1 PHOSPHORUS RECOVERY BY HYBRID ANION EXCHANGE AND STRUVITE PRECIPITATION: APPLICATIONS TO SOURCE SEPARATED URINE AND COMBINED WASTEWATER STREAMS By A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2013
2 201 3
3 ACKNOWLEDGEMENTS T hank you to my research advisor Dr. Boyer my parents committee members Dr. Chadik and Dr. McLamore, and members of my research group for all of your help and support I would also like to thank all of the professors in the Environmental Engineering Department at the University of Florida for their impartment of knowledge and guidance. I would like to thank employees at the Howard F. Curren Advanced Wastewater T reatment Plant in Tampa, FL for providing water samples. I would also like to thank the National Science Foundation for funding this work under grant number CBET 1150790
4 TABLE OF CONTENTS Page 7 9 ..11 CHAPTER 1 ... 1 3 2 PHOSPHATE RECOVERY USING HYBRID ANION EXCHANGE: APPLICATIONS TO SOURCE SEPARATED URINE AND COMBINED WASTEWATER STREAMS ... ...16 2.1 Phosphate Recovery from Waste Streams by Hybrid Anion Exchange 2.2 Exp 1 9 2.2.1 .. 1 9 184.108.40.206 Domestic w aste s ... .. .... 1 9 220.127.116.11 Diluted and m ixed u .. 2 1 18.104.22.168 Phosphate s elective r .. 2.2.2 Experimental Methods.. 2.2.3 Equilibrium Isotherm Models.. 2.2.4 2.3 5 2.3.1 Domestic Waste Streams..... ...2 5 22.214.171.124 Phosphate so 126.96.36.199 Equilibrium i 27 2.3.2 ..28 188.8.131.52 Phosphate s 184.108.40.206 Equilibrium i sotherms... 2.3.3 Phosphate Recov ery Potential and Regeneration.. 2.4 Discussion .33 2.4.1 .33 2.4.2 Phosphate Recovery an 2.4.3 Cha 2.5 Summary of Phosphate Recovery from Major Waste Streams by Hybrid Anion Exchan
5 3 PHOSPHORUS RECOVERY FROM URINE AND ANAEROBIC DIGESTER FILTRATE: 1 3.1 Comparison of Hybrid Anion Exchange with Direct Precipitation for Phosphorus Recovery Over 1 3.2 Experiment 4 3.2.1 Wast e Streams. 4 3.2.2 Column Experiments an 5 3.2.3 Precipi ... 7 3.2.4 An .5 8 3.3 Results and Disc 59 3.3.1 Column 59 3.3.2 Regeneration Ef 1 3.3.3 Mineral Prec ipitation 3 3.3.4 Mass Balance on P Recove 4 3.4 6 6 4 79 APPENDIX X RAY DIFFRACTION D 1 LIST OF REFERENCES 8 8 9 5
6 LIST OF TABLES Table page 2 1 Chemical compositions of fresh urine (Urine F), ureolyzed urine (Urine H 1, Urine H 2), anaerobic digester supernatant (ADS 1, ADS 2), greywater (GW) and secondary wastewater effluent (WW) used in batch equilibrium exp eriments. 2 2 Water characteristics of diluted waters 2 3 Results for initial values of measured phosphate and pH and calculated values for ionic strength and species fractions for all synthetic waters and subsequent dilutions.. 2 4 Langmuir and Freundlich isotherm parameters determined by linear regression corresponding to Fig ure 2 3 for synthetic waters tested 2 5 Langmuir and Freundlich isotherm parameters determined by linear regression corresponding to Figure 2 4 for diluted waters tested, dilution factor (DF) where DF = (V urin e +V flush )/V urine sum of square errors (SSE) ..44 2 6 Phosphate recovery characteristics for all waste streams and diluted waste streams tested. .. 3 1 Chemical compositions of synthetic fresh and ureolyzed urine. Ureolyzed urine composition assumed complete urea hydrolysis with subsequent precipitation of calcium and magnesium with 68 3 2 69 3 3 P capacity of HAIX Fe resin, mass loading of P on resin and mass of P recovered during .. 0 3 4 Regeneration of HAIX 1 3 5 Recovery of P through struvite precipitation directly in each waste stream compared to precipitation in regeneration solution from the HAIX ................7 2 3 6 Chemicals added to each waste stream and waste regeneration solution to precipitate .. 3
7 LIST OF FIGURES Figure page 2 1 Removal of phosphate as a function of resin dose for a) fresh urine (urine F) b) ureolyzed urine (urine H 1 and urine H 2) c) anaerobic digester supernatant (ADS 1) d) (ADS 2) e) greywater (GW) f) wastewater (WW) 2 2 Experimental data and adsorption isotherms determined by linear regression for a) fresh urine (urine F) b) ureolyzed urine (urine H 1) c) (urine H 2) d) anaerobic digester supernatant (ADS 1) e) (ADS 2) f) greywater (GW) g) s eco ndary wastewater effluent 2 3 Removal of orthophosphate as a function of resin dose for a) urine F, DF=3.5 b) urine F, DF=21 c) urine F, DF=31 d) urine H 1, DF=3.5 e) urine H 1, DF=21 f) urine H 1, DF=31 g) GW + urine H 1, DF=1 and GW + urine H 1, DF=3.5 h) GW + urine H 1, DF=21 i) GW + urine H 48 2 4 Experimentally derived data and adsorption isotherms for a) urine F, DF=3.5 b) urine F, DF=21 c) urine F, DF=31 d) urine H 1, DF=3.5 e) urine H 1, DF=21 f) urine H 1, DF=31 g) GW + urine H 1 DF=1 h) GW + urine H 1, DF=3.5 i) GW + urine H 1, DF=21 j) GW + urine H 1, DF=31 .49 2 5 Flow diagram illustrating the movement of phosphorous from a household utilizing source separation through a wastewater treatment plant... .5 0 3 1 Effluent over influen t P concentration and pH against increasi ng bed volumes for fresh ... 7 4 3 2 Effluent over influent P concentration (C/C 0 ) with increasing bed volumes from column experiments for a) fresh urine b) ureolyzed urine and c) anaerobic digester filtrate... 7 5 3 3 Effluent P concentration with increasing bed volumes for regeneration of HAIX Fe 7 6 3 4 Effluent P concentration with increasing bed volumes for regeneration of HAIX Fe exhausted with a) fresh urine b) ureolyzed urine c) anaerobic 7 7 3 5 Column set up used for all breakthrough experiments and regenerations with subsequent precipitation of MAP by chemical addition of H + Mg 2+ and NH 4 + to the used regeneration solution. 7 8 A 1 X Ray diffraction results for struvite precipitation in an 1 A 2 X Ray diffraction results for struvite precipitation in anaerobic digester filtrate waste regeneration soluti 2
8 A 3 X Ray diffraction results for potassium struvite pre 3 A 4 X Ray diffraction results for potassium struvite precipitation in fresh urine waste regeneration soluti 4 A 5 X Ray diffraction results for struvite precipitati on in ure 5 A 6 X Ray diffraction results for potassium struvite precipitation in ureolyzed urine waste regeneration solution 6 A 7 X Ray diffraction results for struvite precipitation in ureolyzed urine waste regeneration 7
9 LIST OF ABBREVIATIONS ADS Anaerobic digester supernatant ARE Average relative error BV Bed Volume C e Equilibrium concentration DF dilution factor DI Deionized DOC Dissolved organic carbon EBCT Empty bed contact time GW Greywater HAIX Hybrid anion exchange resin HAIX Fe Hybrid anion exchange resin with immobilized iron oxide HAP Hydroxyapatite K F F reundlich constant K L Langmuir constant KMP Potassium magnesium phosphate MAP Magnesium ammonium phosphate P Phosphorus PO 4 P Phosphate as P q e Equilibrium resin capacity q max Langmuir sorption capacity RO Reverse osmosis rpm Rotations per minute
10 RSD Relative standard deviation SLV Superficial linear velocity SSE Sum of squared error Urine F Fresh urine Urine H Ureolyzed urine WW Secondary wastewater effluent WWTP Wastewater treatment plant
11 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering PHOSPHORUS RECOVERY BY HYBRID ANION EXCHANGE AND STRUVITE PRECIPITATION: APPLICATIONS TO SOURCE SEPARATED URINE AND COMBINED WASTEWATER STREAMS By August 2013 Chair: Treavor H. Boyer Major: Environmental Engineering Sciences The overall goal of this work was to find the most favorable means of recovering phosphorus (P) from major waste streams in order to provide a sustainable source of P and reduce reliability o n phosphate rock, a non renewable resource Batch equilibrium tests and continuous flow column test s were used to evaluate th e efficacy of a hybrid anion exchange resin (HAIX) as a means for recovering P from major waste streams. A HAIX resin containing hydrous ferric oxide was used in this study because of its selectivity for phosphate and the option to precipitate P minerals i n the waste regeneration solution. Batch equilibrium tests with HAIX resin generated isotherm models used to evaluate: fresh urine, ureolyzed urine, anaerobic digester supernatant, secondary wastewater effluent, greywater and various dilutions of urine. The maximum loading of P determined from the isotherm models was fresh urine > ureolyzed urine > ana r e ywater > secondary wastewater effluent. The P recovery potential was fresh urine > ureolyz ed urine > greywater > secondary wastewater effluent > anaerobic digester supernatant. Results indicated that the sorption capacity of HAIX resin for phosphate and total P recovery potential were greater for source separated urine than the combined wastewa ter streams of secondary wastewater effluent or anaerobic digester supernatant. Dilution of urine with tap water decreased the P loading on HAIX resin.
12 C ontinuous flow column operation with HAIX was used to remove P from fresh urine, ureolyzed urine and s ludge belt press filtrate from a full scale anaerobic digester and subsequently to recover P by struvite crystallization within the waste regeneration solution. The recovery of P using the HAIX process was compared with direct precipitation within each ind ividual waste stream considering chemical addition requirement for precipitation of struvite. Complete exhaustion of the HAIX resin was at ~11, ~12, and ~ 50 bed volumes (BV) for fresh urine, ureolyzed urine, and anaerobic digester filtrate respectively w ith operating resin capacit y of 10.1, 5.9, and 6.4 mg P/g resin. Regeneration efficiency of HAIX exhausted with P from each waste stream was > 94% using a caustic brine solution with diminishing efficiency as the regeneration solution was reused. Precipita tion of struvite resulted in > 96% P recovery for direct precipitation within each waste stream and at the end of the HAIX process within the waste regeneration solution. Both direct precipitation and HAIX are viable options for P recovery, however direct precipitation in fresh and ureolyzed urine is favored when considering chemical addition requirements. Alternatively, HAIX is recommended over direct precipitation in anaerobic digester filtrate due to mineral impurities associated with direct precipitatio n in this waste stream. The results of this work advance the current understanding of nutrient recovery from complex wastewater streams by sorption processes.
13 CHAPTER 1 INTRODUCTION Phosphate rock is a non renewable resource under threat of depletion (Van Vuuren et al., 2010) and the current major source of P in fertilizers which are necessary for global food security (Childers et al., 2011) A potential solution to this dilemma is th e recovery of P from wastewaters (Cabeza et al., 2011). An added value to P recovery from wastewaters is reduced nutrients loading to surface waters from wastewater treatment plants (WWTP) which are known point sources of excess P (Carey and Migliaccio, 2 009) can cause adverse effects such as eutrophication. Of all potential wastewater stream s source separated urine is particularly suitable for P recovery due to high P concentration (200 800 mg P/L) and the ability to precipitate phosphate minerals such a s struvite (Wilsenach et al., 2007) due to high concentrations of P and ammonium Furthermore urine accounts for ~1% of wastewater flows but ~50% of the total P load to WWTPs (Larsen and Gujer, 1996; Wilsenach and van Loosdrecht, 2006). Nutrient recovery f rom source separated urine is predicted to decrease energy usage and increase capacities at WWTPs (Wilsenach and van Loosdrecht, 2006). Urine source separation technology utilizes waterless urinal and no mix toilets, which can be used to collect and store urine (Rossi et al., 2009). P can then be recovered from the stored urine by direct precipitation of struvite (Antonini et al., 2011) or in the case of this study, by sorption to hybrid anion exchange ( HAIX ) resin However, treatment of urine poses many challenges due to its complex chemistry (Maurer et al., 2006) confounded by high ionic strength and highly variable concentrations Another challenge with P recovery from urine is urea hydrolysis which changes fresh urine to ureolyzed urine and is caused b y urease positive bacteria (Mobley and Hausinger, 1989) which are assumed to be ubiquitous in urine collection systems. The hydrolysis of urea produces ammonium, bicarbonate, and an increase in pH from ~6 to ~9.3
14 (Mobley and Hausinter, 1989) which is favo rable for the precipitation of hydroxyapatite (HAP, Ca 10 (PO 4 ) 6 (OH) 2 ) and struvite (MAP, MgNH 4 PO 4 6H 2 O) (Udert et al., 2003a). Subsequently, P is lost throughout the collection system and further losses occur at the WWTP due to biological uptake. When dilut ed with tap water, the additional calcium and magnesium can cause even further precipitation and loss of P (Uder t et al., 2003b). Precipitation and recovery of P minerals is possible and can be maximized (> 95% P rec overy) by addition of magnesium (Wilsen ach et al., 2007). Another promising waste stream is anaerobic digester effluents which c an be high in P (Botinni and Rizzo, 2012). Excess P in sludge management processes at WWTPs can result in fouling of pumps and pipes due to precipitation (Doyle et al., 2002) which could be abated by removal of P from anaerobic digester waste streams (Lew et al., 2011) In addit ion to direct precipitation, sorption processes can be used to recover P from a variety of waste streams. Hybrid anion exchange (HAIX) resins are particularly ideal for removal of P due to their selectivity of P over other competing anions such as chloride sulfate and bicarbonate (Pan et al., 2009). HAIX resin is strong base anion exchange resin that ha s been infused with metal oxide nanoparticles. R esin s make use of Zr(IV) (Awual et al., 2011; Zhu and Jyo, 2005), Cu(II) (Zhao and Sengupta, 1998), and Fe( III) (Blaney et al., 2007) all of which form inner sphere complexes with phosphate. Additionally HAIX resins have the ability to precipitate minerals such as struvite and hydroxyapatite within the waste regeneration solution (Sengupta et al., 2011) which has the potential to circumvent problems associated with direct precipitation such as biological contamination (Decrey et al., 2011) and poor mineral quality (Sakthivel et al, 2012). P recovery from wastewaters offers a promising solution to phosphate rock depletion however it cannot be efficiently impl emented due to several gaps in knowledge which are
15 preventing engineering strategies from being employed. Foremost, there is no compa rison of P recovery by direct precipitation in concentrated waste streams with HAIX processes There is some research on the effect of diluted urine on sorption to clinoptilolite (Kocatrk and Baykal, 2012) but none on removal of P from human urine using H AIX resin with subsequent regeneration and precipiation of P minerals. Additionally there is no previous work that has examined which waste stream is the most effective for P recovery using HAIX resin when considering P recovery potential and sorption chemistry. Accordingly, the goal of this study was to evaluate the potential to recover P using selective sorption and direct precipitation within a variety of waste streams recovery using hybrid anion exchange: Applications to source the sorption chemistry associated with P recovery using a HAIX resin and the P recovery potential from: fresh urine, ureolyz ed urine, urine diluted with tap water and mixed with greywater anaerobic digester supernatnat, greywater, and secondary wastewater effluent. press filtrate: A P recovery using continuous flow column operation with HAIX resin and subsequent regeneration and precipitation of P minerals. Chapter 3 also compares the sorption process wit h direct precipitation considerig chemical addition requirements for precipitation of P minerals. Chapter 4 contains major conclusions and recommendations resulting from the findings in Chapter s 2 and 3.
