runs, were kept virtually constant. Figure 5.8 represents the daily pPs-eff values versus the daily operational pH for both runs. The regression lines for both runs are almost parallel to one another. At a lower temperature, the regression line shifts downwards, indicating that more struvite has formed. This is substantiated by the recorded P-removals for runs L and A , 27.1% and 37.8%, respectively. It was also observed that the rate of struvite growth was faster for run A , an observation corroborated by the size distributions by weight of the harvested crystals, shown in Table 5.8. These data indicate that a temperature of 15 °C is more beneficial to struvite formation than a temperature of 25 °C. Although a lower temperature is known to decrease struvite solubility 71 R E S U L T S A N D D I S C U S S I O N (Section 3.3.7), these data provide a quantification of the extent of the said decrease. Figure 5.8: The effect of temperature on struvite solubility Table 5.8: Size distributions by weight of crystals harvested from runs with constant operational parameters at 15 °C and 25 °C Sieve size (mm) Weight of harvested crystals retained on different sieves (g) Run L (25 °C) Run A (15 °C) 2.362 2.09 24.42 2.0 1.30 7.17 1.0 23.85 5.24 0.5 2.83 4.23 0.25 1.42 6.65 0.125 3.29 11.97 72 R E S U L T S A N D D I S C U S S I O N The conditional solubility product curve from the U B C experiments appears in Figure 5.8, merely to illustrate its position with respect to the pPs-eff data points from run L (25 °C). The positioning of these data points below the curve is certainly odd. Theoretically, once the maximum possible P-removal has been achieved, the said data point would have to lie on the curve. An attempt has been made to account for the possible reasons for such a discrepancy in the following section. 5.8.4 An elaboration on the discrepancies between graphical presentation of the data and actual observations The clustering / positioning of the data points for runs with similar ionic strength is consistent and easily discernible from Figure 5.9. This bears out the assumption made earlier that appreciable differences in struvite solubility do not exist at temperatures between 20 °C and 25 °C (Section 5.8.1). 7.0 7.5 8.0 8.5 10.0 10.5 11.0 9.0 9.5 pH • pPs-inf * pPs-ml • pPs-eff Conditional solubility product curve from UBC experiments Linear relationship: pPs-ml - - - Linear relationship: pPs-eff Figure 5.9: The relationship between pP s values from different runs (20 °C - 25 °C) 73 R E S U L T S A N D D I S C U S S I O N However, the fact that this clustering occurs at positions below the curve is unsettling. Further, the solubility curve resulting from UBC's experiments exhibits an ionic strength of about 0.01 M in the pH range 7 - 9 . The runs depicted in Figure 5.7 exhibited virtually the same ionic strength and yet, its data points for pP s . m i lie below the curve. The possibility of the curve being incorrect has already been ruled out (Section 3.2.4). This suggests that there may be a different factor or combination of factors that need to be taken into consideration. The possible factors have been listed as follows. It is important to mention here that no supporting data in relation to any of these factors exists specifically for struvite, but these factors represent trends that repeatedly occur through all crystallization studies reviewed to date for this work. • The differences in fluid motion between the dissolution study at U B C and inside the crystallization reactor: The dissolution study at U B C was carried out in a stirred beaker, the fluid motion of which would resemble a completely stirred reactor. The flow inside the crystallization reactor closely resembles a plug flow. It is common knowledge that fluid motion would control the kinetics of the reaction and hence the rate of dissolution. This would naturally mean that the faster the fluid motion, the less the time needed to reach chemical equilibrium. Fairly complex hydrodynamic studies (77 - 80) have proven that the fluid motion also controls the pattern in which the given crystal dissolves. This implies a complicated relationship exists between both the chemical and the physical components of a crystallization-dissolution system. Whether the differences between the physical components of systems (in this case mixed flow versus plug flow) affects the chemical equilibrium (and hence the P s . e q value) is questionable and perhaps, even specious. This doubt remains unanswered, since studies of a hydrodynamic nature do not focus on the chemical aspects that this work is obviously interested in. What is known with sufficient clarity is that the constituent ions of a crystal require a certain amount of energy to be supplied to them, so that they may arrange themselves in an orderly manner in the crystal lattice (47, 81, 74 RESULTS AND DISCUSSION 82). Dissolution of a crystal involves displacing these well-arranged ions from the solid and conducting them into the solution. Differing rates of dissolution would probably cause differences in the manner of displacement and possibly, its extent (83). • Differences in crystal morphology: The morphology or structure of the crystal is known to cause differences in the equilibrium concentration of its constituent ions in solution (84, 85). The solubility experiments at UBC made use of a powdered form of struvite (Ping Liao, pers. comm.). The size distribution of the crystals harvested from different runs in this work varied from greater than 2.362 mm to less than 0.125 mm. Figure 5.10 gives an example of a precipitation diagram for a better-studied crystal compound CaHPO ,^ one of the forms of calcium phosphate. The different boundaries T r Figure 5.10: Precipitation diagram for the formation of CaFfPC^ from the CaCh-NasPCVHaO system in 0.15 M NaCl solution at pH 5 and 32 °C. Region 1 - unsaturated solution; 2 - metastable supersaturated solutions; 3 - rhombohedral CaHP04.2H20; 4 - intergrown and twinned CaHP04.2H20; 5 - CaHP04.2H20 and spherical agglomerates of CaHP04 (84) 75 R E S U L T S A N D D I S C U S S I O N in this diagram demarcate the concentration and where appropriate, the pH and temperature regions within which precipitates having similar features such as morphology, colour, chemical composition or solid phase structure are formed (84). • The differences between the activity / thermodynamic property of the crystals used in the dissolution study at U B C and the crystals grown from solution in the reactor: This factor has already been highlighted in previous sections. It is notable that there is no quantifiable definition for the activity of a crystal form. The activity of the solid phase is said to be "implicitly contained in the solubility equilibrium constant" (39) and hence, in its P s . e q value. • A leading textbook on water chemistry (32) notes the following in relation to crystallization and dissolution: "Since very small particles have a higher surface energy than larger particles, the solution concentration in equilibrium with small particles is higher than that in equilibrium with larger particles. Consequently, in a mixture of particle sizes, the large particles wil l continue to grow because the solution is still supersaturated with respect to them. As the concentration of the solution is lowered through the growth of the larger particles, the smaller particles dissolve because the solution concentration is now below their saturation value. Conversion of small particles to larger particles is also enhanced by the agglomeration of smaller particles to form larger particles". Therefore, it is possible and quite probable, that a comparison of "like-to-like" is not being made in plotting the data points for pPs-mi and pPs-eff on the given solubility product curve and hence the reason for the data points appearing below the said curve. 76 R E S U L T S A N D D I S C U S S I O N 5.8.5 pH versus P-removal andpPs.mi Figure 5.11 shows the relationship between the pHum, P-removal and pP s - m i for runs conducted between 20 °C - 25 °C. Figure 5.11: The relationship between pHijm, P-removal and pP s-mi (20 °C - 25 °C) It can be seen that an increase in P-removal is possible with an increase in pH and an increase in pPs-mi- This underscores the importance of increasing the pH to increase the P-removal. The circled outlier for P-removal is the result of a run conducted with a Mg:P molar ratio of 0.34:1. P-removal was suppressed, since magnesium became the limiting ingredient. 5.8.6 The utility of the pPs versus pH plot As mentioned earlier (Section 4.2.2), not all runs were operated at a Mg:P molar ratio of 1.5:1 and a N:P molar ratio of 6:1. The pP s . m i values from these runs (runs A , B, D, G, H 77 R E S U L T S A N D D I S C U S S I O N and K-Append ix H) have already been incorporated into Figures 5.7 and 5.9, along with other runs operated at a Mg:P molar ratio of 1.5:1 and a N :P molar ratio of 6:1 (runs M , O and P - Appendix H). There appear to be no differences among data points for differently supplemented runs. The pPs-mi values for runs with low Pmfiuent (Section 5.7) have also been included in Figures 5.7 and 5.9. Again, there appear to be no differences among data points for different Pinfluent concentrations. In fact, the regression line fit values (R 2 = 0.95) are very encouraging. This suggests that, as long as the pPs-mi value is somewhere along the plotted bold line (line AB) given in Figure 5.9, one would be able to predict the pH requirement of the reactor. Since the plotted values are a result of operating the reactor at a value of pHij m , this pH requirement would correspond to pHij m . In fact, such a prediction was attempted for runs O and Q and it worked successfully""1. In general, a lower pPs-mi value (or higher concentration of mother liquor inside the reactor) translates to a lower pHum requirement, which in turn, becomes detrimental to the desired increase in P-removal. Figure 5.12 shows plotted pP s.mi values using data obtained from certain days that exceeded the pHnm value. In all these cases, the reactor would be plugged / be close to plugging after 24 hours - 30 hours of operation. The calculated ionic strength for all these cases was the same as that for the runs along line A B , i.e. 0.01 M (Appendix J), so an association between the data points and the line would not be incorrect. x x " Acknowledgement and thanks are due to Ahren, for suggesting that these predictions be attempted and for helping in attempting them. 78 R E S U L T S A N D D I S C U S S I O N 6.5 -r 7.0 -7.0 7.5 8.0 8.5 9.0 pH A Run I - daily record ° Run D - day 1 Linear relationship: pPs-ml Figure 5.12: Tolerance window for pP s-mi values The plot and observations suggest that the tolerance window for pP s-mi values at the bench-scale is very narrow. The higher the position of the point above line A B , the greater the danger of plugging. In other words, line A B is essentially a graphical depiction of the upper limit of pP s. m i value that would need to be maintained inside the bench-scale reactor, so as to ensure a smooth operation. Section 3.2.3 referred to the existence of a theoretical metastable zone. According to that definition, line A B portrays the upper boundary of that zone (depicted by C-D in Figure 3.2). The above discussion brings into question the applicability of using solubility criteria as a control parameter for struvite crystallization at the bench-scale. The following points need to be considered in this respect. • It has already been shown that the data does not relate to the conditional solubility product curve from the U B C experiments in the expected manner, depicted by 79 R E S U L T S A N D D I S C U S S I O N Figure 3.2. • Use of p P s - i n f as a control parameter would be incorrect, since in reality, this value never exists inside the reactor. It "exists" at the injection port. Limited data from the follow-up pilot-scale reactor suggests that the influent concentration may give an idea of the recycle ratio required to reduce the pP s - m i value to a manageable level inside the reactor, with a lower pP s - i n f value (or higher influent molar concentration product) probably requiring a higher recycle ratio. • The behaviour of the entire process of crystallization has been shown to be highly reactor-dependent, because of the noted differences in physical parameters such as the turbulence inside the reactor and the reactor seed / crystal loading at any given time. Limited data at the pilot-scale shows that a gradual increase in the operational pH is needed to achieve a P-removal of 89% - 96% (Section 5.6.2). Since this translates to a gradual accompanying change in the pP s - m i values, the prospect of applying an ever-changing value as a control parameter becomes dim. Only when the process reaches a true steady state, with respect to the maintenance of a constant total crystal load inside the reactor and a peak operational pH value, wil l the pP s - m i stabilize to some constant value. • The U B C experiments on struvite solubility were carried out by dissolving struvite in distilled water, as per the norm for all solubility experiments performed in a laboratory (Ping Liao, pers. comm.). Such experiments cannot incorporate the effect that any impurities or extraneous substances may have on the overall solubility of the substance under study. Any wastewater would possess different ions that may be classified as impurities, as far as the compound struvite is concerned. Such extraneous matter could change the overall chemistry of the solution and bring about a change in solubility of a compound (31, 37, 86 - 91) and possibly in the crystal morphology (87 - 89, 92). 80 R E S U L T S A N D D I S C U S S I O N • A wastewater would also possess a population of varied microorganisms. The available literature shows that different microorganisms are capable of reducing the solubility of struvite to differing extents (93 - 97). The solubility criteria provided by the dissolution experiments at U B C have been very useful in forming a hypothesis with respect to the overall chemistry of the struvite reaction using synthetic feed. However, its original motive, i.e. its use as a process control parameter for any struvite crystallizer operation, regardless of the reactor scale and feed characteristics, may not be realized due to the considerations outlined above. It wil l be possible to say for certain once more data from two ongoing pilot-scale operations, one using synthetic feed and the other sludge digester supernatant, are available. 81 S U M M A R Y A N D C O N C L U S I O N S C H A P T E R 6 S U M M A R Y A N D C O N C L U S I O N S 6.1 The broader picture The main challenge in the design of a crystallization reactor is to predict the influence of crystallizer geometry, scale, feed characteristics and operating conditions on the process behaviour and product quality. Some ideas of the crystallizer geometry were tested and discarded in the earlier phases of the U B C Phosphate Recovery Project. Numerous shades of gray still existed as far as the overall understanding of the struvite crystallization reaction was concerned. Increasing the yield of the operation in terms of P-removal, was of paramount importance. However, it was not possible to accomplish removal to the desired degree in a controllable manner. The earlier days of the U B C Phosphate Recovery Project produced P-removals of over 95%, but the form of the resulting crystalline material was extremely poor. The nature of the produced crystalline form was important in the contexts of a smooth system operation, as well as convenient harvesting. Production of very fine crystalline material was extremely conducive to daily reactor plugging / injection port cementation problems. The significance of a smooth reactor operation / control can only be appreciated in a study that employs a more "sophisticated" reactor design. This "sophistication" is manifest in the P-recovery operations based in Japan and the Netherlands, and also, in the work being conducted at U B C . The other studies referred to in the literature review do not touch on this aspect in any way. The focal point of those studies has been merely to test the chemical requirements of the reaction, using simple jar tests. Factors such as the degree of turbulence inside the reactor, seeding characteristics, and the resulting rates of nucleation and growth 82 S U M M A R Y A N D C O N C L U S I O N S have more of a physical connotation to them. Such factors demand attention at a fairly advanced state in the project. The observations made and the data collected at bench-scale were very useful in terms of laying the groundwork concepts for the overall process. A bench-scale operation was certainly less tedious to manage than a pilot-scale operation and had a lower chemical requirement. The learning curve was also very steep with respect to the subject of this research and the bench-scale studies facilitated this learning process. Nevertheless, there were serious drawbacks. In hindsight, the bench-scale reactor never reached a true steady state, which would be marked by continuous crystal withdrawal after the reactor limbs fill up entirely with crystals™". Unlike the pilot-scale reactor, it was not possible to continue a bench-scale reactor run for a time period of 30 days - 40 days, for reasons that have been already outlined in Section 4.2.1. It was not possible to quantify the recycle ratio necessary for dilution of the incoming streams of different concentrations. A l l this detracts from the ability of the bench-scale operation to approach conditions that would mirror a full-scale P-recovery project. Based on the data / results obtained from the bench-scale work, it would be prudent to concentrate future investigations concerning struvite crystallization at the pilot-scale, and preferably, with real supernatant. After a careful scrutiny of the data, experimental observations and available literature, it appears that the application of the presented solubility criteria wil l probably not act as a panacea for all struvite crystallization operations, simply because the criteria can change on a case-by-case basis. However, this should not be looked on as a discouraging "development". The entire field of crystallization is notorious for such complexities and the general consensus has been "to define such operational conditions for an already existing apparatus, under which satisfactory crystals will be produced" (56). XX1" It is for this reason alone, that the efficiency for all runs in this work has been reported as "P-removal" and not as "P-recovery". 83 S U M M A R Y A N D C O N C L U S I O N S A limited amount of data presented in this report shows that P-removal of over 90% in the desired crystalline form is attainable under certain conditions at the pilot-scale, with Pinfluent values as high as 70 mg ' L" 1 . There is every indication that such encouraging numbers are becoming more of a rule than an exception (Ali Adnan, pers. comm.). It then follows that future efforts must concentrate on duplicating those conditions, refining them and putting them to the ultimate use, i.e. P-recovery from wastewater. 6.2 Conclusions The following conclusions are drawn, based on the presented work conducted at the bench-scale and limited data from the pilot-scale crystallization reactor operation. • The tested reactor design worked admirably well, in terms of its ability to provide a good size classification in the fluidized crystal bed, by virtue of differing upflow velocities in each limb. It confirmed the conceptual utility of allowing each crystal enough time to grow to a larger size, thus making it more amenable to harvesting. • For any given set of conditions, there was a specific pH value (pHiim) beyond which the bench-scale reactor could not operate smoothly. There was clearly more potential for P-removal beyond a value of pHijm , but it was not possible to exploit it in a manner conducive to smooth reactor operation. • The operational pH is the single most important parameter with respect to its influence on controlled P-removal. An increase in the operational pH resulted in an increase in the P-removal. • The mode of pH operation plays a major role in the quality of the harvested crystals at the bench-scale. Runs operated at a pH value as close as possible to or equal to pHii m , produced crystals of good quality. Runs operated at a pH value exceeding pHij m produced dendritic / brittle crystals; i.e. crystals of an unacceptable quality. 