16 CHAPTER 2 PHOSPHATE RECOVERY USING HYBRID A NION EXCHANGE: APPLICATIONS TO SOURCE SEPARATED URINE AND COMBINED WASTEWATER STREAMS 2.1 Phosphate Recovery from Waste Streams by Hybrid Anion Exchange Overview Phosphate rock is the main source of phosphorus (P) in fertilizers and is essential for high crop yields in agriculture (Schrder et al., 2011). Phosphate rock is also a non renewable resource that is under threat of depletion (Van Vuuren et al., 2010). Confounding this problem is a growing world population and greater food demands, which wil l require phosphate based fertilizers (Childers et al., 2011). There is the potential, however, to recover P from wastewater streams for use as fertilizer (Cabeza et al., 2011) in order to lessen the problem of limited phosphate rock supply. Not only could P recovery from wastewater contribute to more sustainable P management (Cordell et al., 2011), but it could also reduce P loading to receiving waters. It is known that wastewater treatment plants are a point source of excess P to receiving waters (Carey a nd Migliaccio, 2009), which can cause eutrophication and degraded water quality. One approach to P recovery is the precipitation of struvite at the wastewater plant, e.g., membrane concentrate from aerobic or anaerobic treatment (Bradford Hartke et al., 20 12) or filtrate from sludge belt filter press (Lew et al., 2011). A more radical approach to P recovery is source separation and treatment of urine, which accounts for ~1% of wastewater by volume, yet contributes ~50% of the total P load to wastewater (Lar sen and Gujer, 1996; Wilsenach and van Loosdrecht, 2006). Recent studies on source separation and treatment of urine outline many of the potential benefits that could come from this technique such as lower energy requirements for wastewater treatment (Wil senach and van Loosdrecht, 2006), water conservation, nutrient recovery, and decreased loading of nutrients (Larsen et al., 2009) and pharmaceuticals (Lamichhane and Babcock, 2012)
17 to the environment. Urine source separation is accomplished using waterless urinals and no mix toilets, which can be used to collect undiluted or low diluted urine in storage tanks (Rossi et al., 2009). However, treatment of urine poses challenges due to its complicated chemistry (Maurer et al., 2006). The composition of urine ch anges from fresh urine to ureolyzed urine once it leaves the human body and flows through urinals, toilets, and wastewater piping. The change in composition is due to urea hydrolysis that is catalyzed by urease positive bacteria (Mobley and Hausinger, 1989 ), which are assumed to be present in toilets, urinals, and wastewater collection systems. The hydrolysis of urea creates ammonium, bicarbonate, and increases pH Eq. 2 1 (Mobley and Hausinger, 1989). Urea hydrolysis can occur in minutes to hours in undilut ed urine, and causes loss of phosphate through p recipitation of hydroxyapatite Eq. 2 2 and struvite Eq. 2 3 (Udert et al., 2003a). In conventional urinals and toilets, dilution of urine with tap water can cause phosphate precipitation if calcium or magnesi um are present (Udert et al., 2003b). Thus the potential to recover P from urine changes as the composition of urine changes, e.g, fresh urine, ureolyzed urine, urine diluted with tap water or mixed with greywater. NH 2 (CO)NH 2 + 3H 2 O 2NH 4 + + HCO 3 +OH (Eq. 2 1) 10Ca 2+ + 6PO 4 3 + 2OH Ca 10 (PO 4 ) 6 (OH) 2(s) (Eq. 2 2) Mg 2+ + NH 4 + + PO 4 3 +6H 2 O MgNH 4 PO 4 6H 2 O (s) (Eq. 2 3) The focus of current research on P recovery from urine is through struvite precipitation, e.g., in ful ly automated reactors (Antonini et al., 2011), in simple hand operated reactors (Etter et al., 2011), or by electrolytic magnesium dosage (Hug and Udert, 2013). A promising approach for P recovery from urine is hybrid anion exchange (HAIX), which has not b een tested with urine but has been evaluated in wastewaters (Blaney et al., 2007; Martin et al., 2009). Because conventional strong base anion exchange resins have a higher selectivity for sulfate over
18 phosphate (Gregory and Dhond, 1972), HAIX resins were developed for selective removal of phosphate over competing anions like sulfate (Blaney et al., 2007; Pan et al., 2009; Sengupta and Pandit, 2011). HAIX resin consists of strong base anion exchange resin with immobilized metal or impregnated metal oxide na noparticles. Previous research has tested HAIX resins with Zr(IV) (Awual et al., 2011; Zhu and Jyo, 2005), Cu(II) (Zhao and Sengupta, 1998), and Fe(III) (Blaney et al., 2007) all of which form inner sphere complexes between metal and phosphate. The inner s phere complex can preferentially remove phosphate in the presence of competing anions such as chloride, sulfate, or bicarbonate (Pan et al., 2009). This is contrary to typical ion exchange which forms outer sphere complexes and favors the ion with the high er valence, typically sulfate (Helfferich, 1995) at pH 7 HAIX resin containing Fe(III) (hereafter HAIX Fe) was selected for this study because it is commercially available, and has been tested for phosphate removal from lake and stream water (Boyer et al. 2011), domestic secondary wastewater effluent (Blaney et al., 2007; Martin et al., 2009), industrial wastewater effluent (Pan et al., 2009), reverse osmosis concentrate from wastewater treatment (Kumar et al., 2007), and sludge liquor from wastewater tre atment (Bottini and Rizzo, 2012). HAIX Fe resin exhausted with phosphate is effectively regenerated using caustic brine solution with > 80% phosphate recovery (Blaney et al., 2007; Sengupta and Pandit, 2011) and solid phase fertilizers (struvite and calciu m phosphate) can be precipitated from the waste regeneration solution (Kumar et al., 2007; Sengupta and Pandit, 2011). Despite the opportunities for P recovery from source separated urine, several gaps in knowledge are preventing the implementation of th is strategy. Foremost, there is no previous research on the removal of phosphate from urine using HAIX resin. While some research has evaluated the effect of diluted urine on ion exchange using clinoptilolite (Kocatrk and Baykal,
19 2012), there are no data on how dilution of urine affects phosphate removal using HAIX resin. There is also no previous work that has examined which wastewater stream is most effective for P recovery when considering maximum recovery potential for P and sorption chemistry. Accordi ngly, the goal of this study was to evaluate the potential to recover P from source separated urine and combined wastewater streams that included undiluted urine, urine diluted with tap water, greywater, mixture of urine and greywater, anaerobic digester s upernatant, and secondary wastewater effluent. The specific waste streams were chosen because they comprise the major waste streams that make up domestic wastewater (Kujawa Roeleveld and Zeeman, 2006). The specific objectives were: (1) to evaluate the sorp tion capacity and selectivity of HAIX Fe resin for phosphate in fresh and ureolyzed urine, greywater, anaerobic digester supernatant, and secondary wastewater effluent; (2) to evaluate the effect of diluting urine with tap water and mixing urine with greyw ater on the sorption capacity and selectivity of HAIX Fe resin for phosphate; and (3) to identify the most efficient location to recover P starting at the point of urine generation and through wastewater treatment. 2.2 Experimental 2.2.1 Materials 220.127.116.11 Domestic w aste s treams Synthetic wastewaters including fresh urine (urine F), ureolyzed urine (urine H), anaerobic digester supernatant (ADS), household greywater (GW), and secondary wastewater effluent (WW) were used in this study (Table 2 1). Synthetic wastewaters were chosen to ensure minimal variation in composition because real wastewaters can vary greatly. The compositions of synthetic urine were modeled after previous literature containing measured concentrations from real urine (Ronte ltap et al., 2007; Wilsenach et al., 2007; Udert and Wchter, 2012)
20 Synthetic urine has been used successfully in previous studies showing similar chemical thermodynamics during precipitation when compared to real urine (Ronteltap et al., 2007) It was as sumed that synthetic urine and real urine would exhibit similar chemical thermodynamics during sorption based on previous studies showing comparable sorption results for the removal of ammonium from synthetic urine (Lind et al., 2000) and real urine (Bayka l et al., 2009). Fresh urine represents urine composition before urea hydrolysis and precipitation has occurred, while ureolyzed urine represents urine composition following complete urea hydrolysis and spontaneous precipitation of struvite and hydroxyapat ite due to the associated pH. Two recipes for ureolyzed urine were used in this study, urine H 1 and urine H 2, with slightly different compositions but with the same phosphate concentration. Urine H 2 assumed complete urea hydrolysis from the fresh urine composition while urine H 1 was prepared separately and contains slightly higher ammonium and bicarbonate concentrations. Sodium concentrations vary with each urine composition because all other concentrations were fixed while sodium was allowed to vary be cause it does not compete with phosphate for sorption sites. All pH adjustments were made using 1 M NaOH or HCl. Daily flow per person for urine was assumed to be 1.4 L p 1 d 1 based on void volume of 200 mL void 1 p 1 and 7 void d 1 (Chung and van Mastrig t, 2009) The composition of anaerobic digester supernatant (ADS 1) was modeled after supernatant characteristics from a real anaerobic digester (Battistoni et al., 2001) A second anaerobic digester supernatant (ADS 2) was used in order to incorporate org anic constituents and to establish a reliable flow. The inorganic concentrations for ADS 2 were modeled after previous work (Battistoni et al., 2001; Stratful et al., 2001) with organic constituents based on Wei et al. (2011). For ADS 2, the daily flow per person of 1.11 L p 1 d 1 was determined to be 0.42% the daily flow per person for secondary wastewater effluent (Stratful et al., 2001). For accurate
21 calculation of mass flow of P, both the phosphate concentration and flow were based on Stratful et al. (2 001) which represents the mass flow of P from the centrate produced from an anaerobic digester, however, does not account for all streams produced from an anaerobic digester such as filtrate produced during the dewatering process. The greywater composition was intended to represent typical household greywater consisting of kitchen refuse, laundry water, shower water and sink water (Daiper et al., 2008) with a daily flow per person of 91.3 L p 1 d 1 from Kujawa Roeleveld and Zeeman (2006) The secondary wast ewater effluent was designed to represent biologically treated domestic wastewater (Seo et al., 1996) ; the daily flow per person of 262 L p 1 d 1 is an average value for the United States (Dziegielewski and Kiefer, 2010) The compositions and flows for all synthetic waste streams were best estimates and can vary widely in real world situations. The maximum potential for P recovery, or total P load, was calculated as the product of the daily flow per person and initial phosphate concentration. 2.2. 1.2 Diluted and mixed u rine Characteristics of diluted and mixed urine are shown in Table 2 2 The urine dilutions were based on flush volumes found in urinals and toilets manufactured after 1994, and assumed 7 urinal or toilet flushes per day to match 7 voi ds per day. Urine was diluted with tap water in order to simulate real world conditions. The dilution factor (DF) was defined as the total volume of urine and flush water divided by the volume of urine. The DFs corresponded to high efficiency urinal (0.5 L flush 1 DF = 3.5), standard urinal (4 L flush 1 DF = 21), and standard toilet (6 L flush 1 DF = 31). The DF only applied to urine; the volume of greywater remained constant in the mixture of urine and greywater. Fresh urine was not mixed with greywater due to the rapid hydrolysis (within hours) of urea that would occur in bathroom plumbing. The mixture of urine and greywater was based on the amount of ureolyzed urine and greywater that 1 person
22 produces per day with varying flush volumes. Thus the volum e ratio of greywater to urine was 91.3:1.4 and tap water was added to achieve different DFs for urine. All batch equilibrium tests and phosphate measurements were performed on the same day the dilutions were made. 18.104.22.168 P hosphate selective r esin The HAIX Fe resin used in this work has the commercial name PhosX np or LayneRT and is manufactured by SolmeteX (Northborough, MA). HAIX Fe resin is a strong base anion exchange resin impregnated with hydrous ferric oxide nanoparticles (Blaney et al., 2007) The hydrous ferric oxide sites in HAIX Fe resin undergo ligand exchange between the oxygen atom attached to iron and phosphate in solution; Fe(III) serves as the Lewis acid (electron pair acceptor) which forms an inner sphere complex with the Lewis base (electron pair donor) phosphate (Stumm and Morgan, 1996). HAIX Fe resin was used as received. The resin was stored in deionized (DI) water and measured in a graduated cylinder as volume of wet settled resin for volumes of 1 mL or greater. For resin volumes less than 1 mL, the resin was dried for a minimum of 48 h in a desiccator and weighed on an analytical balance where 1 mL of wet settled resin was equal to 0.389 g of dry resin. 2.2.2 Experimental Methods Batch equilibrium tests were performed on all syn thetic wastewaters using HAIX Fe resin. Tests were performed in amber bottles containing 100 mL of solution and varying amounts of resin. The samples were collected after 2 h of mixing on an Innova 2000 Platform Shaker at 200 rpm, which was determined to b e the equilibrium time from previous kinetic tests (Sendrowski and Boyer, 2013). The doses used for different synthetic wastewaters were adjusted based on initial phosphate concentration to attain phosphate removal in the range of 10% to 99%. All resin dos es were performed in triplicate and phosphate concentrations are reported as mean
23 value of triplicate samples. The precision of dosing the resin was verified by calculating the relative standard deviation (RSD = 100 times the ratio of the standard deviatio n to the mean) of measured phosphate concentrations for triplicate samples. The RSD was <10% for 90% of all samples and no sample exceeded a RSD of 20%. The pH of each wastewater was not adjusted to the same value because the goal of this research was to i dentify which waste stream is the most effective for P recovery and not to evaluate the resin on its own. In addition, the pH increased with increasing resin dose and subsequent phosphate removal due to hydroxide release during ligand exchange. It is recog nized that not being able to maintain a constant pH is a limitation of the batch experiments. Batch regeneration of the resin was performed in triplicate using 15 mL of HAIX Fe resin that was saturated with phos phate and 100 mL of a 2.5% NaCl/ 2.0% NaOH by weight solution. The HAIX Fe resin was saturated during the batch equilibrium test using ADS 2. The HAIX Fe resin and regeneration solution were mixed at 200 rpm for 2 h. The regeneration solution was analyzed for phosphate to determine the amount of phos phate desorbed from the resin. 2.2.3 Equilibrium Isotherm Models The parameters for the Langmuir isotherm model Eq. 2 4 and the Freundlich isotherm model Eq. 2 5 were determined by linear regression after linear transformation of the models and equilibriu m data. The Langmuir model assumes monolayer coverage and homogenous adsorption due to equally distributed adsorption sites and energy which creates a maximum resin capacity (Foo and Hameed, 2010). The Freundlich model can apply to multilayer adsorption an d assumes a distribution of adsorption sites and energy across a heterogeneous surface (Foo and Hameed, 2010). A thorough discussion on the Langmuir and Freundlich models is described