84 S U M M A R Y A N D C O N C L U S I O N S • The quality of the growing / harvested crystals was independent of the shape of the seed used at the start-up of a run. • The quality of the harvested crystals was not related to the magnesium feed concentration. • Once a seeded struvite crystallization reaction is underway (smaller struvite crystals being used as the seed), it eventually becomes self-seeding. There were no indications of a decrease in P-removal with an increase in crystal size. Hence, there was no requirement for reseeding the reactor with smaller seed crystals at a later stage in the run. • The recycle stream can be used as an effective diluent for a high incoming Pmfiuent concentration. Limited results from the pilot-scale reactor showed that a Pinfiuent concentration of 60 mg ' L" 1 - 70 mg ' L" 1 , required a recycle ratio of 6:1 and an operational pH of 8.2 to achieve 85% - 90% P-removal. The crystals obtained at such operational conditions were of a good quality. • Low Pinfiuent concentrations (20 mg ' L" 1 - 30 mg ' L"1) exerted a very high demand for NaOH (pHiim - 9.0). Thus, it is probable that P-recovery through struvite crystallization wil l not be a cost-effective process for such low Pmfiuent concentrations. The observed pace of crystal growth was 2 times - 3 times slower than that for higher Pinfiuent concentrations (40 mg ' L" 1 - 80 mg ' L" 1), thus pointing to a higher requirement of crystal retention time at the full-scale. • Struvite crystallization is favoured at low temperatures. The kinetics of struvite crystallization, in terms of the rate of crystal growth and P-removal, showed a marked increase at a lower temperature (15 °C), compared to that at a higher temperature (25 °C). 85 S U M M A R Y A N D C O N C L U S I O N S • A lower pP s-mi value (or higher molar product concentration inside the reactor) translates to a lower pHum requirement, which in turn becomes detrimental to the desired increase in P-removal. • The data collected at the bench-scale did not relate to the conditional solubility product curve from the U B C experiments, in the theoretically predicted manner. Since the process of struvite crystallization is believed to be highly dependent on reactor hydrodynamics and seed / crystal loading, it is very probable that the solubility criteria supplied by the conditional solubility product curve from the U B C experiments may not serve as a process control parameter for all struvite crystallization systems. 86 R E C O M M E N D A T I O N S C H A P T E R 7 R E C O M M E N D A T I O N S The following recommendations are made based on the knowledge gained from this work and from the work conducted by other members of the U B C Phosphate Recovery Project. • Some studies have been conducted at the bench-scale, using wastewater sourced by the U B C Pilot Plant and the same reactor design as that tested in this work (Ping Liao, pers. comm.)™v. During the course of such experiments, it became evident that any changes in the influent composition (for example/a surge or drop in the N:P or Mg:P molar ratio) could cause drastic imbalances in the reactor. It is essential that the composition of the supernatant feed (P0 4-P and NH 4 -N) be known and controlled carefully before the feed enters the reactor, in order to avoid any system upsets. It is recommended to have holding / equalization tanks for the supernatant feed, at this stage at least. • It is advisable to shorten the time needed for the pilot-scale crystallization system to reach a true steady state. Indications are that this may be considered as the time needed for the reactor to fill up completely with crystals. Since the restriction afforded by pHijm is removed at such a stage, it follows then, that starting a run with a fully loaded / fully seeded reactor would be in the operator's best interests. • The possibility of starting a pilot-scale run at an operational pH that exceeds the respective pHnm value, with a fully loaded / fully seeded reactor, should be explored. X X 1 V Results from this work have not been included here, since the Pinfluent concentrations, Mg:P and N ; P molar ratios were very different from those tested in this work. 87 R E C O M M E N D A T I O N S It would be interesting to see i f the system can handle an immediate increase in the operational pH, without exhibiting plugging problems and i f so, to what extent. Another important point to be aware of in this respect, would be the question of the sufficiency of reactor volume. Starting the run at a higher pH would bring about a relative increase in the number of crystallites nucleating out of the solution. Added to that is the requirement of maintaining each crystal inside the reactor for some definite time (crystal retention time), until it is sufficiently hard to withstand a harvesting operation. It is very likely that extra reactor volume / space may need to be provided for these reasons. • The observations at the bench-scale have shown that crystals of an unacceptable quality are produced i f the reactor is subjected to an immediate increase in operational pH. It appears that in following the third recommendation, there is a possibility that crystals of a poorer quality may be produced. If so, there are two options open to the operator of the pilot-scale reactor. The first option would be to allow the system to reach its true steady state, by gradually nudging the pH to a higher value (as is being done at the present time). Since this would take some time (up to 3 weeks, as a rough indication), arrangements would need to be made to deal with that part of the PO4-P that would escape removal till the system reaches a true steady state. This may be accomplished by returning the reactor effluent to the equalization basin. Of course, the extent of the required P-removal may also vary on a case-by-case basis, so it may be that P-removals exceeding 90% are not necessary. The second (and perhaps less desirable) option would be setting up reactors in series and following a stepped-pH control. A stepped pH-control (gradual rise) would undoubtedly ensure a good quality of crystals. The logic of such an exercise would be to allow some P-removal at a lower pH (pHijm) in the first reactor, so that the 88 R E C O M M E N D A T I O N S second reactor, which is subjected to less of a P-load, can operate at a relatively higher p H x x v . A l l accounts at the pilot-scale operation point to a comfortable P-removal of about 40%, when a partially loaded reactor is operated at a value of pHi i m (Pinfiuent - 40 mg ' L" 1 - 60 mg 'L" 1 ) . Thus, the cumulative P-removal for two reactors in series may be estimated to be at least 75% - 80%. • Since Vancouver has very soft water, the concentration of magnesium and calcium ions in the generated wastewater is extremely low. The magnesium requirement for the struvite reaction was supplied externally and the concentration of calcium in the resulting solutions was not detectable. As a result, it was possible to obtain virtually pure struvite. However, in cases where both calcium and magnesium ions are present, calcium will tend to participate / interfere in the overall crystallization reaction and the resulting precipitate will not be pure struvite (19, 20). Alkalinity also tends to affect the struvite crystallization reaction (18, 19, 20). If the requirements of a P-recovery project demand that only pure struvite be produced, it wi l l be necessary to explore when, and to what extent, calcium ions and alkalinity affect the crystallization reaction. • It has been noted that the process of P-recovery through struvite crystallization is certainly highly reactor-specific (with respect to the degree of generated turbulence inside the reactor and the reactor geometry) and probably very site-specific as well (with respect to the characteristics of the feedwater / influent / supernatant). It is also very likely that requirements of P-recovery may vary from one EBNR WTP to another. With the existence of so many factors having the potential to affect the x x v Limited experience with stepped pH reactors in series does exist. A stepped pH-three-phase reactor using chemical feed was used for a very limited time at a preliminary stage in the crystallization studies at the UBC Pilot Plant (Frederic Koch, pers. comm.). It managed to produce cumulative P-removals of up to 70% and would have probably managed better, if the operational pH in the final reactor had been higher than 8.2. The quality of crystals obtained from this set-up was good. However, the stepped pH set-up was shelved, in the hope that a one-point NaOH injection would suffice with the newer reactor design (tested in this work). Results from the stepped pH set-up have not been included here, since its Mg:P and N:P molar ratios were very different from those tested in this work. 89 R E C O M M E N D A T I O N S overall dynamics of the crystallization process, it would be prudent to establish a protocol with respect to the manner in which any P-recovery operation be conducted in the future. 90 R E F E R E N C E S REFERENCES 1. Watanabe Y . and others (2000) Phosphorus recycling from pre-coagulated wastewater sludge. In Chemical Water and Wastewater Treatment VI. Proceedings of the 9 t h Gothenburg Symposium, October 2 - 4, 2000, Turkey, 359 -371. 2. Driver J., Lijmbach D. and Steen I. (1999) Why recover phosphorus for recycling and how? Environmental Technology. 20:7, 651 - 662. 3. Sawyer C. (1968) The need for nutrient control. Journal of the Water Pollution Control Federation. 40:3, 336 - 337. 4. Clark J., Viessman Jr. W. and Hammer M . (1977) Water Supply and Pollution Control. 3 r d edition. Publishers Harper and Row. 5. Stockner J., Rydin E. and Hyenstrand P. (2000) Cultural oligotrophication: Causes and consequences for fisheries resources. Fisheries. 25:5, 7-14. 6. The Fertilizer Industry of the European Union (1997) World Fertilizer Supply and Demand. Retrieved October 1, 2001, from the World Wide Web: http://www.efma.org/Publications/ETJBook/Section 07.asp 7. Crites R. and Tchobanoglous G. (1998) Small and decentralized wastewater management systems. McGraw-Hill Series in Water Resources and Environmental Engineering, 180-181. 8. Statistics Canada (October 1, 2001) Canadian statistics menu: Population. Retrieved October 1, 2001, from the World Wide Web: http://www.statcan.ca/english/Pgdb/People/Population/demo02.htm 9. U.S. Census Bureau (October 1, 2001) U.S. POPClockprojection. Retrieved October 1, 2001, from the World Wide Web: http://www.census.gov/cgi-bin/popclock 10. Stratful I., Brett S., Scrimshaw M and Lester J. (1999) Biological phosphorus removal, its role in phosphorus recycling. Environmental Technology. 20:7, 681 - 695. 11. Edge D. (1999) Perspectives for nutrient removal from sewage and implications for sludge strategy. Environmental Technology. 20:7, 759 - 763. 91 R E F E R E N C E S 12. Woods N . , Sock S. and Daigger G. (1999) Phosphorus recovery technology modeling and feasibility evaluation for municipal wastewater treatment plants. Environmental Technology. 20:7, 663 - 679. 13. Abe S. (1995) Phosphate removal from dewatering filtrate by M A P process at Seibu Treatment Plant in Fukuoka city. Sewage Works in Japan. 59 - 64. 