24 elsewhere (Weber and DiGiano, 1996) The fit of the isotherm models to t he equilibrium data were statistically evaluated using t he sum of squared errors (SSE) Eq. 2 6 an d average relative error (ARE) Eq. 2 7 where the better fitting model has lower SSE and ARE (Foo and Hameed, 2010). Parameters for Langmuir and Freundlich mode ls are discussed in sections 3.1.2 and 3.2.2.Parameter n in Eq. 2 6 and Eq. 2 7 is number of experimental data points. (Eq. 2 4) (Eq. 2 5) (Eq. 2 6) (Eq. 2 7) 2.2. 4 Analytical Methods Chemicals of ACS grade purity were combined with DI water when preparing all standard solutions and synthetic wastewaters. Phosphate was measured using the ascorbic acid method following Standard Method 4500 P (Eaton et al., 2005) using a Hitachi U 2900 sp ectrophotometer and a 1 cm quartz cuvette. In this method a reagent containing sulfuric acid, ammonium molybdate, antimonyl tartrate, and ascorbic acid was combined with samples containing phenolphthalein, which forms a blue color that can be measured at a n absorbance of 880 nm. All samples were diluted in order to attain concentrations within the calibration curve of 0.15 1.2 mg P/L. Calibration check standards and matrix spikes were performed with all phosphate measurements to ensure accuracy. The relativ e difference between measured and known values was <5% for all calibration check standards for all tests. Matrix spike recoveries
25 were between 95% and 105% for all tests. All phosphate concentrations are as P. pH was measured using an Accumet AB 15 pH mete r (Fischer Scientific), which was calibrated with buffer solutions of pH 4, 7, and 10 prior to use. 2 .3 Results 2.3.1 Domestic Waste streams 22.214.171.124 Ph osphate s orption The decrease in phosphate concentration as a function of resin dose is shown in Fig ure 2 1 for undiluted urine, anaerobic digester supernatant, greywater, and secondary wastewater effluent. Greater than 90% phosphate removal was achieved in all wastewater streams, with the required resin dose increasing as the initial phosphate concentra tion increased. For example, fresh urine had the highest initial phosphate concentration of 668 mg P/L and showed > 94% removal at a resin dose of 300 mL/L. Anaerobic digester supernatant (ADS 2) had an initial phosphate concentration of 82.5 mg P/L and > 94% removal for a resin dose of 150 mL/L. Greywater had an initial phosphate concentration of 5.2 mg P/L and > 94% removal for a resin dose of 5 mL/L. The pH of the wastewater streams was ordered from most acidic to most basic as follows: fresh urine < se < ureolyzed urine (Table 2 3). The pH of the wastewater streams increased as the HAIX Fe resin dose increased because of hydroxide release due to ligand exchange, especially fo r unbuffered wastewater streams. For example, pH after HAIX Fe resin ranged from pH 6.0 8.9 for fresh urine, pH 7.8 8.5 for greywater, and pH 6.7 7.6 for secondary wastewater effluent. The pH after HAIX Fe resin changed only slightly ( 0.1 pH units) for u reolyzed urine and anaerobic digester supernatant because of high concentration of carbonate and ammonia alkalinity. The higher pH
26 (9.7) for urine H 1 could result from carbon dioxide stripping after urea hydrolysis which would lower the buffer capacity of bicarbonate. This scenario could occur if ureolyzed urine was stored for a prolonged period of time where the volatilization of carbon dioxide could occur. The higher pH in urine H 1 could have a negative impact on the performance of HAIX Fe resin due to the increased hydroxide ions competing with phosphate for ligand exchange sites and the more negative surface charge of hydrous ferric oxide as shown by Morel and Hering ( 1993). Ionic strength was calculated using Visual MINTEQ based on the compositions in Table 2 1 (Gustafsson, 2012) The ionic strength was examined because it affects the interactions between charged solutes and charged surfaces. For instance, phosphate adsorption to iron (hydr)oxide, at pH above isoelectric point, has been shown to increa se with increasing ionic strength because of charge screening (Geelhoed et al., 1997). The ionic strength of the waste streams was ordered: wastewater effluent (4.1 10 4 M) < greywater (1.4 10 3 M) < fresh 10 1 M) < ureolyzed urine (~5 10 1 M) (Table 2 3). Hence, the high ionic strength of urine and anaerobic digester supernatant may give some advantages to phosphate sorption, especially at high pH. Also related to ionic strength, the waste streams cover different activity coefficient regions from sufficiently dilute such that activity is approximately equal to concentration to high ionic strength requiring the Davies equation for calculation of activity coefficients (Morel and Hering, 1993) Following fr om the pH and ionic strength, the distribution of phosphate between monovalent (H 2 PO 4 ) and divalent (HPO 4 2 ) was calculated based on the compositions in Table 2 1 using Visual MINTEQ, which uses the Davies equation to calculate all activity coefficients. The phosphate distribution (H 2 PO 4 / HPO 4 2 ) was fresh urine (0.94/0.06), secondary wastewater effluent (0.73/0.23), anaerobic digester supernatant (0.28/0.72), greywater (0.2/0.8), ureolyzed
27 urine (<0.01/0.99) (Table 2 3). The hydrous ferric oxide will als o vary as a function of pH and 2 + ) with increasing pH (Morel and Hering, 1993). Thus, the predominant inner sphere complexes between iron and phosphate 2 PO 2 ) 3 2 ) in anaerobic digester supernatant, greywater, and ureolyzed urine (Blaney et al., 2007; Zeng et al., 2008). 2.3. 1.2 Equilibr ium i sotherms Equilibrium data for phosphate sorption to HAIX Fe resin showed a favorable nonlinear relationship between resin phase equilibrium concentration ( q e ) and solution phase equilibrium concentration ( C e ) for all wastewater streams, except secondary wastewater effluent which showed approximately linear sorption behavior (Fig ure 2 2). A favorable nonlinear isotherm shows preferential accumulation of solute on the solid phase relative to a linear isotherm. The Langmuir and Freundlich parameter s along with corresponding SSE and ARE are listed in Table 2 4. The best fit isotherm model to phosphate sorption data was determined based on SSE and ARE. The Langmuir model was the best fit in fresh urine. The Freundlich model was the best fit in ureolyz ed urine, anaerobic digester supernatant, and greywater. The Langmuir, Freundlich, and linear models showed similar fit in secondary wastewater effluent. The Langmuir model parameters provide insights on sorption capacity and sorption affinity between pho sphate and hydrous ferric oxide. The monolayer sorption capacity ( q max ) of HAIX Fe resin for phosphate was fresh urine > ureolyzed sing initial phosphate concentration. Increasing value of the Langmuir constant ( K L ) indicates
28 increasingly favorable sorption. The Langmuir constant was greywater > anaerobic digester supernatant > secondary wastewater effluent > ureolyzed urine > fresh u rine. The Freundlich model parameters provide a different interpretation on sorption capacity and sorption affinity than the Langmuir model. The Freundlich constant K F approximates the sorption capacity (similar to q max ) when 1/ n approaches 0 as shown by the units of K F when 1/ n is equal to 0 (i.e., mmol g 1 ). Based on K F the sorption capacity of HAIX Fe resin for phosphate ureolyzed urine, which is a different trend than the Langmuir monolayer sorption capacity ( q max ). The Freundlich constant K F is not a good approximation of the sorption capacity in secondary wastewater effluent because 1/ n approaches 1, so it is not correct to compare K F for secondary wastewater eff luent with the other wastewater streams. As the Freundlich constant 1/ n approaches 0 it indicates favorable, nonlinear sorption whereas 1/ n approaching 1 indicates less favorable, linear sorption. Based on 1/ n sorption of phosphate to HAIX Fe resin showe d similar level of favorable, nonlinear sorption in fresh urine, ureolyzed urine, anaerobic digester supernatant, and greywater. Also based on 1/ n phosphate sorption to HAIX Fe resin was less favorable in secondary wastewater effluent compared to the othe r wastewater streams. 2.3.2 Diluted and Mixed Urine 126.96.36.199 Phosphate s orption The decrease in phosphate concentration as a function of HAIX Fe resin dose is shown in Fig ure 2 3 for urine diluted with tap water and urine mixed with greywater. As the DF increased for fresh urine and ureolyzed urine, the initial concentration of phosphate decreased and the HAIX Fe resin dose required for > 90% phosphate removal decreased. For exam ple, as the DF for fresh urine increased from 3.5 (high efficiency urinal) to 31 (standard toilet) the initial
29 phosphate concentration decreased from 175 to 22 mg P/L and the resin dose required for > 90% phosphate removal decreased from 100 to 20 mL/L (fi rst row of Fig ure 2 3). A similar trend for ureolyzed urine diluted with tap water is shown in the second row of Figure 2 3. The initial phosphate concentration in the mixture of ureolyzed urine and greywater did not decrease as the DF increased as was exp ected based on calculations in Table 2 2. The initial phosphate concentration was ~8 mg P/L for the mixture of ureolyzed urine and greywater at DFs equal to 1, 3.5, and 21 (Fig ure 2 3 and Table 2 3). The initial phosphate concentration increased to 11 mg P/L for the mixture of ureolyzed urine and greywater at DF equal to 31 (Fig ure 2 3 and Table 2 3). Thus the initial concentration of phosphate measured (Fig ure 2 3 and Table 2 3) and calculated (Ta ble 2 2) for the mixture of ureolyzed urine and greywater disagreed by ~30%. Two likely explanations for the disagreement are precipitation of phosphate minerals in the mixture of undiluted ureolyzed urine and greywater and hydrolysis of phosphorus compoun ds in the mixture of diluted ureolyzed urine and greywater. Dilution of fresh urine with tap water increased the pH, whereas dilution of ureolyzed urine with tap water did not change the pH because the urine was highly buffered by ammonium The mixture of ureolyzed urine and greywater had the same pH as ureolyzed urine and a greater pH than greywater. For fresh urine diluted with tap water the pH after HAIX Fe resin increased because the solution was weakly buffered; the only carbonate alkalinity was fro m tap water. The pH in diluted fresh urine after HAIX Fe resin ranged from 6.3 9.0 (DF = 3.5), 6.9 8.4 (DF = 21), and 7.1 8.4 (DF = 31). The pH after HAIX Fe resin remained approximately constant ( 0.1 pH units) for ureolyzed urine diluted with tap water and ureolyzed urine mixed with greywater because of carbonate and ammonia alkalinity.
30 The ionic strength of urine and urine mixed with greywater decreased as the DF increased (Table 2 3). For example, urine highly diluted with tap water (e.g., DF = 31) had ionic strength between 5 10 3 M (fresh urine) and 1 10 2 M ( ureolyzed urine). The mixture of urine and greywater had an ionic strength approximately 9 higher than greywater and similar ionic strength as highly diluted urine. 188.8.131.52 Equilib rium i sotherms Equilibrium data for phosphate sorption to HAIX Fe resin showed a favorable, nonlinear relationship between solid phase concentration ( q e ) and solution phase concentration ( C e ) for all diluted and mixed urine streams (Fig ure 2 4 and Table 2 5). T he Langmuir model did not show as good of a fit to the equilibrium data for diluted urine or mixed urine as it did for undiluted urine, anaerobic digester supernatant, and greywater. For example, most ARE values were > 8% and as high as 24% for diluted and mixed urine. Nevertheless, examining the values and trends of the Langmuir parameters can provide insights on the sorption process. The monolayer sorption capacity ( q max ) of HAIX Fe resin for phosphate was lower in diluted urine than undiluted urine. For example, q max in diluted fresh urine was 0.149 to 0.196 mmol g 1 compared with q max of 0.291 mmol g 1 in undiluted fresh urine. In comparing diluted fresh urine there was no obvious relationship between q max and DF, e.g., q max (DF = 21) > q max (DF = 31) > q max (DF = 3.5). Diluted ureolyzed urine showed the same trends as diluted fresh urine regarding the effect of dilution on q max Ureolyzed urine (undiluted or diluted) mixed with greywater had similar q max as greywater and lower q max than undiluted ureolyzed urine. The Langmuir constant ( K L ) generally increased in diluted and mixed urine as the DF increased, and showed increasingly favorable sorption in the order of undiluted fresh urine < undiluted ureolyzed urine < diluted
31 ureolyzed urine < diluted fresh urine < mixture of diluted ureolyzed urine (DF = 1, 3.5) and greywater < greywater < mixture of diluted ureolyzed urine (DF = 21, 31) and greywater. The Freundlich model showed a better fit to the equilibrium data when compared to the Langmuir model for almost all diluted and mixed urine streams with lower SSE and ARE < 10%. Based on K F the sorption capacity of HAIX Fe resin for phosphate in urine increased as the DF increased. For example, the trend for K F in ureolyzed urine was undiluted (DF = 1) DF 21 < DF 31. In ureolyzed urine mixed with greywater, K F and the sorption capacity decreased as the DF increased. The mixture of undiluted ureolyzed urine and greywater had a greater K F than either undiluted ureolyzed urine or greywater. The F reundlich constant 1/ n indicated favorable, nonlinear sorption in all diluted and mixed urine streams: 0.15 < 1/ n < 0.4. There was no obvious trend for the effect of dilution on 1/ n For example, 1/ n was approximately constant in undiluted and diluted fres h urine, 1/ n increased as DF increased in ureolyzed urine, and 1/ n decreased as DF increased in mixture of ureolyzed urine and greywater. 2.3.3 Phosphate Recovery Potential and Regeneration A comparison of all wastewater streams shown in Fig ure 2 5 was made to quantify the maximum potential for P recovery and corresponding HAIX Fe resin requirements at each location (Table 2 6). Phosphorus recovery potential (mg p 1 d 1 ) from greatest to least was: mixture of ureolyzed urine and greywater > fresh u rine > ureolyzed urine > greywater > secondary wastewater effluent > anaerobic digester supernatant. The total P load in secondary wastewater effluent is contributed by the P from urine and greywater but is less than the sum of these waste streams due to p recipitation and biological uptake. Dilution of urine with tap water did not affect the P recovery potential because the mass balance on P remained the same assuming no precipitation. Ureolyzed urine had a lower P recovery potential than fresh urine due
32 to urea hydrolysis and the subsequent precipitation of calcium and magnesium phosphate minerals, which causes loss of phosphate in terms of the P mass balance. It is also possible for precipitation to occur in urine diluted with tap water and urine mixed wit h greywater, which would decrease the P recovery potential for these waste streams. The P recovery potential of anaerobic digester supernatant was low due to the composition and flow of supernatant modeled for anaerobic digestion in this work (Table 2 1). Data on the phosphate concentration in anaerobic digester supernatant vary considerably in the literature, e.g., approx. 20 mg P/L (Battistoni et al., 2001) to 300 mg P/L (Lew et al., 2011). Phosphorus recovery potential could be increased by combining the anaerobic digester supernatant with other streams such as belt filter press supernatant from waste activated sludge liquor (Stratful et al., 2001). The phosphate loading on HAIX Fe resin ( q ) was calculated by substituting values for the Freundlich isoth erm parameters (Tables 2 4 and 2 5) and initial phosphate concentration (Table 2 3) into Eq. 2 5. Phosphate loading on HAIX Fe resin (mg g 1 ) from greatest to least ureolyzed urine > undiluted ureolyz ed ureolyzed ureolyzed urine diluted with tap water > secondary wastewater effluent. A trend observed was that loading increased the phosphate concentration. The final column in Table 2 6 is the volume of HAIX Fe resin required for maximum P recovery on a per person per day basis and assumes that sorption capacity is reached and complete phosphate removal occurs. This was calculated by dividing maximum P recovery by phosphate loading on HAIX Fe resin and converting units from mass of resin to volume of resin. HAIX Fe resin requirements were expressed as a volume to illustrate the amount of resin required and reactor size for each wastewater stream. Anaerobic digester supernatant had the
33 lowest resin requirement but also the lowest P recovery potential. Fresh urine and ureolyzed urine had the lowest resin requirements among the wastewater st reams that had high potential for P recovery due to high phosphate loading on HAIX Fe resin in urine. Dilution of urine increased resin requirements due to decrease in phosphate loading. Secondary wastewater effluent had the highest resin requirement due t o the low phosphate loading. Resin requirements for the other wastewater streams were bracketed between fresh urine (0.22 L p 1 d 1 ) and secondary wastewater effluent (0.71 L p 1 d 1 ). Regeneration results confirmed that HAIX Fe can be effectively regene rated using a 2.5% NaCl/2.0% NaOH solution. Regeneration resulted in > 92% desorption of phosphate from HAIX Fe resin, which is in agreement with previous studies on HAIX Fe regeneration (Blaney et al., 2007). The ultimate recovery of phosphate is achieved by precipitating phosphate minerals in the waste regeneration solution (Sengupta and Pandit, 2011). 2.4 Discussion 2.4.1 Phosphate Sorption The Freundlich isotherm was a better fit than the Langmuir isotherm to equilibrium data for phosphate sorption to HAIX Fe resin considering all wastewater streams. The better fit of the Freundlich isotherm suggests that the HAIX Fe resin has sorption sites with a distribution of energies, which is consistent with the combination of electrostatic and Lewis acid base interactions that occur between phosphate and the quaternary ammonium and iron (hydr)oxide exchange sites on the surface of the HAIX Fe resin, respectively (Blaney et al., 2007; Sengupta and Pandit, 2011) The Freundlich isotherm can also represent the su mmation of two or more Langmuir isotherms (Weber and DiGiano, 1996) e.g., there could be three Langmuir isotherm models describing phosphate sorption to HAIX 3 2 ), bidentate
34 2 PO 2 ), and electrostatic (quaternary ammonium) which sum up to Freundlich isotherm behavior again, representing different interaction energy between sorption sites and phosphate. There is limited equilibrium isotherm data published on phosphate sorption to HAIX Fe resin. Sengu pta and Pandit (2011) fit equilibrium data on phosphate sorption to HAIX Fe resin to the Freundlich isotherm model, but they did not show the goodness of fit between data and model or evaluate the fit of other isotherm models. Bottini and Rizzo (2012) used the Generalized Langmuir Freundlich isotherm to model phosphate sorption to HAIX Fe resin. The maximum loading of phosphate on HAIX Fe resin in this study was 10 mg P/g resin in fresh urine and 5 7 mg P/g resin in ureolyzed urine (Table 2 6). Pan et al. (2009) showed a maximum phosphate loading of ~10 mg P/g resin for water containing phosphate (approx. 80 mg P/L; initial concentration not reported), sulfate (500 mg/L), and pH 6.4 6.7, and Bottini and Rizzo (2012) showed a maximum phosphate loading of 10 12 mg P/g resin for real sludge liquor (475 mg P/L, 10 mg/L sulfate, 303 mg/L chloride, and pH 8). HAIX resin containing Zr(IV) showed a maximum phosphate loading of ~16 mg P/g resin for water containing 80 mg P/L at pH 6 (Zong et al., 2013). HAIX resin co ntaining Cu(II) showed a maximum phosphate loading of ~2 mg P/g resin for water containing 18 mg P/L and 200 mg/ L sulfate at pH 7 (Zhao and Seng upta, 1998). Modeling phosphate sorption to HAIX Fe resin in the wastewater streams used in this work is compli cated because of varying pH, ionic strength, and competing anions, and because all measurements were made in the bulk solution phase and could be much different inside the resin because of the Donnan co ion exclusion effect (Cumbal and Sengupta, 2005) The Donnan co ion exclusion effect causes repulsion of co ions which can lower the permeability of the resin and impact adsorption. It is not expected that other inorganic substances (SO 4 2 Cl HCO 3 ) had
35 much of an effect on the fitted isotherms due to pre vious research by Pan et al. (2009). Bottini and Rizzo (2012) observed earlier breakthrough of phosphate during HAIX Fe packed column tests for real sludge liquor than synthetic solution, which they attributed to presence of complex organic matter in real sludge liquor. Therefore, organics present in greywater, secondary wastewater effluent, and anaerobic digester supernatant could have an effect on the fitted isotherms. Furthermore, biofilm formation would cause additional reduction in sorption efficiency (Lahav and Green, 2000). The pH also plays a significant role by changing speciation of hydrous ferric oxide surface site, speciation of phosphate in solution, and hydroxide concentration. The experimental results indicate that both phosphate species, H 2 PO 4 and HPO 4 2 are selectively removed by HAIX Fe resin. 2.4.2 Phosph a te Recovery and Implementation HAIX Fe resin showed > 90% removal of phosphate for all wastewater streams and > 92% phosphate desorption from HAIX Fe resin during regeneration. Based on the results in this work and previous results in the literature (Kumar et al., 2007; Sengupta and Pandit, 2011) it is expected that essentially all phosphate removed using HAIX Fe resin can be recovered as solid phase phosphate minerals from the waste regeneration solution. Assuming a population of 50,000, the greatest potential for P recovery was ureolyzed urine mixed with greywater (55 kg d 1 ), fresh urine (43 kg d 1 ), and ureolyz ed urine (30 kg d 1 ). The actual P recovery from urine mixed with greywater is likely to be lower because of phosphate precipitation with calcium and magnesium from tap water and household products (Udert et al., 2003b). Secondary wastewater effluent and a naerobic digester supernatant had the lowest potential for P recovery: 21 and 4.5 kg d 1 respectively. One reason for the lower P recovery potential for wastewater effluent and anaerobic digester supernatant is soluble phosphate converting to solid phase minerals by
36 spontaneous precipitation throughout the wastewater collection system and treatment plant (Houhou et al., 2009) and through assimilation and biological growth during wastewater treatment (Bitton, 2011). Bradford Hartke et al. (2012) showed P re covery of 20 kg d 1 for reverse osmosis concentrate from biological nutrient removal plant and 59 kg d 1 for nanofiltration concentrate from anaerobic treatment plant; both scenarios were based on centralized wastewater plant with population of 50,000. Wil liams (1999) showed P recovery of 20 kg d 1 for anaerobic digester supernatant, which was based on centralized wastewater plant serving population of 165,000. Thus, P recovery from source separated urine is on the higher end of P recovery from wastewater e ffluent and sludge supernatant at centralized plants. Fresh urine had the highest recovery potential for P considering the source separated waste streams and the highest phosphate loading on HAIX Fe resin. However, treating a large volume of fresh urine at a centralized location is not practical unless urea hydrolysis is inhibited by the addition of chemical inhibitors (Mobley and Hausinger, 1989) Ureolyzed urine had lower recovery potential for P and lower phosphate loading on HAIX Fe resin than fresh uri ne, but it is more favorable when compared to secondary wastewater effluent or anaerobic digester supernatant. Ureolyzed urine would be easier to treat compared to fresh urine because urea hydrolysis can be ignored. Larsen and Gujer (1996) proposed source separation and storage of urine at each individual buildings; the urine would then be released to the wastewater collection system at a time of low flow, such as the middle of the night, and treated at the centralized wastewater plant, e.g., using a fixed bed or fluidized bed reactor containing HAIX Fe resin. From Table 2 6, a resin volume of 11 m 3 would be required for ureolyzed urine from a population of 50,000 with at least one regeneration per day. The exact reactor size would depend on number of reacto rs in service and down flow or up flow operation. Ureolyzed urine mixed
37 with greywater had the highest recovery potential for P; this mixing already occurs at houses and could be diverted and treated in the same manner described for ureolyzed urine. Howeve r, precipitation of phosphate with calcium or magnesium could lower the recovery potential for P and result in treatment complications such as blockages and clogging. Dilution of wastewater streams is of practical importance because almost all current wast ewater collection systems contain dilute wastewater. Dilution of ureolyzed urine with tap water containing calcium and magnesium will decrease P recovery potential due to precipitation. HAIX Fe resin had a lower phosphate loading in diluted urine than undi luted urine. For example, ureolyzed urine diluted with tap water (DF = 31) would require twice as much resin as undiluted ureolyzed urine. Thus dilution of urine with tap water decreases the effectiveness of P recovery in two ways: P available for recovery is decreased due to losses from precipitation and resin requirements are increased due to lower phosphate sorption to HAIX Fe resin in dilute wastewater. As a result, the effectiveness of phosphate recovery can be increased by decreasing the level of dilu tion in urine and wastewater collection systems, which could be achieved by installing more water efficient urinals and toilets. Bradford Hartke et al. (2012) reached a similar conclusion that decreasing per capita water use would decrease the energy requi rements for P recovery from centralized wastewater plant effluent. 2.4.3 Challenges Several challenges are anticipated regarding urine source separation and phosphate recovery. One major challenge is that source separated urine would require separate piping and storage tanks. Decentralized treatment of urine would require more reactors th an centralized treatment; an extreme situation would be a reactor at every house. Centralized treatment of urine could be achieved by releasing the urine into the wastewater collection system during low flow
38 periods (Larsen and Gujer, 1996); however this w ould require sophisticated controls and detailed understanding of the collection system. While urine source separation has the potential to lower energy requirements for wastewater treatment (Wilsenach and van Loosdrecht, 2006) it is not known whether thi s could offset the cost of infrastructure upgrades. It is also unclear whether adsorption processes are the most effective means of phosphate recovery from urine. For example, direct precipitation of struvite in ureolyzed urine can be accomplished with 90% recovery efficiency (Etter et al., 2011). Furthermore, the theoretical composition of struvite is 12.6% by mass as P, whereas the maximum load of P on HAIX Fe was ~1%. Phosphate recovery and use as fertilizer could potentially generate revenue, but has no t been shown to be economically sustainable on its own (Martin et al., 2009) Public perception could be another barrier to urine source separation but has the potential to be remedied through education (Lienert and Larsen, 2009) The next steps would incl ude life cycle assessment of phosphate recovery from ureolyzed urine using decentralized and centralized approaches to treatment (i.e., adsorption, precipitation, and membrane separation) and economic analysis of urine derived fertilizer products (Lundin e t al., 2000; Kumar et al., 2007) 2.4 Summary of P Recovery from Major Waste Streams by Hybrid Anion Exchange A hybrid anion exchange resin containing hydrous ferric oxide (HAIX Fe) was able to selectively remove phosphate from source separated and combi ned wastewater streams that included fresh urine, ureolyzed urine, urine diluted with tap water, greywater, urine mixed with greywater, secondary wastewater effluent, and anaerobic digester supernatant. Phosphate sorption on HAIX Fe resin showed favorable, nonlinear relationship between solid phase concentration and solution phase concentration for all wastewater streams except secondary wastewater effluent which showed linear sorption behavior at a relatively lower phosphate concentration The no nlinear sorption of phosphate on HAIX Fe resin was better fit by the Freundlich isotherm than the Langmuir isotherm considering all wastewater streams however the best fit model for these wastewater streams did vary
39 Source separation of fresh urine and ureolyzed urine had higher recovery potential for P (868 and 590 mg p 1 d 1 respectively) and showed greater phosphate loading on HAIX Fe resin (10.1 and 6.9 mg g 1 respectively) than the end of pipe waste streams of secondary wastewater effluent (414 m g p 1 d 1 ; 1.5 mg g 1 ) and anaerobic digester supernatant (89 mg p 1 d 1 ; 5.2 mg g 1 ). This resulted in lower HAIX Fe resin requirements (L p 1 d 1 ) for urine than secondary wastewater effluent. Dilution of urine with tap water decreased the phosphate lo ading on HAIX Fe resin. As a result, diluted urine required more HAIX Fe resin than undiluted urine to recover the same amount of phosphate. Diluted urine had higher phosphate loading on HAIX Fe resin and lower HAIX Fe resin requirements than secondary was tewater effluent. The results of this work indicate that more phosphate can be recovered and it can be recovered more effectively from building level wastewater streams, i.e., undiluted urine, urine diluted with tap water, greywater, and urine mixed with greywater, than wastewater streams typical of a central treatment plant.
40 Table 2 1. Chemical compositions of fresh urine (Urine F), ureolyzed urine (Urine H 1, Urine H 2), anaerobic digester supernatant (ADS 1, ADS 2), greywater (GW) and secondary wastewater effluent (WW) used in batch equilibrium experiments. Values determined from amount of salts and products added. Chemical/Product Unit Urine F Urine H 1 Urine H 2 ADS 1 ADS 2 GW WW Na + mmol/L 74 83.6 104 90.8 108.5 0.97 K + mmol/L 50 50 40 0.051 Ca 2+ mmol/L 4 5.0 5.0 Mg 2+ mmol/L 4 2.2 0.5 0.0041 NH 4 + N a mmol/L 722 500 58.6 41.7 0.36 Cl mmol/L 100 100 100 14.4 11.1 HCO 3 C a mmol/L 267 250 149 149 0.30 0.25 SO 4 2 mmol/L 10 10 15 0.25 0.058 PO 4 3 P a mmol/L 20 13.6 13.6 0.72 2.58 0.18 0.051 Urea N a mmol/L 500 Boric Acid (H 3 BO 3 ) mmol/L 0.023 Lactic Acid (C 3 H 6 O 3 ) mmol/L 0.31 Moisturizer mg/L 10 Toothpaste mg/L 32.5 Deodorant mg/L 10 Vegetable Oil mg/L 7 Shampoo mg/L 720 Laundry Detergent mg/L 150 Beef Extract mg/L 0.25 1.8 Peptone mg/L 2.7 Humic Acid mg/L 4.2 Tannic Acid mg/L 4.2 Sodium Lignin Sulfonate mg/L 2.4 Sodium Lauryl Sulfate mg/L 0.94 Acacia Gum Powder mg/L 4.7 Arabic Acid mg/L 5 pH 6 9.7 9.3 7.6 7.6 7.8 6.7 Volume L p 1 d 1 1.4 c 1.4 c 1.4 c 1.11 d 91.3 e 262 f Ionic Strength mol/L 0.145 0.528 0.476 0.160 0.160 0.000406 a Total species. b Hydrolysis of urea causes the precipitation of calcium and magnesium phosphate minerals. c Assumes 7 voids per day and 200 mL of urine per void. d Considered 0.43% the flow of WW (Stratful et al., 2001) e (Kujawa Roeleveld and Zeeman, 2006) f (Dziegielewski and Kiefer, 2010)
41 Table 2 2. Water characteristics of diluted waters. Flow assumes 200 mL of urine per void and 7 voids per day where each void assumes to 1 flushes. Wastewater Diluted DF a Flow (L p 1 d 1 ) PO 4 3 P b (mmol L 1 ) pH Urine F 3.5 4.9 5.71 6.3 Urine F 21 29.4 0.95 6.9 Urine F 31 43.4 0.65 7.1 Urine H 1 3.5 4.9 3.89 9.4 Urine H 1 21 29.4 0.65 9.3 Urine H 1 31 43.4 0.44 9.3 GW + Urine H 1 1 92.7 0.38 9.3 GW + Urine H 1 3.5 96.2 0.37 9.2 GW + Urine H 1 21 120.7 0.29 9.3 GW + Urine H 1 31 134.7 0.26 9.1 a DF=(V urine + V flush )/V urine b Total species.
42 Table 2 3. Results for initial values of measured phosphate and pH and calculated values for ionic strength and species fractions for all synthetic waters and subsequent dilutions. Water DF a PO 4 P b (mg L 1 ) pH c IS d H 2 PO 4 /Total e HPO 4 2 /Total e Urine F 1 668 6 0.145 0.9403 0.0596 Urine H 1 1 470 9.7 0.528 0.0039 0.9944 Urine H 2 1 457 9.3 0.476 0.0078 0.9913 ADS 1 21.1 7.6 0.160 0.2838 0.7162 ADS 2 82.5 7.6 0.158 0.2838 0.7162 GW 5.17 7.8 0.00144 0.2000 0.8000 WW 1.8 6.7 0.000406 0.7290 0.2316 Urine F 3.5 175 6.3 0.0439 0.8877 0.1123 Urine F 21 31.4 6.9 0.00787 0.6651 0.3349 Urine F 31 22.1 7.1 0.00544 0.5562 0.4438 Urine H 1 3.5 108 9.4 0.155 0.0062 0.9927 Urine H 1 21 15.1 9.3 0.0246 0.0062 0.9927 Urine H 1 31 8.9 9.3 0.0169 0.0078 0.9913 GW + Urine H 1 1 7.7 9.3 0.00924 0.0078 0.9913 GW + Urine H 1 3.5 8.0 9.2 0.00919 0.0078 0.9913 GW + Urine H 1 21 7.8 9.3 0.00889 0.0078 0.9913 GW + Urine H 1 31 11.0 9.1 0.00895 0.0098 0.9895 a DF = Dilution Factor = (V urine +V flush )/V urine b Measured values of total species as P. c Measured values. d IS = ionic strength, calculated using Visual MINTEQ. e Fraction of phosphate species activity to total species activity, calculated using Visual MINTEQ.