14. Fujimoto N . , Mizuochi T. and Togami Y . (1991) Phosphorus fixation in the sludge treatment system of a biological phosphorus removal process. Water Science and Technology. 23:4-6, 635 - 640. 15. Greaves J., Hobbs P. and Haygarth P. (1999) Prospects for the recovery of phosphorus from animal manures: A review. Environmental Technology. 20:7, 697 - 708. 16. Schuiling R. and Andrade A. (1999) Recovery of struvite from calf manure. Environmental Technology. 20:7, 765 - 768. 17. Miles A. and Ellis T. (2001) Struvite precipitation potential for nutrient recovery from anaerobically treated wastes. Water Science and Technology. 43:11, 259 - 266. 18. Angel R. (1999) Removal of phosphate from sewage as amorphous calcium phosphate. Environmental Technology. 20:7, 709 - 720. 19. Battistoni P., Pavan P., Cecchi F. and Mata-Alvarez J. (1998) Phosphate removal from real anaerobic supernatants: Modelling and performance of a fluidized bed reactor. Water Science and Technology. 38:1, 275 - 283. 20. Battistoni P. and others (2001) Phosphorus removal from a real anaerobic supernatant by struvite crystallization. Water Research. 35:9, 2167 - 2178. 21. Momberg G. and Oellermann R. (1992) The removal of phosphate by hydroxyapatite and struvite crystallization in South Africa. Water Science and Technology. 26:5-6, 987-996. 22. Shin H . and Lee S. (1998) Removal of nutrients in wastewater by using magnesium salts. Environmental technology. 19:3, 283 - 290. 23. Giesen A. (1999) Crystallization process enables environmental friendly phosphate removal at low costs. Environmental Technology. 20:7, 769 - 775. 92 R E F E R E N C E S 24. Bridger G., Salutsky M . and Starostka R. (1962) Metal ammonium phosphates as fertilizers. Journal of Agricultural and Food Chemistry. 10:3, 181 -188. 25. Sterling M . (1997) Phosphorus release from a slow-release fertilizer under simulated stream conditions. M.A.Sc. Thesis, Department of Civi l Engineering, University of British Columbia, Vancouver, B.C., Canada. 26. Metcalf and Eddy, Inc. (1991) Wastewater Engineering Treatment Disposal and Reuse. McGraw-Hill Series in Water Resources and Environmental Engineering. 27. Booker N . (1998) Struvite formation in wastewater treatment plants - A n accident waiting to happen? In Phosphate Recovery. Retrieved October 1, 2001, from the World Wide Web: http://www.nhm.ac.uk/mineralogv/phos/booker.htm 28. Borgerding J, (1972) Phosphate deposits in digestion systems. Journal of the Wastewater Pollution Control Federation. 44:5, 813 - 819. 29. Abbona F., Lundager Madsen H . and Boistelle R. (1982) Crystallization of two magnesium phosphates, struvite and newberyite: Effects of pH and concentration. Journal of Crystal Growth. 57:6, 6 -14. 30. Booram C , Smith R., Hazen T. (1975) Crystalline phosphate precipitation from anaerobic animal waste treatment lagoons liquors. Transactions of the ASAE. 18:2, 340 - 343. 31. Buchanan J., Mote C. and Robinson R. (1994) Thermodynamics of struvite formation. Transactions of the ASAE. 37:2, 617 - 621. 32. Snoeyink V. and Jenkins D. (1980) Water Chemistry. John Wiley & Sons, New York. 33. Mamais D., Pitt P., Cheng Y . , Loiacono J. and Jenkins D. (1994) Determination of ferric chloride dose to control struvite precipitation in anaerobic sludge digesters. Water Environment Research. 66:7, 912 - 918. 34. Burns J. and Finlayson B. (1982) Solubility product of magnesium ammonium phosphate hexahydrate at various temperatures. The Journal of Urology. 128:2, 426-428. 35. Taylor A. , Frazier W. and Gurney E. (1963a) Solubility products of magnesium ammonium and magnesium potassium phosphates. Transactions of the Faraday Society. 59, 1580 - 1584. 93 REFERENCES 36. Ohlinger K . , Young T. and Schroeder E. (1998) Predicting struvite formation in digestion. Water Research. 32:12, 3607- 3614. 37. Sawyer C. McCarty P. and Parkin G. (1994) Chemistry for Environmental Engineering. McGraw-Hill Series in Water Resources and Environmental Engineering, New York. 38. Ohlinger K . (1999) Struvite controls in anaerobic digestion and post-digestion wastewater treatment processes. Ph.D. Thesis, University of California Davis, U.S. 39. Stumm W. and Morgan J. (1981). Aquatic Chemistry. Wiley-Interscience, New York. 40. Doyle J., Philp R., Churchley J. and Parsons S. (2000) Analysis of struvite precipitation in real and synthetic liquors. Process Safety and Environmental Protection. 78:B(6), 480 - 488. 41. New Dana Classification of Hydrated Phosphates (1997) VII - Phosphates - Dana. Retrieved October 5, 2001, from the World Wide Web: http ://www. webmineral. com/ dana/VII-40. shtml 42. Webb K. and Ho G. (1992) Struvite solubility and its application to a piggery effluent problem. Water Science and Technology. 26: 9-11, 2229 - 2232. 43. Sleeman K . (1963) Study of the solubility of synthetic hydroxy apatite. M.A.Sc. Thesis, Department of Soil Science, University of British Columbia, Vancouver, B.C., Canada. 44. Johnson R. (1959) The solubility of magnesium ammonium phosphate hexahydrate at 38 °C, with considerations pertaining to urine and the formation of urinary calculi. The Journal of Urology. 81:5, 681 - 690. 45. Ohlinger K. , Young T. and Schroeder E. (1999) Kinetic effects on preferential struvite accumulation in wastewater. Journal of Environmental Engineering. 125:8, 730 - 736. 46. Ohlinger K. , Young T. and Schroeder E. (2000) Postdigestion struvite precipitation using a fluidized bed reactor. Journal of Environmental Engineering. 126:4, 361 - 368. 47. Strickland - Constable R. (1968) Kinetics and Mechanism of Crystallization. Academic Press, London. 94 R E F E R E N C E S 48. Nyvlt J. (1985) The Kinetics of Industrial Crystallization. Elsevier Science Pub. Co. Inc., New York. 49. Bouropoulos N . and Koutsoukos P. (2000) Spontaneous precipitation of struvite from aqueous solutions. Journal of Crystal Growth. 213:3- 4, 381 - 388. 50. Pace G., Berton A. and Mantovani A . (2000) Ammonia removal from aqueous solutions by MgNH4PO4.6H.2O precipitation. Annali di Chimica, 90:7-8, 443 - 453. 51. Tsuno PL, and Somiya I. And Yoshino M . (1999) Production of struvite from phosphate and ammonium in supernatant of anaerobic digestion. (In Japanese) Journal of Japan Sewage Works Association. 28:324, 68 - 77. 52. Seckler M . , Leeuwen M . , Bruinsma O. and Rosmalen G. (1996) Phosphate removal in a fluidized bed - I. Identification of physical processes. Water Research. 30:7, 1585 -1588. 53. Seckler M . , Leeuwen M . , Bruinsma O. and Rosmalen G. (1996) Phosphate removal in a fluidized bed - II. Process optimization. Water Research. 30:7, 1589 -1596. 54. Munch E. and Barr K. (2001) Controlled struvite crystallization for removing phosphorus from anaerobic digester sidestreams. Water Research. 35:1, 151 -159. 55. Matsumiya Y . , Yamasita T. and Nawamura Y . (2000) Journal Of The Chartered Institution Of Water And Environmental Management. 14:4, 291 - 296. 56. Nyvlt J. (1982) Industrial Crystallization: The State of the Art. Verlag Chemie, Basel. 57. Siegrist and others (1992) In Chemical Water and Wastewater Treatment II. Proceedings of the 5 t h Gothenburg Symposium, September 28 - 30, 1992, Nice, France, 457 - 466. 58. Rousseau R., L i K . and McCabe W. (1976) The influence of seed crystal size on nucleation rates. In Analysis and Design of Crystallization Processes. AIChE Symposium Series. 153:72, 48 - 52. 59. Roques FL, Nugroho-Jeudy L. and Lebugle A . (1991) Phosphorus removal from wastewater by half-burned dolomite. Water Research. 25:8, 959 - 965. 60. Exceptional Minerals and Hall of Fame (1997) Struvite Mineral Data - General Information. Retrieved October 5, 2001, from the World Wide Web: http ://www. webmineral. com/ data/S truvite. shtml 95 R E F E R E N C E S 61. Mineral Help (1997) Mineral Hardness - Definition. Retrieved October 8, 2001, from the World Wide Web: http://www.webmineral.com/help/Hardness.shtml 62. Berg E. (1982) Handbook for Sampling and Sample Preservation for Water and Wastewater. United States Environmental Monitoring and Support Laboratory, United States Environmental Protection Agency (USEPA), Cincinnati, Ohio. 63. A P H A , A W W A and WPCF (1995) Part 4500-H - pH value. In Standard Methods for the Examination of Water and Wastewater, 19 t h edition. American Public Health Association, Washington, D.C. 64. Gunn D. and Murthy M . (1972) Kinetics and mechanisms of precipitations. Chemical Engineering Science. 27:2, 1293 -1313. 65. Golubev S. Pokrovsky O. and Savenko V . (2001) Homogeneous precipitation of magnesium phosphates from seawater solutions. Journal of Crystal Growth. 223:4, 550-556. 66. Ilievski D., Rudman M . and Metcalfe G. (2001) The separate roles of shear rate and mixing on gibbsite precipitation. Chemical Engineering Science. 56:7, 2521 - 2530. 67. Torbacke M . and Rasmuson A. (2001) Influence of different scales of mixing in reaction crystallization. Chemical Engineering Science. 56:7, 2459-2473. 68. Manth T., Mignon D. and Offerman H . (1996) Experimental investigations on precipitation reactions under homogeneous mixing conditions. Chemical Engineering Science. 51:11, 2571 - 2576. 69. Aoun M . Plasari E., David R., Villermaux J. (1996) Are barium sulphate kinetics sufficiently known for testing precipitation reactor models? Chemical Engineering Science. 51:10, 2449- 2458. 70. Aoun M . Plasari E., David R., Villermaux J. (1999) A simultaneous determination of nucleation and growth rates from batch spontaneous precipitation. Chemical Engineering Science. 54:9, 1161 -1180. 71. Pagliolico S., Marchisio D. and Barresi A . (1999) Influence of operating conditions on BaS04 crystal size and morphology in a continuous Couette precipitator. Journal of Thermal Analysis and Calorimetry. 56:3, 1423 -1433. 96 R E F E R E N C E S 72. Tai C , Wu J.and Rosseau R. (1992) Interfacial supersaturation, secondary nucleation and crystal growth. Journal of Crystal Growth. 116:3-4, 294 - 306. 73. Liu Z. and others (1995) Hydrodynamic control of inorganic calcite precipitation in Huanglong Ravine, China - Field measurements and theoretical prediction of deposition rates. Geochimica et Cosmochimica Acta. 59:15, 3087 - 3097. 74. Garside J. (1991) Tailoring crystal products in precipitation processes and the role of mixing. In Particle Design Via Crystallization. AIChE Symposium Series. 284:87, 16-25. 75. Gahn C , Krey J. and Mersmann A. (1996) The effect of impact energy and the shape of crystals on their attrition rate. Journal of Crystal Growth. 166:1-4, 1058 -1063. 76. Takiyama H. , Yamauchi H . and Matsuoka M . (1997) Effects of seeding on start-up operation of a continuous crystallizer. In Separation and Purification by Crystallization. ACS Symposium Series. 667, 172 -186. .11. Sullivan T., Liu Y . . and Ecke R. (1996) Turbulent solutal convection and surface patterning in solid dissolution. Physical Review E. 54:1, 486 - 495. 78. Hanson W. (1982) Handbook of Dissolution Testing. Pharmaceutical Technology Publications, Oregon. 79. Garside J. (1991) The role of transport processes in crystallization. In Advances in Industrial Crystallization. Butterworth-Heinemann Ltd., London, 92 -104. 80. Wilcox W. (1993) Transport phenomena in crystal growth from solution. Progress in Crystal Growth and Characterization of Materials. 26,153 - 194. 81. Benefield L., Weand B. and Judkins J. (1982) Process Chemistry for Water and Wastewater Treatment. Prentice-Hall, New Jersey. 82. Walton A. (1967) The Formation and Properties of Precipitates. John Wiley & Sons, Inc., New York. 83. Sohnel O. and Garside J. (1992) Precipitation: Basic Principles and Industrial Applications. Butterworth-Heinemann Ltd., London, 173 - 177. 84. Sohnel O. and Garside J. (1992) Precipitation: Basic Principles and Industrial Applications. Butterworth-Heinemann Ltd., London, 149 -169. 97 R E F E R E N C E S 85. Burt H . (1980) Effect of Crystal Anisotropy and Crystal Defects on Dissolution. Ph.D. Thesis, Faculty of Pharmaceutical Sciences, University of British Columbia, Vancouver, B.C., Canada. 86. Butler J. (1964) Solubility and pH calculations: The Mathematics of the Simplest Ionic Equilibria. Addison-Wesley Series in the Principles Of Chemistry, Addison-Wesley Inc., Massachusetts. 87. Downey J., Nickel J., Clapham L. and McLean R. (1992) In-vitro inhibition of struvite crystal growth by acetohydroxamic acid. British Journal of Urology. 70:4, 355 - 359. 88. McLean R. and others (1991) Pyrophosphate inhibition of Proteus Mirabilis induced struvite crystallization in vitro. Clinica Chimica Acta. 200:2-3, 107-117. 89. Ayati M . and Madsen H. (2000) Crystallization of some heavy metal phosphates alone and in the presence of calcium ion. Journal of Crystal Growth. 208:1-4, 579 - 591. 90. Salem M . , Mangood A. and Hamdona S. (1994) Dissolution of calcite crystals in the presence of some metal ions. Journal of Materials Science. 29:24, 6463 - 6467. 91. Freche M . , Rouquet N . , Koutsoukos P. and Lacout J. (1992). Effect of humic compounds on the crystal growth of dicalcium phosphate dihydrate. Agrochimica. 36:6, 500-510. 92. Wierzbicki A . and others (1997) Crystal Growth and Molecular Modeling Studies of Inhibition of Struvite by Phosphocitrate. Calcified Tissue International. 61:3, 216-222. 93. Clapham L. and others (1990). The influence of bacteria on struvite crystal habit and its importance in urinary stone formation. Journal of Crystal Growth. 104:2, 475 - 484. 94. McLean R., Lawrence J., Korber D. and Caldwell D. (1991) Proteus Mirabilis biofilm protection against struvite crystal dissolution and its implications in struvite urolithiasis. Journal of Urology. 146:4,1138 - 1142. 98 REFERENCES 95. Dumanski A . and others (1994) Unique ability of the Proteus Mirabilis capsule to enhance mineral growth in infectious urinary calculi. Infection and Immunity. 62:7, 2998 - 3003. 96. McLean R., Fortin D. and Brown D. (1996) Microbial metal-binding mechanisms and their relation to nuclear waste disposal. Canadian Journal of Microbiology. 42:4, 392 - 400. 97. Rivadeynera M . , Perez-Garcia I. And Ramoscormenzana A. (1992) Influence of the ammonium ion on bacterial struvite production. Geomicrobiology Journal. 10:2, 125-137. 99 A P P E N D I X A APPENDIX A Calculations For Recommended Reaction Time In The Crystallization Vessel (Pertaining to Table 3.3: Guidelines for pH control for struvite crystallization Source: Reference 13) Given: Height of the reaction zone = H = 3.5 m Inner diameter of the reaction zone = D = 60 cm Upflow velocity in the reaction zone = 24 m " hr"1 to 78 m ' hr"1 Then, Volume of the cylindrical reaction zone = V = T I X D 2 X H 4 V = 7rX(0.6)2X3.5 V = 0.9896 i n 3 Circulation flow rate =• Q = area of the reaction zone X upflow velocity in the reaction zone Q = n X (0.6)2 X 24 to n X (0.6)2 X 78 4 4 Q = 6.79 cu. m • hr"1 to 22.05 cu. m • hr"1 Recommended time in the reaction vessel = T = V Q T = 0.9896 to 0.9896 22.05 6.79 T = 2.7 min to 8.76 min (i.e. ~ 3 min to 10 min) 100 A P P E N D I X B A P P E N D I X B Calculations For Upflow Velocities In The Reactor Limbs; Corresponding To A Constant Flow Rate Of 400 mL ' min"1 (Pertaining to Table 4.1) Upflow Velocity = flow rate cross-sectional area Table B-l Reactor Nominal diameter (cm) Cross-sectional area Upflow velocity limb (cm"2) (cm' min"1) A 1.54 1.84 217 B 2.06 3.33 120 C 2.62 5.39 74 D 7.81 46.88 8 101 A P P E N D I X C APPENDIX C Calculations For Fluid Reynolds Numbers In Reactor Limbs; Corresponding To A Constant Flow Rate Of 400 mL min"1 And An Ambient Temperature Of 20 + 5 °C (Pertaining to Table 4.1) If: p F = Mass density of the fluid V F = Average velocity of the fluid D = Diameter of the limb in question u,p = Viscosity of the fluid Reynolds number = PF X V F X D ^F As shown below, p F and \xf are governed by the temperature of the fluid (in this case, tap water)(a). Table C-l Temperature (°C) PF (kg m 3 ) HFX103 (N-s m - 2) 15 999.1 1.139 25 997.0 0.890 Then, the corresponding Reynolds numbers may be presented as below. Table C-2 Reactor limb D X 103 (m) U X 102 (m' sml) Reynolds number at 15 °C 25 °C A 15.4 3.6 477 609 B 20.6 2.0 369 471 C 26.2 1.2 282 360 D 78.1 0.1 96 123 From Metcalf and Eddy, Inc. (1991) Wastewater Engineering Treatment Disposal and Reuse. McGraw-Hill Series in Water Resources and Environmental Engineering, 1253. 102 A P P E N D I X D APPENDIX D Seeding Technique Used For The Crystallization Reactor: Size And Quantity (Pertaining to Section 4.2.4) Table D - l Run Weight of seed (g) added for the given seed size identifier 0.25 mm to 0.5 mm 0.5 mm to 1.00 mm 1.00 mm to 2.00 mm Total weight of seed added at start-up A - 3.88 3.11 6.99 B - 3.88 3.11 6.99 C - 1.94 2.40 4.34 D - 1.94 2.40 4.34 E - 3.88 3.88 F - 3.88 3.11 6.99 G - 1.94 3.11 5.05 H - 1.94 2.40 4.34 I - 3.88 3.88 J - 3.88 3.88 K - 1.94 3.11 5.05 L - 3.88 3.11 6.99 M - 3.88 3.88 N 8.44 4.80 13.24 O - 3.88 3.88 P - 3.88 3.88 Q - 3.88 3.88 R - 3.88 3.88 s - 3.88 3.88 T - 3.88 3.88 U - 3.88 3.11 6.99 103 APPENDIX E APPENDIX E Instrument Operational Parameters (Pertaining to Section 4.2.5) Instrument operational parameters for the flame atomic absorption spectrophotometer Table E - l Element analyzed: Magnesium Copper Concentration units: mg ' L" 1 mg • L" 1 Instrument mode: Absorbance Absorbance Wavelength: 285.2 nm 327.4 nm Flame type: N 2 0 / C 2 H 2 N 2 0 / C 2 H 2 Range: O m g - L ' 1 t o l O O m g L " ' 0 mg • L"' to 20 mg • L ' 1 Measurement mode: Integrate Integrate Lamp current: 4.0 mA 8.0 mA Replicates standard: 3 3 Replicates sample: 3 3 Calibration algorithm: New rational New rational Instrument operational parameters for the LaChat QuikChem flow injection analysis instrument (A) Element analyzed: Phosphorus as orthophosphate (PO4-P) Concentration units: mg ' L" 1 Range: 0 mg ' L" 1 to 100 mg • L" 1 Temperature: 63 °C Principle^: The orthophosphate ion reacts with ammonium molybdate and antimony potassium tartarate under acidic conditions, to form a complex. This complex is reduced with ascorbic acid to form a blue complex that absorbs light at 880nm. ^ From LaChat Instruments Methods Manual for the QuikChem® Automated Ion Analyzer (1990) QuikChem method number!0-115-01-1Z. 104 APPENDIX E The absorbance is proportional to the concentration of orthophosphate in the sample. (B) Element analyzed: Ammonia nitrogen (NH4-N) Concentration units: mg' L" 1 Range: 0 mg ' L" 1 to 100 mg • L" 1 Temperature: 64 °C Principle^: Alkaline phenol and hypochlorite react with ammonia to form indophenol blue that is proportional to the ammonia concentration. The blue colour formed is intensified with sodium nitroprusside. w APHA, AWWA and WPCF (1995) Part 4500-NH3 - F. Phenate method. In Standard Methods for the Examination of Water and Wastewater, 19th edition. American Public Health Association, Washington, D.C. 105 A P P E N D I X F APPENDIX F Size Distributions By Weight Of The Sieved Harvested Crystals (Pertaining to Section 4.2.6) Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 A 24.42 2.0 7.17 1.0 5.24 0.5 4.23 0.25 6.65 0.125 11.97 Subtract seed contribution 6.99 Sum 52.69 Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 D 0.00 2.0 0.29 1.0 3.35 0.5 1.85 0.25 1.58 0.125 0.49 Fines 20.20 Subtract seed contribution 4.34 Sum 3.22 2.362 B 2.0 54.06 1.0 16.04 0.5 10.60 0.25 25.74 0.125 22.00 Subtract seed contribution 6.99 Sum 121.45 2.362 E 2.0 2.68 1.0 1.83 0.5 6.85 0.25 17.31 0.125 10.28 Subtract seed contribution 3.88 Sum 35.08 2.362 C 0.22 2.0 0.26 1.0 1.15 0.5 0.98 0.25 5.57 0.125 1.86 Fines 17.43 Subtract seed contribution 4.34 Sum 23.14 2.362 F 2.0 1.90 1.0 36.78 0.5 7.00 0.25 2.95 0.125 5.91 Subtract seed contribution 6.99 Sum 47.55 106 A P P E N D I X F Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 G 48.30 2.0 8.49 1.0 7.21 0.5 2.94 0.25 4.34 0.125 4.09 Subtract seed contribution 5.05 Sum 75.37 2.362 H 24.94 2.0 6.37 1.0 1.99 0.5 4.20 0.25 5.46 0.125 10.25 Fines 23.83 Subtract seed contribution 4.34 Sum 72.70 2.362 I 1.87 2.0 0.75 1.0 4.90 0.5 4.73 0.25 5.24 0.125 4.40 Fines 16.53 Subtract seed contribution 3.88 Sum 34.54 Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 J -2.0 3.21 1.0 2.26 0.5 8.70 0.25 19.36 0.125 9.43 Subtract seed contribution 3.88 Sum 39.08 2.362 K 2.68 2.0 0.13 1.0 1.76 0.5 4.54 0.25 2.68 0.125 4.93 Subtract seed contribution 5.05 Sum 11.67 2.362 L 2.09 2.0 1.30 1.0 23.85 0.5 2.83 0.25 1.42 0.125 3.29 Fines 6.37 Subtract seed contribution 6.99 Sum 34.17 107 A P P E N D I X F Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 M 20.30 2.0 4.86 1.0 7.32 0.5 2.96 0.25 3.57 0.125 4.03 Fines 12.62 Subtract seed contribution 3.88 Sum 51.78 2.362 N 12.39 2.0 3.18 1.0 19.95 0.5 27.21 0.25 15.82 0.125 10.50 Fines 5.43 Subtract seed contribution 13.24 Sum 81.24 2.362 O 0.13 2.0 1.77 1.0 6.49 0.5 5.45 0.25 5.17 0.125 3.19 Fines 15.00 Subtract seed contribution 3.88 Sum 33.31 Sieve size (mm) Run Weight of dry struvite crystals (g) 2.362 P 26.86 2.0 3.25 1.0 2.41 0.5 6.63 0.25 10.73 0.125 7.40 Fines 16.83 Subtract seed contribution 3.88 Sum 70.23 2.362 Q 6.03 2.0 2.07 1.0 10.78 0.5 15.55 0.25 13.85 0.125 6.13 Fines 7.68 Subtract seed contribution 3.88 Sum 58.20 2.362 R 16.53 2.0 2.51 1.0 4.51 0.5 8.20 0.25 4.79 0.125 3.25 Fines 6.60 Subtract seed contribution 3.88 Sum 42.50 108 A P P E N D I X F Note: (1) Monitoring of fines was started once the significance of pHijm on the operational control of the system was suspected. (2) Collected size weight distributions for runs S, T and U are not presented here. Unlike the other runs, dissolution of seed was apparent in runs S, T and U , thus implying that the rate of crystal growth was discontinuous. Presenting the crystal size weight distributions for such cases, would be misleading. 