43 Table 2 4. Langmuir and Freundlich isotherm parameters determined by linear regression corresponding to Fig ure 2 3 for synthetic waters tested. Sum of square errors (SSE) and average relative error (ARE) were used to statistically compare model fits to t he equilibrium data. Synthetic Water Langmuir Freundlich qmax (mmol g 1 ) K L (L mmol 1 ) SSE (mmol 2 g 2 ) ARE (%) K F (mmol 1 1/n L 1/n g 1 ) 1/n SSE (mmol 2 g 2 ) ARE (%) Urine F 0.291 1.18 4.5E 04 3.5 0.149 0.255 2.6E 03 8.1 Urine H 1 0.185 2.87 3.9E 03 11.5 0.113 0.248 2.2E 04 3.3 Urine H 2 0.164 3.71 6.3E 04 5.5 0.109 0.191 2.5E 04 3.9 ADS 1 0.137 50.2 5.0E 04 6.5 0.166 0.237 9.6E 05 3.7 ADS 2 0.123 15.9 2.7E 03 15.4 0.124 0.299 5.7E 05 2.3 Greywater 0.126 262 6.0E 04 7.4 0.235 0.259 2.1E 04 6.4 Wastewater 0.130 9.8 1.4E 05 4.1 0.420 0.765 1.5E 05 5.5
44 Table 2 5. Langmuir and Freundlich isotherm parameters determined by linear regression corresponding to Figure 2 4 for diluted waters tested, dilution factor (DF) where DF = (V urine +V flush )/V urine sum of square errors (SSE), average relative error (ARE). The ratio o f greywater to ureolyzed urine was based on the amount of diluted ureolyzed urine and greywater produced per day; this assumed a void volume of 200 mL of urine (Chung and van Mastrigt, 2009) and 7 voids per day compared to 91 .3 liters per person per day of greywater (Kujawa Roeleveld and Zeeman, 2006) Seven voids per day assumes 7 flushes per day. Synthetic Water DF Langmuir Freundlich q max (mmol g 1 ) K L (L mmol 1 ) SSE (mmol 2 g 2 ) ARE (%) K F (mmol 1 1/n L 1/n g 1 ) 1/n SSE (mmol 2 g 2 ) ARE (%) Urine F 3.5 0.149 59.5 1.77E 03 8.9 0.141 0.22 1.02E 03 11.7 Urine F 21 0.196 101 1.29E 04 2.9 0.217 0.16 9.76E 04 8.5 Urine F 31 0.176 330 1.05E 02 22.1 0.247 0.21 1.38E 03 5.6 Urine H 3.5 0.108 55.8 3.54E 03 19.6 0.112 0.25 2.37E 04 6.2 Urine H 21 0.118 37.1 1.38E 03 14.4 0.183 0.35 1.37E 04 4.3 Urine H 31 0.089 68.6 2.69E 03 20.1 0.186 0.40 2.29E 04 6.1 GW + Urine H 1 0.115 184 8.64E 03 23.9 0.308 0.39 6.91E 04 7.1 GW + Urine H 3.5 0.147 122 5.89E 04 8.7 0.235 0.29 5.30E 04 12.2 GW + Urine H 21 0.121 282 1.56E 03 10.9 0.225 0.27 5.35E 04 8.3 GW + Urine H 31 0.122 292 1.57E 03 11.9 0.181 0.22 4.56E 04 5.1
45 Table 2 6. Phosphate recovery characteristics for all waste streams and diluted waste streams tested. a Resin capacity determined by Freundlich parameters and equation q e = K F C e 1/n b Volume of resin required for max P recovery on a person per day basis. This assumes that adsorption capacity is reached and that complete P removal occurs. c Value denotes the dilution factor of urine. Wastewater Stream Max P Recovery (mg p 1 d 1) q a (mg g 1 ) Resin Required b (L p 1 d 1 ) Urine F 868 10.1 0.22 Urine H 1 590 6.90 0.22 Urine H 2 590 5.65 0.27 ADS 1 4.70 ADS 2 89 5.15 0.04 GW 509 4.58 0.29 WW 414 1.49 0.71 Urine F, 3.5 c 868 6.40 0.35 Urine F, 21 c 868 6.74 0.33 Urine F, 31 c 868 7.11 0.31 Urine H 1, 3.5 c 590 4.74 0.32 Urine H 1, 21 c 590 4.39 0.35 Urine H 1, 31 c 590 3.51 0.43 GW + Urine H 1, 1 c 1100 5.53 0.51 GW + Urine H 1, 3.5 c 1100 4.93 0.57 GW + Urine H 1, 21 c 1100 4.78 0.59 GW + Urine H 1, 31 c 1100 4.46 0.63
46 Figure 2 1. Removal of phosphate as a function of resin dose for a) fresh urine (urine F) b) ureolyzed urine (urine H 1 and urine H 2) c) anaerobic digester supernatant (ADS 1) d) (ADS 2) e) greywater (GW) f) wastewater (WW). All doses were performed in tr iplicate and reported as averages. Values which measured lower than the lowest point on the calibration point, 0.150 mg/L are reported as 0.150 mg/L. Error bars represent upper and lower limits of standard deviation.
47 Figure 2 2. Experimental data and adsorption isotherms determined by linear regression for a) fresh urine (urine F) b) ureolyzed urine (urine H 1) c) (urine H 2) d) anaerobic digester supernatant (ADS 1) e) (ADS 2) f) greywater (GW) g) secondary wastewater effluent (WW). Error bars in vertical and horizontal directions represent the upper and lower limits of standard deviation.
48 Figure 2 3. Removal of orthophosphate as a function of resin dose for a) urine F, DF=3.5 b) urine F, DF=21 c) urine F, DF=31 d) urine H 1, DF=3.5 e) urine H 1, DF=21 f) urine H 1, DF=31 g) GW + urine H 1, DF=1 and GW + urine H 1, DF=3.5 h) GW + urine H 1, DF=21 i) GW + urine H 1, DF=31. Where dilution factor (DF) = (V urine + V flush )/V urine All doses were performed in triplicate and reported as averages. All valu es which measured below the calibration curve value of 0.150 mg/L were reported as 0.150 mg/L. Error bars represent the upper and lower limits of standard deviation.
49 Figure 2 4. Experimentally derived data and adsorption isotherms for a)urine F, DF=3.5 b) urine F, DF=21 c) urine F, DF=31 d) urine H 1, DF=3.5 e) urine H 1, DF=21 f) urine H 1, DF=31 g) GW + urine H 1 DF=1 h) GW + urin e H 1, DF=3.5 i) GW + urine H 1, DF=21 j) GW + urine H 1, DF=31. Where dilution factor (DF) = (V urine + V flush )/V urine All doses were performed in triplicate and reported as averages. All values which measured below the calibration curve value of 0.150 mg /L were reported as 0.150 mg/L. Error bars represent the upper and lower limits of standard deviation.
50 Figure 2 5. Flow diagram illustrating the movement of phosphorous from a household utilizing source separation through a wastewater treatment plant.
51 CHAPTER 3 PHOSPHORUS RECOVERY FROM URINE AND ANAEROBIC DIGESTER FILTRATE: COMPARISON OF SORPTION WITH PRECIPITATION 3.1 Comparison of Hybrid Anion Exchange with Direct Precipitation for Phosphorus Recovery Overview Due to increasing world population a sustainable source of phosphorus (P) is critical to meet agricultural demands and provide global food security (Cordell et al., 2009) Conservative estimates indicate that phosphate rock, the current main source of P, could reach depletion levels of 60 70% by the end of the end of the 21 st century (Van Vuuren et al., 2010) To abate this dilemma recovery of P from wastewaters has been proposed as a viable solution (Guest et al., 2009) Recovery of P from wastewaters is particularly desirable due to the adverse impact s that arise from P in wastewater effluents which can cause significant disruption to valuable ecosystems across the globe such as the Florida Everglades (Noe et al., 2001),Chesapeake Bay (Boesch et al., 2001) freshwater lakes in China (Le et al., 2010), and the Thames catchment in the UK (Young et al., 1999). In addition to this, struvite precipitation due to excess P at wastewater treatment plants (WWTPs) can cause operational problems during sludge management processes due to mineral scaling (Doyle et al., 2002) of pipes and pumps. Remo val and recovery of P from concentrated waste streams such as human urine and sludge belt press filtrate from anaerobic digest ion (hereafter referred to as anaerobic digester filtrate) would provide an alternative to phosphate rock fertilizers while having the additional benefits of decreased nutrient loading to receiving waters and reduced operational problems caused by mineral precipitation at the WWTP. Technologies used to recover P work particularly well in concentrated waste streams such as human urin e and waste streams produced from anaerobic digestion due to their high P content and low flow rates. One such technology used to recover P is direct precipitation by addition of
52 chemicals such as calcium and magnesium to precipitate the minerals: struvite (MgNH 4 PO 4 6H 2 O) also known as magnesium ammonium phosphate (MAP), potassium struvite (KMgPO 4 6H 2 O) also known as potassium magnesium phosphate (KMP), and hydroxyapatite (HAP, Ca 10 (PO 4 ) 6 (OH) 2 ); all of which can all be used as slow release fertilizers (Ueno and Fuji, 2001; de Bashan and Bashan, 2004). Wilsenach et al. (2007) found that > 95% recovery of P by precipitation of struvite and potassium struvite is possible in human urine while anaer obic digester waste streams can yield 94% P recovery by precipitation of struvite (Mnch and Barr, 2001) Hydroxyapatite is typically less desirable compared to struvite since it contains no nitrogen and can be negatively affected by calcium carbonate precipitation. However P removal efficiencies of 85% in sludge side streams (Moriyama et al., 2010) have bee n observed and could potentially be increased in waters with high P and low carbonate concentrations due to the high pK sp (~58) of hydroxyapatite (Bell et al., 1978) One of the major limitations to direct precipit ation is mineral impurity mainly due to calcium which can precipitate calcium carbonate (Hao et al., 2008) and poten tial contamination due to heavy metals (Sakthivel et al., 2012) and pathogens (Decrey et al., 2011) making the precipitates unviable as a fertilizer Human urine and anaerobic digester waste streams are also desirable for hybrid anion exchange (HAIX) processes due to increased resin capacity when compared to biologically HAIX processes utilize strong base anion exchange re sins that have been infused with metal oxides. HAIX resins include the use of Zr(IV) (Zhu and Jyo, 2005) Cu(II) (Zhao and Sengupta, 2000) and Fe(III) (Blaney et al., 2007) all of which form an inner sphere complex with phosphate. This mechanism allows for phosphate to be selectively removed over competing anions such as sulfate, chloride, and bicarbonate (Pan et al., 2009) This study focused on a
5 3 HAIX resin infused with Fe (III), henceforth HAIX Fe. Studies have been performed to investigate the use of HAIX Fe to remove P under continuous flow column operation from wastewater (Blaney et al., 2007; Martin et al., 2009; Pan et al., 2009), anaerobic digester liquor (Bottini and Rizzo, 2012), and reverse osmosis (RO) concentrate from a wastewater treatment plant (Kumar et al., 2007). Blaney et al. ( 2007 ) found that saturation occur red at ~16,000 bed volumes (BVs) for trace phosphate removal (influent P = 0.26 mg/L) while more con centrated streams such as sludge liquor and RO concentrate can experience breakthrough at ~25 BVs (Bottini and Rizzo, 2012) and ~160 BVs (Kumar et al., 2007). Regeneration of HAIX Fe resins using caustic brine can have efficiency > 90% within 10 BVs (Sengu pta et al., 2011) and the option to precipitate P as struvite within the regeneration solution (Kumar et al., 2007; Sengupta et al., 2011) by addition of Mg 2+ and NH 4 + Sengupta et al. (2011) also investigated reuse of the regeneration solution finding tha t regeneration efficiency can still be > 90% after recycling a waste regeneration solution that has been subjected to precipitation of struvite with the addition of fresh caustic brine solution. Recovery of P within the waste regeneration solution is an in direct process and involves removal of P with HAIX Fe followed by regeneration with a caustic brine and precipitation by addition of chemicals to the waste regeneration solution compared to direct precipitation which simply requires addition of chemicals to the initial waste stream P recovery as struvite using HAIX Fe also requires addition of NH 4 + which is not required in ureolyzed urine and some anaerobic digester waste streams. Despite the complications of the HAIX Fe process there is the potential to produce higher purity struvite since the waste regeneration solution will not contain Ca 2+ heavy metals, or pathogens. Due to gaps in the literature the implementation of highly efficient engineering strategies for P recovery using HAIX Fe resin from s ource separated human urine and anaerobic digester
54 filtrate is not achievable at this time. Presently there is no reported comparison of direct precipitation of struvite like minerals within a waste stream with precipitation in the waste regeneration solut ion from the HAIX Fe process when considering efficiency and chemical separated urine using HAIX Fe resin in batch experiments but there is no data on continuous flow column operation with subsequent regeneration and precipitation of struvite when treating human urine. There have been studies on struvite precipitation within regeneration solutions during the HAIX Fe process es (Kumar et al., 2007; Sengupta et al., 2011) but no data on chemical addition requirements such as pH adjustment using HCl or NaOH and Mg 2+ and NH 4 + requirements when compared to direct precipitation have been reported in the literature In order to address the current gaps in knowledge of P recovery the goal of this study was to determine the most efficient approach to P recovery from source separated human urine and anaerobic digester filtrate utilizing HAIX Fe resin under continuous flow column operation. Four specific objectives were defined in order to accomplish this goal: (1) evaluate the capacity of HAIX Fe resin for phosphate under continuous flow column operation; (2) evaluate regeneration efficiency of HAIX Fe resin using a caustic brine under continuous flow column operation and including brine recycling; (3) compare mineral precipitation in waste streams (direct) and waste regeneration solutions (indirect); (4) compare P recovery utilizing a mass balance considering all chemical inputs and amount of P recovered. 3.2 Experimental 3.2.1 Waste Streams Three waste streams were used in this study: fresh urine, ureolyzed urine and anaerobic digester filtrate. Synthetic urine was used in all experiments and the compositions of the fresh
55 and ureolyzed can be found in Table 3 1. Synthetic urine was chosen due its similar precipitation dynamics when compared to real urine (Ronteltap et al., 2007) In addition, compara ble sorption results have been demonstrated for the removal of ammonium in synthetic urine (Lind et al., 2000) and real urine (Beler Baykal et al., 2009) The compositions of fresh and ureolyzed urine were based on previous studies (Ro nteltap et al., 2007; Wilsenach et al., 2007; Udert and Wchter, 2012) The composition of fresh urine assumed that no urea hydrolysis has occurred while the ureolyzed urine assumed complete urea hydrolysis with subsequent precipitation of struvite and hydroxyapatite. Urea hydrolysis occurs due to urease positive bacteria (Mobley and Hausinger, 1989) assumed to be present in urine collection systems and causes an increase in ammonium, bicarbonate and pH thereby causing precipitation of struvite and hydroxyapatite (Udert et al., 2003) The anaerobic digester filtrate was collected from the belt press filtrate of anaerobically digested sludge from the Howard F. Curren Advanced Wastewater Treatment Plant in Tampa, Florida and is the only waste st ream produced from the anaerobic digesters at this plant. The anaerobic digesters are operated at 36.7 C with a typ ical solids residence time of 15 25 d and a combination of mixed sludge and waste activated sludge. When the anaerobic digester filtrate was received it was immediately filtered in series through filters with pore sizes of 5, 3 and 0.45 (Millipore) The characteristics of the anaerobic digester filtrate can be found in Table 3 2. The anaerobic digester filtrate was spiked with NaH 2 PO 4 to b ring the P concentration to ~80 3.2.2 Column Experiments and Regeneration The HAIX Fe resin used in this work is commercially available product (trade names PhosX np or LayneRT manufactured by Solmete X, Northborough, MA) There are procedures in
56 the peer reviewed literature to prepare HAIX Fe resin ( Pan et al., 2009 ) The HAIX Fe resin has is infused with hydrous ferric oxide nanoparticles so that phosphate is selectively removed due to ligand exchange which occurs between iron (Lewis acid) and phosphate (Lewis base) creating an inner sphere complex (Stumm and Morgan, 1996). Breakthrough curves for fresh and ureolyzed urine and anaerobic digester filtrate were characterized using continuous flow column experiments where breakthrough was defined when C/C 0 0.1 and exhaustion was defined when influent P concentration was equal to the effluent (C/C 0 1). G lass columns (Omnifit) were used with a 10 mm inner diameter and an adjustabl e column endpiece. The column was operated up flow using a peristaltic pump (Masterflex) with size 13 tubing at a flow rate of 2.5 mL/min, an empty bed contact time (EBCT) of 3.2 min and a superficial linear velocity (SLV) of 3.1 cm/min. The BV of all col umns was 8 mL of wet settled HAIX Fe resin which was measured using a graduated cylinder. All column experiments were performed at laboratory temperatures 24 C 1C. Effluent samples were collected at predetermined time intervals in order to characteriz e the breakthrough curve to exhaustion. An initial column experiment was performed in triplicate with fresh urine in order to ensure quality control and reproducibility where the relative standard deviation (RSD = standard deviation divided by mean) of mea sured phosphate concentrations were < 10% for 70% of all samples and samples exceeding an RSD of 10% at lower concentrations where small differences in concentration resulted in larger RSD values. In subsequent experiments the resin was used to treat each waste stream three times: first with fresh HAIX Fe resin second with HAIX Fe resin regenerated with ~9 BVs of fresh regeneration solution, and third with HAIX Fe resin regenerated with ~9 BVs of once used regeneration solution. T he column containing HAIX Fe resin was rinsed with DI water (~10 BVs) after each break through and regeneration experiment.