109 A P P E N D I X G APPENDIX G Verified Composition Of The Harvested Crystals (Pertaining to Section 4.2.7) If MW = Molecular weightw, MWMgNH4Po4.6H 2o = M W M g + MW N + 4 X MW H + MW P + 4 X M W 0 + 12 X MW H + 6 X M W 0 = 24+ 14+ 4 X 1 +31 + 4X 16 + 12 X 1 +6X 16 = 245 If 0.25 g of 100% pure struvite were to be dissolved in 1000 mL of deionized water and acid, then: Mg = 0.25 X 24 X 1000 = 24.5 mg' L"1 245 N = 0.25 X 14 X 1000 = 14.3 mg-L"1 245 P =0.25 X31 X 1000 = 31.6 mg-L"1 245 ( d ) From Sawyer C , McCarty P. and Parkin G. (1994) Chemistry for Environmental Engineering. McGraw-Hill Series in Water Resources and Environmental Engineering, New York. 110 A P P E N D I X G For a volume of 250 mL, Mg = 24.5 X 1000 = 24.5 X 4 = 98 mg ' L"1 250 Similarly, N = 14.3 X4 = 57.2 mg-L"1 P =31.6X4= 126.4 mg-L"1 111 a x a w OH OH < DC ••5 "S < -9 ••s "O ^ s S « s -= = 1 s s > Cu <3 " ° i 5 « d o 5 «»- 2 3 a s »> s £ .5 5 S 2 112 A P P E N D I X G Calculated molar ratios Sample calculations - Run A: N:P molar ratio = 120.6 X 31 =1.01 14 264.4 Mg:P molar ratio = 190.1 X 31 = 0.93 24 264.4 Table G-2: Run Crystal Molar ratios (calculated from analytical results) Identifier size (mm) N:P Mg:P A 2.362 1.01 0.93 B 2.0 1.02 1.02 C 0.25 1.07 1.03 D 1.0 1.02 0.99 E 2.0 1.05 1.04 F 2.0 1.02 1.01 G Fines 0.93 1.02 H 2.362 1.03 1.04 I 0.25 0.97 1.03 J 1.0 1.03 1.03 K 0.5 1.04 1.02 L 0.125 0.97 1.02 M 1.0 0.99 1.04 N 0.125 0.96 1.03 0 2.0 1.06 1.02 P 2.0 1.05 1.03 Q Fines 0.99 0.99 R 2.362 1.01 1.02 S Fines 0.98 0.93 T Fines 0.99 1.06 U 1.0 0.99 1.03 V 1.0 1.02 1.04 113 A P P E N D I X H APPENDIX H Daily Record For All Runs The data for the runs has been presented in the order of citation in the body of the report. Since the number of columns is too great to be included in one page, the information has been sub-divided into two parts - Appendix H-I and Appendix H-IJ. The former part contains measured data such as the concentrations of the constituent ions, temperature and the pH. The latter part contains calculations for the data inside the reactor and the subsequent calculations for the pP s values. In Appendix H-II, the pP s values, which have been used in plotting the relevant figures / graphs in the body of the report, have been underlined for the purpose of referencing. The average values for each run have been used in all cases, excepting runs with low Pinfiuem concentrations (20 mg ' L ' 1 - 30 mg "L" 1), since some seed dissolution was apparent in these runs, at an operational pH value below 9.0. Thus, for runs with low Pjnfluent concentrations (runs S, T and U), the plotted values of pP s are the values recorded / calculated at the pH when seed growth was first observed; i.e. at a pH of 9.0. The sign " - " implies that the injection port / reactor was plugged on that particular day and hence, no effluent sample could be taken. The acronym "S.D." stands for standard deviation. Although none of the underlying distributions for run data were perfectly normal distributions^, the average and the standard deviation for all runs were calculated using the formulae for a perfectly normal distributional Such a digression was not entirely unreasonable^, since these values are only meant to serve (e) From Perkins J. (1997) Modem Industrial Hygiene: Recognition and Evaluation of Chemical Agents - Volume 1. Van Nostrand Reinhold Publishing Inc., New York, 147-164. ( f ) From Statistics Every Writer Should Know (1996) Standard Deviation. Retrieved November 19, 2001, from the World Wide Web: http://nilesonline.com/stats/stdev.shtml 114 APPENDIX H as comparisons between runs with similar characteristics - similar P, N and Mg values, but differing modes of pH control or differing seeding techniques; for example, runs M and N (Table 5.1), runs O, M and P (Table 5.6), runs Q and R (Table 5.6), runs H and C (Table 5.6) and so on. In fact, this digression became necessary since there was no other easier and condensed manner in which to depict the comparisons referred to. Sample calculations for Appendix H-II Run A : Pump head configuration / flow-rate ratio (P: Mg + N : NaOH : Recycle) = 33:10:10:33 PO4-P: Calculations for conditions inside the reactor (mg' L" 1): If the contribution of PO4-P from the feed tank = A , then A = PO4-P from the feed tank (g) X flow-rate ratio of the PQ 4-P line total flow-rate ratio A = 80.3 X 33 33 + 10+10 + 33 A = 30.8 mg L"1 If the contribution of PO4-P from the recycle line = B, then B = PO4-P in the filtered effluent sample from the clarifier ( s ) X flow-rate ratio of the recycle line total flow-rate ratio Values provided in Appendix H-I. 115 A P P E N D I X H B = 34.9 X 33 33 + 10 + 10 + 33 B = 13.4 mg L" 1 Therefore, Total PO4-P inside the reactor = A + B, where A + B = 30.8 + 13.4 = 44.2 mg L" 1 Similar logic applies to the columns referring to the Mg and N H 4 - N conditions inside the reactor. Calculations for pP,: If M W = Molecular weight, Ps-inf ( h ) ~Pinfluent X Hnjluem_X Mginf|Uent u M W P M W N M W M g = 61.6 X 76.3 X 15.9 31000 14000 24000 = 7.22X 10"9 (moles-U 1) 3 ^ Values refer to flow-apportioned influent samples at the injection port, without NaOH dilution; values provided in Appendix H-I. 116 A P P E N D I X H PPs-inf = - lOgio(Ps-inf) = -log 1 0(7.22XlO- 9) = 8.14 P s . m i ( i ) = Total PO4-P X Total N H 4 - N X Total M g M W P M W N M W M g = 44.2 X 60.3 X 10.7 31000 14000 24000 Ps-mi = 2.73 X 10"9 (moles 1 L" 1 ) 3 p P s . m | = - logio(Ps-ml) = -log 1 0(2.73XlO- 9) = 8.56 Ps-eff - Peffluent X Neffluent X Mgeffiuent M W P M W N M W M g = 34.9 X 57.6 X 6.9 31000 14000 24000 = 1.34 X IO - 9 (moles-L" 1) 3 pPs-eff = " logio(Ps-eff) = -iog 1 0 (1.34XlO- 9 ) = 8.87 Values refer to total values inside the reactor; values provided in Appendix H-II. Values refer to filtered effluent samples from the clarifier; values provided in Appendix H-I. 117 X X Q z w DH OH < o a O £ B E S B B a ^ Z as E O 1 ' J 4> 3 O O a o z + s a . •s o c 33 O z + OD o c - a, s o z + o. o u a. >> c u c «J s Ci BL. £ a. as OH I s 3 9 t-Q-< a 00 9 D . <: <: o B 3 OS 9 3 Q t/5 X X Q •z, w 0-< • i a O. o a • « E E 2 O a. on E E " a = o 2 « 3 Z c ~ 4> *3 I "5 •8 §* ^ 3 O OS oo \0 f ; ON NO NO NO TT (N oo NO N© <N oo OO oo 00 o © ro r~; ro fN in ro ro ON fN OO in ro in LO ro •<? © fN p ro ro ro fO fN r-- r-- r» r-ro ro ro fO 1.30 1.30 1.30 1.30 oo oo oo 00 ON ON ON ON ro ND ro NO ro NO ro NO . 45.1 45.1 45.1 45.1 in in tn ON ON ON ON ON ON CN fN 90.1 90.1 90.1 o o fN rs o fN 1-01 9 1-01 age 15-Ju 16-Ju 17-Ju Aver S.E u c «j 3 a: o. e 3 ai o z + Of o 2 «5 a. I 3 OS 9 9 o-< 119 X X Q z w cu 2 £ o. • n E M 1 o n a* .a — ai u « 1 I H 13 " a B O a « 3 z c Si. I 1 0> Cu B a s Oi a o z + S PH Oi a. I 120 X E S? Q Z w OH OH < CL o a • t S £ E E £ E E « w " X B O •8 I i * r t o o a. a a o z + OX s CL, K O z + o 00 o BC O z + CI' s o c o . E Oi o-l D 3 a t/5 a «5 121 x x Q 2 w cu cu < X a a O Si —1 E S -Si J= "5 *-i- « I J 8 J I 1" « w £ X B O 3 ™ I i w c <U 3 Ci a. 3 Oi 3 es p OO 00 in N O ro ro p N O O N C O fN O N O N fN fN fN O fN fN fN ro o O N O N O N O N O N 00 00 00 C C 00 C C 00 rS r> A t>- rS •^r oo in 00 N O fN fN C O ro fN ro ro fN cO © ro © ro N O ro rO rO rO ro rO ro rO ro ro O N (N C O in in (--• »n ro r-- in oo O N O N O N N O O N fN O N N O O N O N O N in O N in O N in O N r^' O N ro v~~ N O N O 00 r~- N O ro ro ro CO ro ro fN ro o ro ro ro ro N O fN ro fN ro O N oo O N ro ro fN ro ro ro ro ro fN OO O N fN N O N O in N O N O N O N O N O N O N O N O N O O in ro O fN m O rj- ^ ro ro o ro m ro O p »n OO ro r-- OO N O ro CN fN N O ro in" ro ro rf fN O fN <n •o p r~- OO O N NO N O co od ro ro •n" fN fN ro N O N O fN N O fN O N m o o o O N ro O N ro © O N ro © O N ro o O N ro O N ro oo O N N O -3- fN N O ro O N oo fN O N ro r-© r- oo 00 ro 00 fN C C o 00 fN 00 00 r-; o 00 CN o ro m OO OO fN fN (N ro fN fN fN N O fN fN fN N O fN CN ro fN ro fN fN in fN fN fN fN fN fN fN fN fN fN r-^  fN fN fN fN ro O N in fN 00 O N OO 00 O oo O N r- O N r- © 00 00 r-© 00 O N r- O 00 00 r~ oo r-o CN fS fN o fN fN 00 00 00 00 O N oo ON Aug-Ot o D. op O a. u op 9 a. u op 9 a. op o a op 9 ex cj op 9 o. CJ 00 9 o-OJ 00 o CL. CD 00 o 6-00 cp C L OJ op co o fN O ro O -4 o in O N O o o oo © O N o o a 122 X X Q z P J CM a. • £ a a E S E E £ « I, O E -sH E E o o 2 * t r o o, c a. a. . E £ "2 = i — C K: C 'ing 'ing isolving ;soh isoh isolving N O fN ro ro in fN in CN 1 sip dis dis N O N O O N P O N oo oo 00 od r-^  r-; N O in fN N O m N O rO ro ro ro r--fN ro fN ro m ro ro ro oq © N O ro in ro O N N O O N O N CN O N m O N O N O N N O O N P O fN — fN ro ro ro ro ro O N fN N O fN ro ro O ro O ro OO N O r—• O N OO in O N O N oo r~- oo oo o r-fN in" wS in in" in N O N O N O N O CN N O fN o fN oo r-fN CN ro r» ro ro OO fN (N ro r-^  oq N O 00 ro O N ro r*-ro N O ro O N ro r> ro V~~ Tf in ro CN oo oo oo ro O N O o N O o N O o N O o fN O r-- O N p 39.6 O N in fN O N ro o o 39.6 O N ro o o © O N ro in N O in <n m fN N O fN N O t-- O N r- r- CN r-» oo r-- in r-~ oo oo ro 00 ro 00 ro O N O N N O in N O N O fN fN fN fN fN ro in —• CN CN fN fN CN fN fN fN fN fN O N O N ro O N O N | - » OO o oo O N r- O N r-- O oo o" 00 o 00 od r--fN fN fN fN fN fN CN fN fN o fN fN fN CN CN 2 9 9 2 9 10-! 9 9 ou 9 bO 3 3 3 3 3 < -Au fN fN fN CN fN ro ro O fN O p fN *n ro O N — p O N N O O N oo r-- 00 O N r- O m fN ro ro CN CO CO N O CN O N O N O N O N oo p O N O N O N od od od oo p p ro oo in in «n fN ro ro rO ro rO ro ro t~-fN O N CN fN ro o ro fN ro O N O N O TT, ro CN m in O N 3 ro in © »n O N ' ' O N O N O N oo 00 O N O N O N N O O N fN m oo p O N , fN ro ro ro ro fN ro fN ro fN ro o fN ro CN N O fN N O fN CN oo fN r-- r- OO CN fN OO O N OO O N in t-- OO CN fN fN O N OO N O N O N O N O N O in in in N O N O in in o o OO fN r- r~- r- OO fN ro ro N O ro fN fN fN fN ro fN fN in CN ro ro m in N O m m O N ro O N ro o ro 00 ro ro o O N CN O N O N „ fN N O CN O N O N t— O r— o r— o o O N O N O o o o O p in fN O p ro O N p N O ro N" O N ro O o o oo ro O N ro O N ro O N ro ro ro O N fN ro in in O N ro O N m CN 00 fN 00 O fN OO oo r-- oo r- 00 N O OO N O 00 00 in in OO ro f-; ro fN O N O N ro fN in in o O N CN fN CN CN fN fN fN fN fN CN CN CN O O O O N O N in p in CN OO CN OO O N r- O OO O N O N fN oo oo r- O N r- od r-fN ro o o o o fN CN fN fN CN fN fN fN fN fN CN CN CN 9 9 9 9 p O 9 9 9 9 9 oo oa GO 00 00 00 00 00 00 00 00 < < < < < "1 < < < <, <! 