57 Regeneration was performed in the same column used to characterize breakthrough using a 2.5% NaCl and 2% NaOH by weight solution (pH ~13.7) based on previous work done by Sengupta and Pandit (2011). Effluent samples were taken at various time intervals in order to characterize the entire elution curve for all regeneration tests. An initial regeneration test was performed in triplicate o n the HAIX Fe that had been previously exhausted from fresh urine breakthrough experiments where operating conditions were the same as previously mentioned and RSD values were < 10% for 77% of the samples tested and no sample > 20%. Higher RSD values occur red at the beginning and end of the elution curve where P concentrations are low and small differences result in larger RSD values. In order to decrease the BVs required to elute all of the P from the HAIX Fe resin, the flow rate was lowered to 0.8 mL/min, which corresponded to EBCT of 10 min and SLV of 1.0 cm/min for all subsequent regeneration tests. The regeneration solution was recycled twice for all waste streams tested as described in the previous paragraph, using the lower flow rate. 3.2.3 Precipi tation Experiments Precipitation was performed directly in all waste streams and in waste regeneration solutions after being recycled twice. Two minerals were precipitated in this study: struvite and potassium struvite. Potassium struvite was precipitated in fresh urine and twice recycled regeneration solutions used to regenerate HAIX Fe exhausted with P from fresh and ureolyzed urine. Struvite was precipitated in ureolyzed urine, anaerobic digester filtrate and twice recycled regeneration solutions used to regenerate HAIX Fe exhausted with P from ureolyzed urine and anaerobic digester filtrate. Struvite was not precipitated in fresh urine because it would be im practical to add ammonium to a solution containing high nitrogen (N) content (~7000 mg N/L) in t he form of urea Potassium s truvite precipitation was performed by adding chemicals so that
58 the solution contained a 1.5:1.5:1 molar ratio of K:Mg:P. Struvite precipitation used 1.5:1.5:1 molar ratio for Mg:NH 4 + :P. The molar ratios were based on previous studies (Wilsenach et al., 2007) and were intended to be in excess of what was required in order to ensure maximum precipitation. Precipitation was performed in 125 mL amber bottles with 50 mL of solution where after addition of chemicals the pH was adjusted to 9.3 with 1 M NaOH or 12.1 M HCl and stirred in an Innova 2000 Platform Shaker at 200 rpm for 2 h. The pH adjustment was determined through previous studies (Wi lsenach et al., 2007) and the time of mixing was based on Sakthivel et al. (2012). After precipitation the solution was filtered through a 1.5 micron glass fiber filter (Whatman TM 934 AH TM RTU) and dried in a dessication chamber for a minimum of 48 h. 3.2 .4 Analytical Methods All stock solutions and synthetic urine was prepared with DI water and chemicals of ACS reagent grade purity in volumetric flasks. Phosphate was measured the same as in previous stud y following Standard Method 4500 P (Eaton et al., 2005) using a Hitachi U 2900 spectrophotometer at 880 nm and a 1 cm quartz cuvette. Samples were diluted with DI water to ensure P concentrations were within the calibration curve of 0. 15 1. 2 mg P/L. Standard calibrat ion checks and matrix spikes were performed to ensure accuracy where the relative difference between measured and known concentrations w as < 5% for all standard checks and matrix spike recoveries were between 94 106%. The pH of each sample was measured wit h an Accumet AB 15 pH meter and Accumet combination pH/temperature electrode (Fischer Scientific), which was calibrated with buffer solutions of pH 4, 7, and 10 (Fisher Scientific) prior to use. Dissolved organic carbon (DOC) in the anaerobic digester filt rate was analyzed using a Shimadazu TOC V CPH total organic carbon analyzer with an ASI V
59 autosampler (Apell and Boyer, 2010) Inorganic a nions and cations in the anaerobic digester filtrate excluding phosphate were measured using a Dionex ICS 3000 ion chromatograph described elsewhere (Indarawis and Boyer, 2012). X ray diffraction (XRD) of mineral precipitates was performed as in Harris and White (2008) with a computer controlled X ray powder diffractometer with a stepping motor and graphite crystal monochrometer. 3.3 Results and Discussion 3.3. 1 Column Mode Phosphate Sorption Breakthrough curves for P removal during continuous flow column operation with HAIX Fe resin are shown in Fig ure 3 1 and Figure 3 2. The breakthrough curve in Fig ure 3 1 was performed with fresh urine in triplicate, to ensure good precision, and was reported as an average with error bars of the upper and lower bounds of standard deviation. The change in pH is also shown in Fig ure 3 1 where higher pH values (~9) are observed with greater P removal (C/C 0 < 0.1) due to liga nd exchange of hydroxide with phosphate In Fig ure 3 2, BVs to breakthrough (C/C 0 0.1) were ~22, ~4, and ~2.5 for anaerobic digester filtrate, fresh urine, and ureolyzed urine. The lower breakthrough in ureolyzed urine may be due to higher pH (9.3) where the optimum pH is ~7 (Pan et al., 2009). The BVs to saturation from greatest to least was anaerobic digester filtrate > ureolyzed urine > fresh urine using fresh HAIX Fe resin and values of ~50, ~12, and ~11. The BVs to saturation for the anaerobic diges ter filtrate was reasonable when compared to sludge liquor (~25 BVs) containing 472 mg P/L (Bottini et al., 2011) and RO concentrate (~200 BVs) with 10 mg P/L (Kumar et al., 2007). Both fresh and ureolyzed urine demonstrated much lower BVs to saturation th an typical column experiments using HAIX Fe resin ; Martin et al. (2009) showed that saturation for HAIX Fe resin was ~1000 BVs when treating wastewater (5 mg P/L) and Sengupta et al. (2011) showed saturation at ~600 BVs with a
60 synthetic solution containing 4.25 mg P/L. The comparatively low BVs to saturation is due to the high concentrations of P in fresh and ureolyzed urine (620 and 422 mg P/L respectively). Despite HAIX Fe resin required for P recovery from urine (0.22 L per person per day) was less than secondary wastewater effluent (0.71 L per person per day). The low BVs to saturation for urine and anaerobic digester filtrate coul d result in atypical reactor design with longer EBCT when compared to typical EBCTs (1 5 min). For example, treating 1.1 liters per person per day of once per day would req uire a bed volume of 22 mL resin per person and if treated for a full 24 h would have an EBCT ~ 29 min. The mass of P loaded onto the HAIX Fe resin and resin capacities are found in Table 3 3. The far right column of Table 3 3 contains capacities determine d by batch equilibrium tests and Fe was calculated by subtracting the P measured within the effluent of each waste stream after treatment from the initial P within the influent (influent P effluent P). The P loading onto the fresh HAIX Fe resin for each waste stream tested from greatest to least was as follow: fresh urine > anaerobic digester filtrate > ureolyzed urine. The P loading on the HAIX Fe resin after being regenerated with fresh regeneration solution was very similar to the performance of fresh resin, however the P loaded onto the resin decreased when regenerated with recycled regeneration solution (2 regen in Table 3 3), which was likely due to the presence of P and a lower pH within the regeneration solution which agrees with Sengupta et al. (2011). It was expected that fresh urine would exhibit the greatest capacity due to the high P concentration (620 mg P/L) and favorable pH of 6 (Pan et al., 2009). The higher cap acity of the anaerobic digester filtrate
61 compared to ureolyzed urine may be a result of a more favorable pH when compared to ureolyzed urine (pH of 7 and 9.3 respectively) despite the ureolyzed urine having a much higher P concentration (420 mg P/L compare d to 80 mg P/L). The change in capacity with pH is due to speciation of the hydrous ferric oxide in the resin which shifts from positively charged 2 + ) with increasing pH (Morel and Hering, 1993) and fo 2 PO 2 ) in fresh 3 2 ) in ureolyzed urine (Blaney et al., 2007; Zeng et al., 2008). All HAIX Fe resin capacities are reasonable (Table 3 3) and Martin et al. (2009) exhibiting 7.6 mg P/g resin when treating wastewater (5 mg P/L). The HAIX Fe resin performed less favorably compared to Zr(IV) functionalized graphite oxide which demonstrated a maximum P loading of ~16 mg P/g resin treating water with 80 mg P/L at pH 6 (Zong et al., 2013), however HAIX Fe demonstrated a greater P capacity compared to HAIX resin loaded with Cu(II) which exhibited a loading of ~2 mg P/g resin for water containing 18 mg P/L at pH 7 (Zhao and Sengupta, 1998). 3.3.2 Regeneration Efficiency Elution curves for HAIX Fe regeneration under continuous flow fixed bed co lumn operation are shown in Figures 3 3 and 3 4. The elution curve in Fig 3 3 was performed as a quality control experiment to ensure reproducible results using HAIX Fe exhausted with P from fresh urine performed in triplicate and reported as average values with error bars showing one standard devia tion. The elution curves in Figure 3 4 display regeneration u sing fresh, once used, and twice used regeneration solution where the mass of P recovered can be found in Table 3 3. Subsequent reuse of the regeneration solution resulted in higher P concentration at the end of
62 each elution curve and lower regeneration ef ficiencies (Table 3 4). Regeneration efficiencies using fresh regeneration solution were all > 94% with the majority of P desorption occurring within the first 5 BVs where regeneration efficiency was determined by dividing the amount of P collected within the waste regeneration solution by the amount of P initially on the HAIX Fe resin. These regeneration efficiencies are in agreement with other literature reporting typical efficiencies of 98% (Sengupta et al., 2011), 90% (Kumar et al., 2007), and 95% (Bott ini and Rizzo, 2011). Sengupta et al. (2011) demonstrated 93% regeneration efficiency after one recycling of regeneration solution, which is higher than the values in Table 3 4 and was likely due to the precipitation of P minerals and addition of 1.5% NaOH to compensate for the loss of OH Chemicals were not added to the recycled regeneration solution in this study because an objective was to account for chemicals added during the precipitation process, which could have been affected by these steps. The pH decreased slightly with each reuse of the regeneration solution owing to the ligand exchange mechanism (Table 3 4), which would likely have an adverse impact on regeneration efficiency. The P concentration in the regeneration solutions increased with reus e which is more favorable for precipitation where the final waste regeneration solution from anaerobic digester filtrate was ~700 mg P/L compared to ~80 mg P/L within the actual waste stream. However, when comparing the waste regeneration solutions from ur eolyzed urine (~650 mg P/L) and fresh urine (~780 mg P/L) the increase in concentration from the original waste streams was not as substantial (~420 and ~620 mg P/L, respectively). Reuse of regeneration solutions may pose difficulties in implementing full scale operations at WWTPs due to the lower regeneration efficiencies and multiple chemical adjustments. Precipitation of struvite like minerals would require pH adjustment of the regeneration solution below pH 10 to favor struvite or potassium struvite ove r magnesium
63 hydroxide. Following precipitation the pH of the regeneration solution would need to be raised to allow for reuse and efficient phosphate desorption. If pH adjustment is not feasible after the precipitation step the waste regeneration solution could be disposed of or potentially used to regenerate other ion exchange resins which utilize the mobile counter ions Na + or Cl 3.3.3 Mineral Precipitation P recipitation results (Table 3 5) exhibited P recovery of > 96% for all samples with precipitation of struvite yielding the greatest P recovery (> 99%) where P recovery was calculated by the difference of the initial (before precipitation) and final (after precipitation) P concentrations in solution divided by the initial P concentration for each waste stream and waste regeneration solution. Anaerobic digester filtrate demonstrated the lowest final P concentration after precipitation (0.51 mg P/L); this was likely due to the presence of Ca 2+ in solution causing add itional precipitation of P minerals such as hydroxyapatite. Results agree with previous literature where P recovery from urine can be as high as 99% for struvite precipitation and 95% for potassium struvite precipitation (Wilsenach et al., 2007) and P reco very up to 95% for struvite crystallization at an anaerobic digestion pilot plant (Pastor et al., 2010). Slightly higher recovery of struvite occurs due to the differences in solubility where pK sp ~13.2 for struvite and ~ 10.6 for potassium struvite (Taylo r et al., 1963). XRD results (Appendix) confirmed the precipitates to contain struvite or potassium struvite. Direct precipitation within each waste stream yielded lower final P concentrations compared to precipitation within regeneration solutions, for e xample fresh urine had a final P concentration of 15 mg/L after precipitation of potassium struvite and the waste regeneration solutions exhausted with P from fresh urine and ureolyzed urine exhibited 16.8 and 21.3 mg/L after precipitation of potassium str uvite, respectively. The lower final P concentration in fresh urine relative to ureolyzed urine was
64 possibly due to the presence of Ca 2+ which can cause hydroxyapatite precipitation as was assumed with the anaerobic digester filtrate. The lower P concentra tions could be advantageous considering the final disposal of the waste streams, e.g., averting eutrophication in surface waters where P is the limiting nutrient. However, the different P concentrations could be negligible when diluted with other high volu me waste streams at a WWTP which could result in dilution The effluent P concentration f rom HAIX Fe column mode sorption attained values < 0.15 mg/L (lowest point of the calibration curve) which is lower than any of the final P concentrations after direct precipitation in each waste stream. The effluent of the waste streams treated with HAIX Fe or after direct precipitation could be sent to a WWTP with much l ower P concentrations. However, management of the waste regeneration solution after precipitation may require proper disposal due to high salt content (2.5% NaCl). Additional studies on regeneration solutions excluding NaCl could circumvent this problem since on ly OH is required to regenerate ligand exchange sites. 3 .3.4 Mass Balance on P Recovery and Chemical Addition All required chemical additions for precipitation of struvite minerals within each waste stream and waste regeneration solution is displayed in Table 3 6. The percentage of P in solids was calculated by dividing the mass of P recovered by precipitation per unit volume by the mass of solid recovered by precipitation per unit volume. Fresh urine and anaerobic digester filtrate required addition of 1 M NaOH to bring to a pH of 9.3 while ureolyzed urine required no pH adjustment due to the rise in pH that occur s during urea hydrolysis. The anaerobic digester filtrate required the least amount of magnesium per mass of P produced, however had the lowest percentage of P within the solids (7.2% by mass ) which indicates additional mineral precipitation
65 beyond struvit e since the P content in pure struvite is 12.6% by mass Fresh and ureolyzed urine both required less chemical addition when compared to the waste regeneration solutions. Struvite precipitation within ureolyzed urine is likely the most favorable process si nce the only chemical required is magnesium. If potassium recovery is desired then fresh urine would be the most favorable stream for precipitation of potassium struvite due to the lower chemical addition required when compared to the waste regeneration so lutions. All waste regeneration solutions required similar additions of 12.1 M HCl (~53 mL/L) and yielded minerals with P content comparable to potassium struvite (11.9% by mass ) and struvite (12.6% by mass ). The mass of solids and P collected per unit vol ume did not include the entire volume of urine treated but instead the volume of waste regeneration solution used for precipitation. This study did not focus on maximizing P recovery from each waste stream since HAIX Fe since an objective was to saturate t he HAIX Fe resin with P and to recycle the regeneration solution. However, total P recovery could be maximized by performing column experiments to breakthrough as opposed to saturation and utilizing a fresh regeneration solution. Due to the ability to eas ily precipitate P minerals within urine, HAIX Fe resin is a less favorable approach for P recovery when considering the process complexity and additi onal chemicals required (Figure 3 5). However, there are potential hazards associated with direct precipitation in human urine. Decrey et al. (2011) found that struvite made from urine contained viable Ascaris eggs and infective phages even after several days of drying. The HAIX process ma y circumvent this contamination by separating P from the waste stream and recovering it in a caustic brine. HAIX Fe process is considered favorable for P recovery from anaerobic digester filtrate since direct precipitation produced a less pure mineral in t he study and previous literature ( Mnch and Barr, 2001) However, if the goal i s to simply remove P to prevent fouling due to
66 precipitation during sludge management processes at a WWTP then direct precipitation would be favored over HAIX Fe due to lower ch emical addition requirements. Similar to this, a study by Lew et al. (2011) found that struvite precipitation of belt press filtrate from an anaerobic digester can be a solution to prevent clogging due to precipitation in pumps and pipes during sludge dewa tering processes but produces precipitates that could not be used for land application in agriculture due to impurities. The HAIX Fe process may also be more favorable for P recovery from wastewater since the P concentration is typically too low (~1 10 mg P/L) for efficient struvite precipitation. Future studies are planned to investigate P recovery from wastewater using HAIX Fe resin The biggest challenge in P recovery from waste streams is undoubtedly cost which currently cannot rival the low cost of f ertilizer production from phosphate rock (Etter et al., 2011). Etter et al. (2011) found that reactor costs can be inexpensive; however costly magnesium sources are the cause for the high price of struvite precipitation. Alternative magnesium sources such as wood ash have been investigated by Sakthivel et al. (2012) but were not financially viable and produced poor quality struvite. Magnesium oxide is another lower cost magnesium source that has the potential to produce high quality struvite within urine (W ilsenach et al., 2007). Molinos et al. (2011) showed that P recovery from wastewaters are feasible if external benefits are considered, such as the environmental benefits of reducing P loading to surface waters. Despite the issue of cost eventually a sust ainable source of P for fertilizer production is required to counteract the depletion of phosphate rock. 3.5 Comparison of Hybrid Anion Exchange with Direct Precipitation Summary P can be recovered efficiently (> 96%) as struvite within the waste regeneration solution used to regenerate HAIX Fe resin exhausted with fresh urine, ureolyzed urine, and anaerobic digester filtrate as well as through direct precipitation within each individual waste stream by addition of appropriate chemicals.