00 O N o CN fN fN fN ro fN fN m fN N O CN CN 00 CN 123 X X Q z w OH < 2 S. a O a. -s: E E £ I .1 2 a = o 2 * r t ^ 3 OS a. a 3 2 « a o CD s z + 0 . S o ex C c 3 22 o< 124 X X o — I Q Z tu OH CL. < 2 £ a O 2 J £ E E £ £ E Z a B O «j ra = 5 a o z + a o z + EH o c OH S-I H B 3 125 X X Q z a a. OH < c O i i J £ E E S S -5" 1 .2 E E « w <» "B p O I 5 2 S t r o o, o. o. a 60 •a s S o z + s w O o 3 ei a eo 126 X Q a OH OH < « : o ""C H JL a 5 z -_ < n o +3 re u. O o >. o 0) 0. o o 01 I'-ll o o OH s CH IT) N O in in m m 00 C C od od od od od ool l/-) »n «fr IT/1 00 od od od od od od ool O N O N O N O N O N O N O N ONI r> oo oq p 00 N O od © © © —4 © cs CN ro ro ro ro C O CO ro CN N O ro O N od O N © © © —« CN CN CN CN ro ro ro CN CN CN O N O N O N O N O N O N O N od o CN — in — N O m oq O CN oi ro C O ro ro CN 1- TT o ^— o T— II a> o O r-; — — o >» O O N © —'. © O N rec ro CN ro ro ro ro CN X o re z p O O O CN CN CN CN CN CN CN + £ Q. .2 ro in oo iri p CN V~~ If) £ in iri iri in N O vi n CN CN CN CN CN CN CN CN i o •—. c o RI L- CN in CN p TT_ CN 00 Ifl 3 N O N O N O od N O od r> c o o ~c ro ro ro ro o O o O N p a O N O N O N O N O N O N 00 ON E 3 a. 4* p 9 9 p 9 9 9 n 3 3 3 3 3 oo O N o JL, CN ro CN CN CN CN CN 127 a x Q Z w OH OH < fc f a. PH C PH a Li a 5 o 8 -B « ^ o ta. •- ea E w i 3 c o -3 ra i_ 3 «- O)] ^ c " o o o IS " ° 2 ra a> a> x: & E' CU 3 DC 0. a s 3 SH pi ro <n O N | oo oo oql co »n 001 od 00 O N O N M r-^  fN CN ro fN ro © m fN fN fN fN p p O o o O in N O ro oq t> co —- © m p fN O N in m OO fN CN fN CN N O fN oq IT) 00 fN in i/i ro fN fN \D fN oq VJ O N ro fN N O N O O O p o O N O N O N O N 9 9 9 o &x a ~3 en m N O r» > 3 • g o o !* ra a> " a o E <J> 3 o. a. B 3 II o .3 2 0) o >» o 0) a. + I" CL g -3 2 <u 2 3 o CM II O o >» o o a. o B 3 OH fN CN ro ro <n ro 00 OO oo od 00 cc 00 00 00 fN fN fN fN fN ro fN CO fN CO 00 CC 00 00 CC 00 CC 00 oq oq oq oq oq O N 00 00 oq r> rS fN fN fN O N N O NO O N oq © O N CO in O N od ro ro ro ro ro CO f i OO <n r- in O N fN © O N 00 N O O 00 fN fN fN fN O N O p O N 00 O N N O O N © 00 O N N O O N O N fN to u> OO N O r> —« fN N O oq *n o © o O N o o 00 00 ON* in in in in ^ o T — ro CO II o o fN p fN N O r- ro O N 00 00 in N O in £ fN CN fN fN fN fN fN fN o ra z y O N N O p ro ro N O O N O CN CN fN •—< CN fN + fN fN ! fN fN fN fN fN S a. ~~~ o ra k_ O N 00 N O o in 00 00 a ro fN fN fN o d O N fN re ro ro ro ro ro ro CO fN CO w 3 o H-c o •3 s ro O N •^r oq — CN OO r> in in in ro ro vi O) c o u T3 ro fN p ro N O in O CO oq Q. N O N O N O N O N O N O N O N O NO E a. 9 9 © 9 9 9 9 9 a. 00 a c c c a c £ a 3 3 =3 3 3 3 3 3 3 »—> >. © fN ro in N O 128 X X Q z w OH OH < fc a ox -5 ? « : o •< H JL - IB B < a 5 2 SH O ea CM II O o a: 3 ro CN CO ro CO) 00 od od 00 od 00 ool CN CN CO CN ro ro CNl od od 00 00 od od col oo oq 00 oo oo oo 001 r> in CN in o m Ifi • NO in in" CO CO ro ro ro ro CO NO CN ON ON O ON ON NO ON CN NO ro in rS (O ro NO NO NO NO CN O © o ON © ON © © in in in in ID o T— CO II a> o oo — ON — i g> • od t< od £ CN CN CN CN CN CN X o ra z ^ ro ON NO ON 1 CN CN CN CN CN ro + CN CN CN CN CN CN s a o at ON NO CN ON ON Tf o 1 in CO ro © —' CO -rat CO CO CO CO CO CO CO ow —~ c o !—* 2 m o in © ro CN CO 1 00 ON NO in NO gi nf o o •a ra a .c NO ND o OO r> NO Q. 1 NO NO NO N  NO E 3 0. p O 9 9 9 9 p <u b£> rt 3 3 3 3 3 u *~? f—< CN CO in NO rA >• j*« O O o o O O o <4 II £ l o o X O ra I'-ll o O c O 3 CE 0. ro ro ro NO NO in od od od od od od od ro ro CN CO od od od 00 od od 00 NO NO NO NO NO NO r> r> ON ON ro oq ND od NO CO NO NO NO NO od NO "3- CN CN NO in NO in" r-- ON OO © CN © CN © " CN CN CN CN CO oq p CN p ON ON ON "3" ON NO NO ON ND oq NO CO ro in ro rO ro CN ro NO in" ON O NO O co NO od CO CN T—1 10.9 11.0 12.1 NO ND p od ro ON m ON p ON ON ON O NO in NO in in in NO O p 9 9 9 p erage llil 3 3 3 3 erage ON O © CN ro •4 Av. 2 o >. o o 0. o o ° ra + ra 5 O —^• re o 3 U) —I c o o •a ra a E 3 a. 129 X X Q z w cu cu < o < H JJJ i a. x 5 z -_ < E Si o O N O N CO <n © fN 00 od od 00 O N O N fN O N oo 00 rS od N O N O N O N O N O N O r-^  N O fN O N oq oq —- —t in fN O N in • ^ CO CO CO co m O N oo rs fN N O CO © N O • fN fN fN O N O N O N O N O N O N • to fN N O r - 00 © co fN 00 CN O N • v- N O N O in »n N O m o C O ro II 0) o r-; O — — OO co >» vi iri © fN 0) co co CO CO CO CO X O ro Z N O N O N O N O N O N O z r^  + fN fN fN fN fN CN s o oq N O CO N O 0) © cc © fN < rat in CO o c o ra »n CO — od N O od in in © fN fN — — fN fN c o o TJ ro CD CO CO CO CO CO CO Q_ fN fN fN CN fN CN • £ fN fN fN fN fN fN 0_ © © ©> 9 9 9 9 C C C c c C c 3 3 3 3 3 3 3 cn m N O r- OO O © © © © © O CO CO II o >v O CD O ra Z + at g ra 3 o c o 3 Ul 13 C o o a o >l o o oc s 3 0. o u >. o a oc. II o > o a> O re Z Z + g re 3 OI •3 C O o •o re o ^ a E 3 a. col 130 33 X Q Z PJ eu cu < j CL a a. a oo -* ? A Z © < H JL o 1 La LL. a 5 z -_ < E £ l E S , O *- i II o a o o 0. Z c 3 p p O p p p p CN pi od od od od od 00 od 00 00 00 od ccl CN p p ON p p p p pi od 00 od 00 od od od 00 od ool r-- 00 oo r- t-» oo CO oq r-» M iS t-^  iS M CN fN p rt o ON CN as NO rt 1 O rt rt rt rt CN CN CO ro m ro ro ro ro rO CO ro m ro CO NO m O ON NO oq in in ON rt CN ON ON od ON ON ON ON od NO p p in rt NO CN in as NO in ro CN in in rt rt rt rt rt o o •"1 rt in O oq in NO ro m NO r» o NO NO 1 in CN in CO in rt rt rt ro Tt —^ ON ON ON ON ON ON ON ON ON ON ON ON o ro o CO il 0) u — rn oq rt ON NO m ro — CN rt >* od od 1 NO rt NO in NO NO NO tn rec in m in in in in in in in in »n z o ra Z Ol OO ON p oq r-^  in rt CN m 00 rt od od 1 od od od od od co CO ro m ro ro ro ro CO ro m Z + o at L_ oq oq rt oq ro O r-; ON oq oo oo CN rO m I CN CN ro CN CN CN CN ON" fN 2 m ro ro CO ro ro ro ro ro CN CN CO ow> >*— — c o •3 C3 i_ NO r- CO r-; CN ON NO l> NO l> NO p 3 ON ON od od ON t> od od 00 in ro GO Ol •c: c o o •a ra <v m ro p p ro 00 CN -^ —- ON OO Q. rt rt rt rt rt ro rt rt rt ro ro rf E a. 9 9 9 p 9 9 9 9 9 9 O p Of CL a. O. a. a. a CL Cu a. a. « 3 <D « <U <u > <u CL> a> <D <o <u U < °? op op C/3 cfl 00 op op op op op ,—, CN rO rt in NO r- 00 ON o _1 m o o O O o o o O o o >• o II o O ra Z + S o •3 ra L_ o 2 i o c ol 3 Ol i£ c o o XI ra a> sz a E 3 a. O a 3 Oi 00 < X X Q z PJ OH DU < O < H JL 0> -*" as I s i 2 JL s O co S ^ 1 II o 2 a> o >. u a> BC a. s s OS p O N p O O CN CN CO CO od CO od 00 00 od 00 od odl O N O N p p p CN © p p i l> od od od 00 od CO od od 00 odl r- r- r- 00 oo oo oo oq oq CO 001 t-^  r-^  I-* r-* r-^  rH N O oq in r- O N 1/5 in in rn © CN © rn CN cn C O cn cn cn m m cn cn m m m cn cn CN p- in CN rn N O CO in cn © CN © m CN cn CN CM CN CN CN CN CN CN CN CN CN © N O r-O N O N O N 00 od O N O N 00 O N O N r-^  O N O N N O rn CN fN ro" d © _J CN CN Tt O N O N O N 00 O N O N O N O N O N O p N O N O O N © OO in CN r-" O N © N O rn N O 00 00 r- r-- N O t-- N O N O N O N O N O r- N O N O N O r~~- p © 00 rn r-; N O cn rn m" rn rn CN CN CN CN CN CN CN CN CN CN CN OO O 00 p CN r-- N O CN rn © CN m cn CN cn cn rn N O 00 00 O N © © cn m cn cn cn CN CN CN CN CN m cn O N O N o cn 00 cn © rn cn in cn O N O N © —-CN CN CN CN CN CN CN fN oq O N O N O N CN N O OO O N O N O N od 00 od od O N od od 00 od od 9 9 9 9 © 9 9 9 9 © 9 4> 00 00 00 00 00 oo 00 00 00 00 00 73 <l < < <! < < < < <, < *~ - CN CN CN CN CN (N CN CN CN ra II o o o T3 ra " a j? E 2 3 0. a. © © © © 00 O N O N 00 00 00 O N 00 N O N O N O N O N O N O N O N O CN O N r-- O N CN CN r-f-N O N O N O r~- 00 N O N O O N N O CN rn O N OO fN © fN © fN fN fN CN O N © CN O N O N rn CO 00 oq O N Tf © in N O N D fN m fN in 00 O N 00 fN CN O N O N O N OO 00 r~-00 O N OO O N OO 00 00 00 00 O N m oq O N © © N O r~~-in N O m r~-in in in in © N O N O r~- CN CN m m r-; m m f-m N O m O N CN m m O N cn in m N O cn oq CN rn O N CN O N N O © O N N O ON* N O O N N O 00 N O 00 N O CO N O O N N O oq O N t-; fN CN CN © CN © fN O N © CN O N m CO m rn O N © in © in OO © m m d m 9 9 9 9 © 9 9 9 9 00 DJQ 00 00 00 00 DO 60 < < < < <! < < < © OO © O N © © CN cn 4* in 132 a X Q Z w OH 0-< 1= X a eu a. 5 ? H JLI O B9 o a> DC O ra Z I" z + O « o 2 i o 3 Ol «C c o o •a re .c a E 3 a. p p p p p oo oo oq C O O N oq oo CO N O N O N O N O N O N O N O N O CN r t CN in rn CN p O O r-o r- O N N O CN r-N O m N O OO N O in fN CN N O r-; oo in © © r t © r t CN r t in m cn ro fN o O N N O r t in O N © ro O N CN O N fN in fN CN O N fN C N © oq r f O N o 00 O N © O N od oo O N O O O N O O N O O N © O N © O N r-» in m O N r t m N O r t m' CN CN m' O N ro fN r t r~- oq N D O N 0 0 o oo 0 0 O O o 0 0 O O r-O O p N O in in O N 00 ©' r- O N N O O N N O O N N O O N N O r— N D in N O od N O r t n-l CN r~~ r t N O r t r t r ? r t r t O N ro CO o r t O N N O O 0 0 O fN O N fN O N CN © m O m O N CN O N CN O N CN 00 C N 9 G O 9 0 0 ig-01 9 0 0 9 0 0 9 0 0 ig-01 ig-01 •age < < < < < <, < < > O ro II o o re Z + s o *3 re o 3 o> c o o T J ra o o o o O 3 0. 0. 9 >, 9 2 133 I X a. Is V aT a. 00 ON od O N od O N od O N od O N 00 OI ONI p O N O O N ON 00 00 00 00 O N 00 ONI OO oo OO oo OO 00 oo 00 in 00 ulations for 1 aT a 00 od ON 00 O N 00 O N od O N 00 O N 00 o| ON| O N 00 O N od o O N oq od oq od O N od odl r-; od r~; 00 od od »n od a U c sT a. 00 OO od oo od oo 00 oo 00 oo 00 odl OO 00 oo od O N od in od in od m od ml odl od CO od CO od ro 00 CO 00 eactor (mg' L"1) OD S "3 o H a + '< fN fN fN d fN d fN fN d fN fN d fN O N O N o d fN fN O N r-O N O N O N fN od p od cn N O N O tn fN O N CN CN »n" CN in CN r-^  od fN nditions inside the n V V E o u so O d od rn 00 00 od oo O N N O oq o d 00 O N O N O N cn r>; CN O ro ON CN in N O nditions inside the n Mg:Co rt E o La '< N O d ro d d d d d N O d p od rn 00 od O od in r- CN 00 o fN reactor (mg' L"1) z z "c3 o H + JL O O O o ro o o d O N ro o O N m in od m od m p od m o ro in m oo m m m CO CO d o CO CO O N od m 00 cn CN 00 m m 00 cn ro CO CO od CO od CO tn CN J= — M II ions inside t >> u a* L. E o b M aT o >> o £ p O N fN N O rn oo N O N O rn N O CN in oq in N O aT o >» u a> i_ ' X O N CN N O CN CN o d CN d CN ro d CN CN fN ro fN in CN •o s o y z 1 as z B a I o b. o ra z Ul S z + ro fN oq fN fN fN fN CN fN O N fN fN r--d fN N O fN CN fN fN fN fN O CQ z - z + a) E O N N O N O r-N O 00 CO O reactor (mg' L"1) 0. • o cu "3 © + a o E 0) ra ta. i o r-; 00 rS rn r> O 00 r- fN OO in m NO fN N O m rS CL O ra o ra i-5 o in rn CN CN rn CN CN CN cn d CN p CN p CN d fN fN d fN p in fN »ta. ditions inside thi >> u ai £ o £ II o c o 2 a) C o u r» rn r-' fN in rn O N in p N O N O N O c o H3 ra ^- Ul CN ^ ii s o o O N CN \>-CN oo d CN O N d d O d in ON CO a o U CL. 6 a. E o u. u. 2 a o >. o ai Ct TJ ra a E CL p d d N O d d d oo O N m d d N O d '•s "°*- a> o AZ E LL N O d d d CN d OO d N O d d d Date H B 3 OS 9 00 < ro 9 CL CO CO © 9 CL <D OO fN o o a. CU on m O 9 CL o CO •4 o 9 CL <L) 00 in O 9 CL <L> 00 N O o 9 CL 0) 00 r^-O 9 CL <L> CO OO O 9 CL <L> 00 O N O •5 B ZZ o c D O N o c 3 O fN 9 c 3 fN O c 3 CN fN cp c 3 CO CN cp c 3 CN Cp c 3 m CN Cp c 3 N O fN Cp £= 3 r-CN 134 A P P E N D I X I APPENDIX I Attempts To Measure The Velocity Of The Reaction (Pertaining to Section 5.6.3) Generally, a crystallization process reaction is considered to reach steady state within 10 system hydraulic retention times (k ). If: The total hydraulic residence time (HRT) of the crystallization system = T = V Q where T = hydraulic residence time in the reactor + hydraulic residence time in the clarifier = T, + T 2 = v L + v 1 Q = (2.5 L X 1000) + [7t X (0.5 X 9.9)2X 33] cm' 3 X(10' 2) 3X 1000X 1000 400 mL • min"1 = 12.6 min 10 system HRTs translate to approximately 2 hours. Accordingly, it was decided that the From Takiyama H, Yamauchi H. and Matsuoka M. (1997) Effects of seeding on start-up operation of a continuous crystallizer. In Separation and Purification by Crystallization. ACS Symposium Series. 667, 172-186. 135 APPENDIX I monitoring for the velocity of the reaction for a run, should not exceed 2 hours, following a run start-up. Also, for the measurements to be representative, it was important to start sample collection only after the recycle stream had started functioning. This would be the case, after at least 2 system HRTs has elapsed - one system HRT for the reactor and the clarifier to fill up with mother liquor, and the second system HRT for the recycle stream to complete its circulation through the entire system. Accordingly, the first sample was collected after 2 system HRTs had elapsed (Tables 1-1,1-2 and 1-3). In theory, while measuring the velocity of a reaction, it is assumed that the reaction is strictly irreversible1^. However, in reality, this is mostly not the case. Thus, it is becomes very important to exercise control on the reaction conditions used to collect the rate data, so that the assumption of irreversibility is valid. A leading book on water chemistry(1) states ". . . One way of doing this is to collect rate data only during a very short time period following the mixing of the reactants. Even i f the reaction is reversible, there will be little back reaction during this time period, because the products that participate in the back reaction, are present only in small quantities." Accordingly, it was decided to collect samples at short intervals (5 minutes - 10 minutes) and continue doing so for about 75 minutes (approximately 6 system HRTs) after run start-up. Based on previous experience, it was considered that the instrument would be able to record the rate of change in the PO4-P concentration most accurately, compared to the Mg and NH4-N concentrations (Paula Parkinson, Environmental Engineering Laboratory Technician, ® From Snoeyink V. and Jenkins D. (1980) Water Chemistry. John Wiley & Sons, New York, 34. 136 A P P E N D I X I Department of Civil Engineering, U B C , pers. comm.). Accordingly, effluent PO4-P concentrations were monitored first. From Tables 1-1, 1-2 and 1-3, it becomes evident, that the rate of change of effluent PO4-P concentration is nonetheless too small for the instrument to record accurately. This may be taken as an indication of the kinetics of the reaction being particularly slow at the bench-scale. Table 1-1: Run R, pH = 7.3 Sample taken at the end of _ minutes after 2 system HRTs had elapsed Recorded analytical P0 4 -P concentration (mg ' L"1) 15 75.4 20 74.9 30 74.6 40 74.8 50 75.5 Effluent P0 4 -P concentration^ at steady state for a pH of Table 1-2: Run N , pH = 7.8 Sample taken at the end of _ minutes after 2 system HRTs had elapsed Recorded analytical P0 4 -P concentration (mg ' L*') 0 36.5 5 36.3 10 35.4 20 36.1 30 35.9 40 35.7 7.3 = 68.9 + 2.6 mg ' L" 1 Effluent P0 4 -P concentration^ at steady state for a pH of 7.8 + 0.1 = 30.7 ± 3.1 mg ' L' Values taken from Appendix H-I. Effluent P04-P concentrations measured 24 hours after run start-up and / or change in operational pH. 137 A P P E N D I X I T a b l e 1-3: Run L, pH = 8.4. Sample taken at the end of _ minutes Recorded analytical after 2 system HRTs had elapsed PO^-P concentration (mg ' L*1) 0 49.7 5 49.5 20 49.5 30 48.9 40 48.6 50 48.4 Effluent PO4-P concentration1"1' at steady statefor a pH of 8.4 + 0.1 = 43.4 ± 2.8 mg - L' Values taken from Appendix H-I. Effluent P0 4-P concentrations measured 24 hours after run start-up and / or change in operational pH. 138 APPENDIX J APPENDIX J Calculated Ionic Strength For AH Runs (Pertaining to Section 5.8.2) Sample calculation If: C = Concentration of the feedwater constituent inside the reactor (moles / L) Z = Charge on the ion Then, I = Ionic strength of the mother liquor inside the reactor ( M ) ( n ) = 0.5 X Z(C X Z 2 ) Run A: Table J - l ( o ) Feed Ion C(moles/L) 0») Z CZ? P source N H 4 + 2.94E-03 1 2.94E-03 (NH 4 ) 2 HP0 4 H P 0 4 2 " 1.47E-03 -2 5.87E-03 N source N H 4 + 4.36E-03 1 4.36E-03 NH 4 CI Cl" 4.36E-03 -1 4.36E-03 Mg source M g 4.65E-04 2 1.86E-03 MgCl 2 Cl" 9.31E-04 -1 9.31E-04 E (C X Z 2 ) = 2.0 X 10"2 = 0.02 From Sawyer C, McCarty P. and Parkin G. (1994) Chemistry for Environmental Engineering. McGraw-Hill Series in Water Resources and Environmental Engineering, New York, 108. (o)2.94E-03 = 2.94Xl0"3 ^ Values calculated from Appendix H-II. 139 A P P E N D I X J Therefore, Ionic strength of the mother liquor inside the reactor for run A = 0.5 X 0.02 = 0.01 M Table J-2 Run Ionic strength identifier (M) A 0.01 B 0.01 C 0.01 D 0.01 E 0.01 F 0.01 G 0.01 H 0.01 I 0.01 J 0.01 K 0.01 L 0.01 M 0.01 . N 0.01 O 0.01 P 0.01 Q 0.03 R 0.03 s 0.01 T 0.01 U 0.01 140 A P P E N D I X K APPENDIX K Attempts To Predict The Seed Size Based On Stokes' Law / The Theory Of Discrete Particle Settling (Pertaining to Section 4.2.4) Fw Figure K-1: Free body diagram of a crystal inside the reactor(q) where FD = Drag force on the crystal F w = Weight of the crystal F B = Buoyant force on the crystal F F = Force offered by the surrounding fluid flow It is required to find that diameter, for which the crystal would stay suspended in the fluid, for a given fluid velocity. This diameter may be termed as the critical particle (crystal) diameter. For the purposes of this calculation, it will be assumed that the crystals are spherical in shape and that all crystals act as discrete particles. ^ All values and equations pertaining to this Appendix taken / developed from Metcalf and Eddy, Inc. (1991) Wastewater Engineering Treatment Disposal and Reuse. McGraw-Hill Series in Water Resources and Environmental Engineering, 222 - 223,1253. 141 APPENDIX K For the particle to remain suspended in a column of flowing fluid, Velocity of the particle = Velocity of the fluid From the free body diagram, F w = F D + F B + F F (i) If: .-3 p F = Mass density of the fluid, k g ' m" p = Mass density of the particle, k g ' m •3 V F = Average (upflow) velocity of the fluid, m ' s - l V P = Average velocity of the particle, m ' s A P = Area of the particle, m 2 j i p = Viscosity of the fluid, N-s ' m"2 C D = Coefficient of drag for the particle Rep = Particle Reynolds number Dp = Critical diameter of the particle, m Volp = Volume of the particle, m"3 g = Acceleration due to gravity, m ' s" Then, F w = p p X g X Vol? (ii) F D = C D X A P X p F X V P 2 (iii) 2 F B = p F X g X Volp (iv) 142 A P P E N D I X K F F = p F X V F X Q F (v) R e p = p F X V F X D p (vi) Volp = 7 t X D P 3 ...(vii) 6 A P = u X Dp2 (viii) C D = 24 + J_ + 0.34 (ix) R e p ( R e p / 5 ' » — ' 1 » — ' I II Sample calculation Consider an ambient temperature of 15 °C. Then, P F = 999.1 kg-ra" 3 MF= 1.139 X 10"3 N-s-m" 2 Consider limb A of the reactor. Then, Upflow velocity = 217 cm " min"1 ( r ) = 0.036 m ' s"1 ( r ) From Appendix B, Table B-1 143 A P P E N D I X K Combining and substituting Equations (ii) to (viii) in Equation (i), pp § 71 Dp3 = Cp 71 Dp2 PF V F 2 + PF g rt Dp3 + p F V F Q F (x) 6 2 6 - v Y y~ 111 IV V Also, LHS = Left hand side of Equation (x) RHS = Right hand side of Equation (x) Solving for Cp from Equation (ix): Table K-1 Particle diameter (mm) Cd = I + II + 0.34 I II 0.1 9.65 7.60 1.71 0.2 5.35 3.80 1.21 0.3 3.86 2.53 0.99 0.4 3.10 1.90 0.86 0.5 2.63 1.52 0.77 0.6 2.31 1.27 0.70 0.7 2.07 1.09 0.65 0.8 1.90 0.95 0.61 0.9 1.76 0.84 0.57 1.0 1.64 0.76 0.54 2.0 1.10 0.38 0.38 3.0 0.91 0.25 0.31 3.9 0.81 0.19 0.27 4.0 0.80 0.19 0.27 4.1 0.79 0.19 0.27 4.2 0.79 0.18 0.26 4.3 0.78 0.18 0.26 4.4 0.77 0.17 0.26 4.5 0.76 0.17 0.26 144 A P P E N D I X K Solving for the only unknown (D P)by trial and error, using values for C D from Table K - l , such that LHS = RHS in Equation (x): T a b l e K-2 Particle diameter (mm) LHS RHS = III + IV + V L H S - R H S III IV V 0.1 8.7E-09 2.4E-04 -2.4E-04 2.0E-07 5.1E-09 2.4E-04 0.2 7.0E-08 2.4E-04 -2.4E-04 4.4E-07 4.1E-08 2.4E-04 0.3 2.4E-07 2.4E-04 -2.4E-04 7.1E-07 1.4E-07 2.4E-04 0.4 5.6E-07 2.4E-04 -2.4E-04 1.0E-06 3.3E-07 2.4E-04 0.5 1.1E-06 2.4E-04 -2.4E-04 1.3E-06 6.4E-07 2.4E-04 0.6 1.9E-06 2.4E-04 -2.4E-04 1.7E-06 1.1E-06 2.4E-04 0.7 3.0E-06 2.4E-04 -2.4E-04 2.1E-06 1.8E-06 2.4E-04 0.8 4.5E-06 2.4E-04 -2.4E-04 2.5E-06 2.6E-06 2.4E-04 0.9 6.4E-06 2.5E-04 -2.4E-04 2.9E-06 3.7E-06 2.4E-04 1.0 8.7E-06 2.5E-04 -2.4E-04 3.3E-06 5.1E-06 2.4E-04 2.0 7.0E-05 2.9E-04 -2.2E-04 9.0E-06 4.1E-05 2.4E-04 3.0 2.4E-04 3.9E-04 -1.6E-04 1.7E-05 1.4E-04 2.4E-04 3.9 5.2E-04 5.7E-04 -5.1E-05 2.5E-05 3.0E-04 2.4E-04 4.0 5.6E-04 5.9E-04 -3.5E-05 2.6E-05 3.3E-04 2.4E-04 4.1 6.0E-04 6.2E-04 -1.8E-05 2.7E-05 3.5E-04 2.4E-04 4.2 6.5E-04 6.5E-04 -8.4E-07 2.8E-05 3.8E-04 2.4E-04 4.3 6.9E-04 6.8E-04 1.8E-05 2.9E-05 4.1E-04 2.4E-04 4.4 7.4E-04 7.1E-04 3.7E-05 3.0E-05 4.4E-04 2.4E-04 4.5 8.0E-04 7.4E-04 5.7E-05 3.1E-05 4.7E-04 2.4E-04 From Table K-2, it can be seen that theoretically, LHS = RHS for a critical particle / crystal diameter between 4.2 mm - 4.3 mm. Similarly, the corresponding numbers at 15 ° C , for the other limbs of the reactor are shown in Table K-3. T a b l e K -3 Reactor limb Calculated critical diameter (mm to mm) B 3.3-3.4 C 2.8-2.9 145 APPENDIX K However, the calculated values for the critical particle diameter always exceeded the observed values of the seeded crystal diameters, which could withstand the fluid upflow velocity in the respective limbs of the reactor. This discrepancy could have been the result of an incorrect assumption regarding the shape of the crystals, and the added fact that the crystals do not behave as discrete particles in reality. 146 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0063895/manifest

Comment

Related Items