67 P recovery as struvite is more favorable by direct precipitation within fresh and ureolyzed urine when compared to the waste regeneration solution at the end of the HAIX Fe process due to the additional chemicals required for precipitation in the regeneratio n solution. P recovery from anaerobic digester filtrate is recommended within the waste regeneration solution at the end of the HAIX Fe process, despite higher chemical addition requirements, because the final P mineral is higher quality struvite than form ed by direct precipitation. Regeneration of HAIX Fe resin exhausted with fresh urine, ureolyzed urine, and anaerobic digester filtrate using caustic brine solution (2.5% NaCl, 2% NaOH by mass ) yielded high regeneration efficiencies (> 94%) with diminishin g efficiency as the solution was recycled. Recycling of the regeneration solution resulted in higher P concentrations in the regeneration solution than the original waste stream, which could be utilized for P recovery of waters with lower P concentrations where direct precipitation is not feasible (e.g., wastewater effluent, greywater, and nutrient rich surface waters). The HAIX Fe resin capacities during continuous flow fixed bed column operation are comparable to capacities calculated by Freundlich isoth erm models from batch equilibrium studies.
68 Table 3 1. Chemical compositions of synthetic fresh and ureolyzed urine. Ureolyzed urine composition assumed complete urea hydrolysis with subsequent precipitation of calcium and magnesium with phosphate. Chemical a Urine Fresh (m mol/L ) Ureolyzed (m mol/L ) Urea N 500 NH 3 N b 500 Cl 100 100 PO 4 3 P b 20 13.6 c SO 4 2 15 15 CO 3 2 C b 250 Na + 94 104 K + 40 40 Ca 2+ 4 c Mg 2+ 4 c a Urine was prepared using the following salts: NH 4 Cl, NH 4 HCO 3 urea, NaCl, NaH 2 PO 4 Na 2 SO 4 KCl, CaCl 2 2H 2 O, MgCl 2 6H 2 O. b Total species. c Hydrolysis of urea precipitates calcium and magnesium phosphate minerals.
69 Table 3 2. Measured characteristic of the anaerobic digester filtrate. Parameter mg/L mM DOC a 84 7 Mg 2+ 48 2 Ca 2+ 174 4.3 NH 4 N b 1880 134 K + 73 1.9 Na + 328 14.3 Cl 716 20.2 SO 4 2 254 2.6 PO 4 P b 77 2.5 pH 7 c a Dissolved organic carbon. b Total species. c Unitless.
70 Table 3 3. P capacity of HAIX Fe resin, mass loading of P on resin and mass of P recovered during regeneration process. Waste stream P on resin, mg P desorbed d mg Capacity a Isotherm capacity b Fresh resin 1 regen 2 regen 1st regen 2nd regen 3rd regen mg g 1 mg g 1 Fresh urine 31.1 33.4 25.9 30.2 15.9 7.2 10.1 10.1 Ureolyzed urine 18.4 18.6 13.6 17.4 15.3 10.1 5.9 5.7 Anaerobic digester filtrate 19.9 18.6 16.1 18.9 14.4 12.1 6.4 5.2 c a Calculated by dividing the P loaded on the fresh resin by the volume of the bed multiplied by the density of the resin (mass of dry resin over volume of wet settled resin). b 2013). c Synthetic anaerobic digester supernatant. d Mass of P desorbed by each regeneration cycle.
71 Table 3 4. Regeneration of HAIX Fe exhausted with various waste streams. Waste stream Regeneration cycle PO 4 P a mg/L Regeneration efficiency pH Fresh urine 1 419 97% 13.1 2 669 47% 12.9 3 783 28% 12.8 Ureolyzed urine 1 241 94% 13.2 2 492 82% 13.1 3 652 74% 13.0 Anaerobic digester filtrate 1 263 95% 13.2 2 502 77% 13.1 3 693 76% 13.0 a Total species.
72 Table 3 5. Recovery of P through struvite precipitation directly in each waste stream compared to precipitation in regeneration solution from the HAIX Fe process. Precipitation method Waste Stream Mineral Initial PO4 P a mg/L Final PO4 P a mg/L P recovered Direct Precipitation Fresh urine KMP 674 15 b 97.8% Ureolyzed urine MAP 487 1.4 b 99.7% Anaerobic digester filtrate MAP 78 0.51 b 99.3% Precipitation in regeneration solution Fresh urine KMP 783 16.8 97.9% Ureolyzed urine KMP 652 21.3 96.7% Ureolyzed urine MAP 652 5.2 99.0% Anaerobic digester filtrate MAP 693 1.2 99.8% a Total species. b Average of triplicate samples.
73 Table 3 6. Chemicals added to each waste stream and waste regeneration solution to precipitate KMP or MAP. NaOH and HCl were added to reach a pH of 9.3. Solid collected and P recovered was from the waste stream or waste regeneration solution Waste stream 1M NaOH 12.1 M HCl MgCl 2 *6H 2 O KCL NH 4 Cl Solid collected P recovered P in solid mL/L mL/L g/g P e g/g P e g/g P e g/L g/L % UF a KMP 28 8 5.3 0.66 12.5% UH b MAP 8.8 3.7 0.49 13.0% ADF c MAP 36 4.5 1.1 0.078 7.2% UF a R d KMP 54 10.0 3.7 6.2 0.77 12.3% UH b R e KMP 54 8.6 3.1 5.2 0.63 12.1% UH b R e MAP 52 8.4 2.2 5.1 0.65 12.6% ADF c R f MAP 52 8.3 2.2 6.0 0.69 11.4% a Fresh urine. b Ureolyzed urine. c Anaerobic digester filtrate. d Twice recycled waste regeneration solution. e Grams of chemical added to produce one gram of P.
74 Figure 3 1. Normalized effluent P concentration and effluent pH for HAIX Fe resin column operation treating fresh urine (1 bed volume = 8 mL resin) Results are mean of triplicate samples with error bars showing one standard deviation. 0 2 4 6 8 10 0 0.2 0.4 0.6 0.8 1 0 5 10 15 20 25 pH C/C 0 Bed Volumes C/C0 pH Flow: EBCT: 3.2 min SLV: 3.1 cm/min Influent: pH: 6 PO 4 P: 20 mM
75 Figure 3 2. Normalized e ffluent P concentration (C/C 0 ) for HAIX Fe resin column experiments for a) fresh urine b) ureolyzed urine and c) anaerobic digester filtrate. One bed volume equal to 8 mL resin. 0 0.2 0.4 0.6 0.8 1 0 4 8 12 16 C/C 0 Bed Volumes Fresh Resin 1X Regenerated 2X Regenerated Flow : EBCT: 3.2 min SLV: 3.1 cm/min Influent : pH: 6 PO 4 P: 20 mM a 0 0.2 0.4 0.6 0.8 1 0 4 8 12 16 20 C/C 0 Bed Volumes Fresh Resin 1X Regenerated 2X Regenerated Flow: EBCT: 3.2 min SLV: 3.1 cm/min Influent: pH: 9.3 PO 4 P: 13.6 mM b 0 0.2 0.4 0.6 0.8 1 0 10 20 30 40 50 C/C 0 Bed Volumes Fresh Resin 1X Regenerated 2X Regenerated Flow: EBCT: 3.2 min SLV: 3.1 cm/min Influent: pH: 7 PO 4 P: 2.6mM c
76 Figure 3 3. Effluent P concentration with increasing bed volumes for regeneration of HAIX Fe resin exhausted with P from fresh urine. Results are mean of triplicate samples with error bars showing one standard deviation. One bed volume equal to 8 mL resin. 0 400 800 1200 1600 0 5 10 15 20 PO4 P, mg/L Bed Volumes Flow: EBCT: 3.2 min SLV: 3.1 cm/min Influent: 2.5% NaCl, 2% NaOH
77 Fig ure 3 4. Effluent P concentration with increasing bed volumes for regeneration of HAIX Fe resin exhausted with a) fresh urine b) ureolyzed urine and c) anaerobic digester filtrate. F irst regeneration used a fresh regeneration solution ; second and third regeneration were recycled solutions One bed volume equal to 8 mL resin 0 400 800 1200 1600 2000 0 2 4 6 8 10 PO4 P, mg/L Bed Volumes 1st Regeneration 2nd Regeneration 3rd Regeneration Flow: EBCT: 10 min SLV: 1.0 cm/min Influent: 2.5% NaCl, 2% NaOH a 0 400 800 1200 1600 0 2 4 6 8 10 PO4 P, mg/L Bed Volumes 1st Regeneration 2nd Regeneration 3rd Regeneration Flow: EBCT: 10 min SLV: 1.0 cm/min Influent: 2.5% NaCl, 2% NaOH b 0 400 800 1200 1600 0 2 4 6 8 10 PO 4 P, mg/L Bed Volumes 1st Regeneration 2nd Regeneration 3rd Regeneration Flow: EBCT: 10 min SLV: 1.0 cm/min Influent: 2.5% NaCl, 2% NaOH c
78 Figure 3 5. Column set up used for all breakthrough experiments and regenerations with subsequent precipitation of MAP by chemical addition of H + Mg 2+ and NH 4 + to the used waste regeneration solution.
79 CHAPTER 4 CONCLUSIONS AND RECOMMENDATIONS Hybrid anion exchange resin containing hydrous ferric oxide (HAIX Fe) can selectively remove phosphate from various waste streams including: fresh urine, ureolyzed urine, urine diluted with tap water, greywater, urine mixed with greywater, secondary wastew ater effluent, and anaerobic digester supernatant. Greater resin capacities were obtained in more concentrated waste streams (urine and anaerobic digester supernatant) when compared to secondary wastewater effluent. Phosphate sorption on HAIX Fe resin demo nstrated nonlinear relationship between solid phase and solution phase concentration excluding secondary wastewater effluent which exhibited a linear sorption relationship over a small range of low P concentration. Dilution with tap water initially decreas ed phosphate sorption loading on HAIX Fe resin, however was more favorable than secondary wastewater effluent. It is recommended that P removal be performed using waste streams more concentrated in P when considering sorption chemistry where higher resin c apacities can be attained. Phosphate recovery potential is higher in fresh and ureolyzed urine, with the greatest potential in greywater mixed with ureolyzed urine. Due to precipitation of P minerals and potential biological uptake, secondary wastewater e ffluent and anaerobic digester supernatant have lower P recovery potential. Additional research on P recovery from real human urine and various dilutions with tap and greywater is recommended. Continuous flow column operation using HAIX Fe can attain effi cient recovery of P (> 96%) as struvite within the waste regeneration solution with fresh urine, ureolyzed urine, and anaerobic digester filtrate. Low BVs to saturation were observed for urine (~12) and anaerobic digester filtrate (~50). Regeneration effic iency was > 94% for all waste streams tested using a caustic brine solution (2.5% NaCl, 2% NaOH by mass) with diminishing efficiency upon
80 subsequent reuse. Reuse of the waste regeneration solution resulted in higher P concentrations which can be utilized f or P recovery of waters with lower P concentrations where direct precipitation is not feasible (wastewater, greywater, natural waters). When comparing direct precipitation with HAIX Fe processes for P recovery from human urine direct precipitation is recommended due to additional chemical requirements of acid, magnesium, potassium, and ammonium sources. If pathogenic contamination occurs w ithin direct precipitation of struvite with fresh and hydrolyzed urine the HAIX Fe may be a potential solution to this complication. HAIX Fe processes are recommended over direct precipitation within anaerobic digester filtrate due to impure mineral precip itation. Additional research on alternative sources of magnesium and pH adjustment are recommended to lower costs of struvite precipitation.
81 APPENDIX X RAY DIFFRACTION DATA Figure A 1. X Ray diffraction results for struvite precipitation in anaerobi c digester filtrate.
82 Figure A 2. X Ray diffraction results for struvite precipitation in anaerobic digester filtrate waste regeneration solution.
83 Figure A 3. X Ray diffraction results for potassium struvite precipitation in fresh urine.
84 Fig ure A 4. X Ray diffraction results for potassium struvite precipitation in fresh urine waste regeneration solution.
85 Figure A 5. X Ray diffraction results for struvite precipitation in ureolyzed urine.
86 Figure A 6. X Ray diffraction results for potassium struvite precipitation in ureolyzed urine waste regeneration solution.
87 Figure A 7. X Ray diffraction results for struvite precipitation in ureolyzed urine waste regeneration solution
88 LIST OF REFERENCES Apell, J.N., Boyer, T.H., 2010. Combined ion exchange treatment for removal of dissolved organic matter and hardness. Water R esearch 44 (8), 2419 30. Antonini, S., Paris, S., Eichert, T., Clemens, J., 2011. Nitrogen and phosphorus recovery from human urine by struvite precipitation and a ir stripping in vietnam. CLE AN Soil, Air, Water 39, 1099 104. Awual, M.R., Jyo, A., Ihara, T., Seko, N., Tamada, M., Lim, K.T., 2011. Enhanced trace phosphate removal from water by zirconium(iv) loaded fibrous adso rbent. Water Research 45, 4592 600. de B ashan, L., Bashan, Y., 2004. Recent advances in removing phosphorus from wastewater and its future use as fertilizer (1997 2003). Water Research 38(19), 4222 46. Battistoni, P., De Angelis, A., Pavan, P., Prisciandaro, M., Cecchi, F., 2001. Phosphorus remo val from a real anaerobic supernatant by struvite crystallization. Water Research 35, 2167 78. Baykal, B.B., Kocaturk, N.P., Allar, A.D., Sari, B., 2009. The effect of initial loading on the removal of ammonium and potassium from source separated human uri ne via clinoptilolite. Water Scie nce & Technology 60(10), 2515 20. Beler Baykal, B., Kocaturk, N.P., Allar, a D., Sari, B., 2009. The effect of initial loading on the removal of ammonium and potassium from source separated human urine via clinoptilolite. W ater Science & Technology 60 (10), 2515 20. Bell, L.C., Mika, H., Kruger, B.J., 1978. Synthetic hydroxyapatite solubility product and stoichiomet ry of dissolution. Archives of Oral B iology 23, 329 36. B itton, G., 2011. Wastewater Microbiology, 4th Edition. Wiley Blackwell, Hoboken, New Jersey. Blaney, L.M., Cinar, S., SenGupta, A.K., 2007. Hybrid anion exchanger for trace phosphate removal from water and wastew ater. Water Research 41, 1603 13. Boesch, D.F., Brinsfield, R.B., Magnien, R.E., 2001. Chesapeake Bay eutrophication: scientific understanding, ecosystem restoration, and challenges for agriculture. Environmental Quality 30, 303 20. Bottini, A., Rizzo, L., 2012. Phosphorus recovery from urban wastewater treatment plant sludge liquor by ion exchange. Se paration Science and Technology 47, 613 20. Boyer, T.H., Persaud, A., Banerjee, P., Palomino, P., 2011. Comparison of low cost and engineered materials for phosphorus removal from organic rich surface w ater. Water Research 45, 4803 14.
89 Bradford Hartke, Z., Lant, P., Leslie, G., 2012. Phosphorus recovery from centralised municipal water recycling plants. Chemical Engineering Research and Design 90, 78 85. Cabeza, R., Steingrobe, B., Rmer, W., Claassen, N., 2011. Effectiveness of recycled P products as P fer tilizers, as evaluated in pot experiments. Nutrient Cycling in Ag roecosystems 91, 173 84. Carey, R.O., Migliaccio, K.W., 2009. Contribution of wastewater treatment plant effluents to nutrient dynamics in aquatic systems: A review. En vironmental Management 44, 205 17. Childers, D.L., Corman, J., Edwards, M., Elser, J.J., 2011. Sustainability challenges of phosphorus and food: Solutions from closing the human phosph orus cycle. BioScience 61, 117 24. Chung, J., van Mastrigt, R., 2009. Age and volume dependent normal frequency volume charts for healthy male s. Journal of Urology 182, 210 4. Cordell, D., Rosemarin, A., Schrder, J.J., Smit, A.L., 2011. Towards global phosphorus security: A systems framework for phosphorus recovery and reus e options. Chemosphere 84 747 58. Cumbal, L., Sengupta, A.K., 2005. Arsenic removal using polymer supported hydrated iron(iii) oxide nanoparticles: Role of donnan membrane effect. Environmental Science & Technology 39, 6508 15. Daiper, C., Toifl, M., Storey, M., 2008. Greywater t echnology testing protocol. CSIRO: Water for a Healthy Country National Research Flagship. ISSN: 1835 095X. Decrey, L., Udert, K.M., Tilley, E., Pecson, B.M., Kohn, T., 2011. Fate of the pathogen production of struvite fertilizer from source separated urine. Wa ter R esearch 45 (16), 4960 72. Doyle, J.D., Oldring, K., Churchley, J., Parsons, S. a, 2002. Struvite formation and the fouling propensity of different materials. Water R esearch 36 (16), 3971 8. Dziegielewski, B., Kiefer, J.C., 2010. Appropriate design and evaluation of water use and conservation metrics and benchmarks. Journal American Water Works Association 102, 1 62. Eaton, A.D., Clesceri, L.S., Rice, E.W., Greenberg, A.E., 2005. Standard methods for the examination of water and wastewater, 21st ed. APHA, AWWA, WEF. Etter, B., Tilley, E., Khadka, R., Udert, K.M., 2011. Low cost struvite production using source separated urine in nepal. Water Research 45, 852 62.
90 Foo, K.Y., Hameed, B.H., 201 0. Insights into the modeling of adsorption isotherm systems. Chemical Engineering Journal 156, 2 10. Geelhoed, J.S., Hiemstra, T., Van Riemsdijk, W.H., 1997. Phosphate and sulfate adsorption on goethite: Single anion and competitive adsorption. Geochimica et Cosmochimica Acta 61, 2389 96. Gregory, J., Dhond, R.V., 1972. Anion exchange equilibria involving phosphate, sulphate, and chloride. Water Research 6, 695 702. Guest, J.S., Skerlos, S.J., Barnard, J.L., Beck, M.B., Daigger, G.T., Hilger, H., Jackson, S.J., Karvazy, K., Kelly, L., Macpherson, L., Mihelcic, J.R., Pramanik, A., Raskin, L., Van Loosdrecht, M.C.M., Yeh, D., Love*, N.G., 2009. A New Planning and Design Paradigm to Achieve Sustainable Resource Recovery from Wastewater. Environmental Science & Technology 43 (16), 6126 30. Gustafsson, J.P., ver. 3.0. 2012. Visual minteq. retrieved from: http://www2.lwr.kth.se/English/OurSoftware/vminteq/ Hao, X. D., Wang, C. C., Lan, L., Van Loosdrecht, M.C.M., 2008. Struvite formation, analytical International Association on Water Pollution Research 58 (8), 1687 92. Harris, W.G., White, G.N., 2008. X ray diffr action techniques for soil mineral identification. In: Ulery, A., Drees, R. (Eds.), Methods of S oil Analysis: Part 5 Mineralogical Methods, Soil Sc i. Soc. Am. Madison, WI. pp. 81 115. Helfferich, F., 1995. Ion exchange. Dover Publications, Inc., New York Houhou, J.; Lartiges, B. S.; Hofmann, A.; Frappier, G.; Ghanbaja, J.; Temgoua, A., 2009. Phosphate dynamics in an urban sewer: A case study of Nancy, Franc e. Water Research 43 (4), 1088 100. Hug, A., Udert, K.M., 2013. Struvite precipitation from urine w ith electrochemical magnesium dosage. Water Research 47, 289 99. Indarawis, K., Boyer, T.H., 2012. Alkaline earth metal cation exchange: effect of mobile counterion and dissolved organic matter. Environmental Science & T echnology 46 (8), 4591 8. Kocatrk, N.P., Baykal, B.B., 2012. Recovery of plant nutrients from dilute solutions of human urine and preliminary investigations on pot trials. CL EAN Soil, Air, Water 40, 538 44. Kujawa Roeleveld, K., Zeeman, G., 2006. Anaerobic treatment in decentralised and source separation based sanitation concepts. Reviews in Environmental Sc ience and Biotechnology 5, 115 39.
91 Kumar, M., Badruzzaman, M., Adham, S., Oppenheimer, J., 2007. Beneficial phosphate recovery from reverse osmosis (ro) concentrate of an integrated me mbrane system using polymeric ligand exchanger ( ple). Water Research 41, 2211 9. Lahav, O., Green, M., 2000. Bioregenerated ion exchange process: the effect of the biofilm on the ion exchange capacity an d kinetics. WaterSA 26 (1), 51 8. Lamichhane, K., Bab cock, R., 2012. An economic appraisal of using source separation of human urine to contain and treat endocrine disrupters in the USA. Journal of Envi ronmental Monitoring 14, 2557 65. Larsen, T.A., Alder, A.C., Eggen, R.I.L., Maurer, M., Lienert, J., 2009. Source separation: Will we see a paradigm shift in wastewater handling? Environmental S cience & Technology 43, 6121 5. Larsen, T.A., Gujer, W., 1996. Separate management of anthropogenic nutrient solutions (human urine). Water Science & Technology 34, 87 9 4. Le, C., Zha, Y., Li, Y., Sun, D., Lu, H., Yin, B., 2010. Eutrophication of lake waters in China: Cost, causes, and control. Environmental Management 45, 662 8. Lew, B., Phalah, S., Rebhum, M., 2011. Controlled struvite precipitation from belt press filt rate of anaerobic digester in a cstr. Environmental Progress & Sustainable Energy 30, 640 7. Lienert, J., Larsen, T.A., 2009. High acceptance of urine source separation in seven european countries: A review. Environmenta l Science & Technology 44, 556 66. L ind, B., Ban, Z., Byden, S., 2000. Nutrient recovery from human urine by struvite crystallization with ammonia adsorption on zeolite and wollastonite. Bior esource Technology 74 (2), 169 74. Lundin, M., Bengtsson, M., Molander, S., 2000. Life cycle assessme nt of wastewater systems: Influence of system boundaries and scale on calculated environmental loads. Environmental Science & Technology 34, 180 6. Martin, B.D., Parsons, S.A., Jefferson, B., 2009. Removal and recovery of phosphate from municipal wastewate rs using a polymeric anion exchanger bound with hydrated ferric oxide nanoparticles. Water Sc ience and Technology 60, 2637 45. Maurer, M., Pronk, W., Larsen, T.A., 2006. Treatment processes for source separated urine. Water Research 40, 3151 3166.Mobley, H .L., Hausinger, R.P., 1989. Microbial ureases: Significance, regulation, and molecular characterization. Mi crobiological Reviews 53, 3151 66 Mobley, H.L., Hausinger, R.P., 1989. Microbial ureases: significance, regulation, and molecular characterization. Microbiological reviews 53 (1), 85 108.
92 Molinos, M., Hernandez, F., Sala R., Garrido, M., 2011. Economic feasibility study for phosphorus recovery processes. Journal of the Human Environment. 40 (4), 408 416. Morel, F.M.M., Hering, J.G., 1993. Principles and applications of aquatic chemistry. John Wiley & Sons, New York. Moriyama, K., Kojima, T., Minawa, Y., Matsumoto, S., Nakamachi, K., 2001. Development of artificial seed crystal for crystallization of calcium phosphate. Environmental Technology 22 (11), 1245 52. Mnch, E. V, Barr, K., 2001. Controlled struvite crystallisation for removing phosphorus from anaerobic digester sidestreams. Water research 35 (1), 151 9. Noe, G., Childer, D., Jones R., 2001. Phosphorus biogeochemistry and the impact of phospho rus enrichment: Why is the Everglades so unique? Ecosystems 4, 603 24. Boyer, T.H. 2013. Phosphate recovery using hybrid anion exchange: Applications to source separated urine and combined wastewate r streams, Water Research http://dx.doi.or g/10.1016/j.watres.2013.05.037 Pan, B.J., Wu, J., Pan, B.C., Lv, L., Zhang, W.M., Xiao, L.L., Wang, X.S., Tao, X.C., Zheng, S.R., 2009. Development of polymer based nanos ized hydrated ferric oxides (HFO s) for enhanced phosphate removal from waste efflue nts. Water Research 43, 4421 9. Pastor, L., Mangin, D., Ferrer, J., Seco, a, 2010. Struvite formation from the supernatants of an anaerobic digestion pilot plant. Bioresource technology 101 (1), 118 25. Ronteltap, M., Maurer, M., Gujer, W., 2007. Struvite precipitation thermodynamics in source separated urine. Water Research 41, 977 84. Rossi, L., Lienert, J., Larsen, T.A., 2009. Real life efficiency of urine source separation. Journal of Envi ronmental Management 90, 1909 17. Sakthivel, S.R., Tilley, E., Udert, K.M., 2012. Wood ash as a magnesium source for phosphorus recovery from source separated urine. The Science of the Total E nvironment 419, 68 75. Schrder, J.J., Smit, A.L., Cordell, D., Rosemarin, A., 2011. Improved phos phorus use efficiency in agriculture: A key requirement for its sustaina ble use. Chemosphere 84, 822 31. Sendrowski, A., Boyer, T.H., 2013. Phosphate removal from urine using hybrid anion exchange r esin. Desalination 322, 104 12. Sengupta, S., Pandit, A., 2011. Selective removal of phosphorus from wastewater combined with its recovery as a solid phase fertil izer. Water Research 45, 3318 30.
93 Seo, G.T., Suzuki, Y., Ohgaki, S., 1996. Biological powdered activated carbon (bpac) microfiltration for wastewater re clamation and reuse. Desalination 106, 39 45. Stratful, I., Scrimshaw, M.D., Lester, J.N., 2001. Conditions influencing the precipitation of magnesium ammonium phosp hate. Water Research 35, 4191 99. Stumm, W., Morgan, J.J., 1996. Aquatic chemistry. John Wi ley & Sons, Inc., New York. Taylor A W Gurney E L Frazier A W. 1963 Solubility pro ducts of magnesiu m ammonium and magnesium potas sium phosphates. Trans Faraday Soc iety 59 1580 4. Udert, K.M., Larsen, T.A., Biebow, M., Gujer, W., 2003a. Urea hydrolysis and precipitation dynamics in a urine collecting sy stem. Water Research 37, 2571 82. Udert, K.M., Larsen, T.A., Gujer, W., 2003b. Estimating the precipitation potential in urine collecting systems. Water Research 37, 26 67 77. Udert, K.M., Wchte r, M., 2012. Complete nutrient recovery from source separated urine by nitrification and distil lation. Water Research 46, 453 64. Ueno, Y., Fujii, M., Three years experience of operating and selling recovered struvite from full scale plant. Environmental T echnology 22, 1373 81 Van Vuuren, D.P., Bouwman, A.F., Beusen, A.H.W., 2010. Phosphorus demand for the 1970 2100 period: A scenario analysis of resource depletion. Globa l Environmental Change 20, 428 39. Weber, W.J., DiGiano, F.A., 1996. Process dynamics i n environmental systems. John Wiley & Sons, Inc., New York. Wei, L.L., Zhao, Q.L., Hu, K., Lee, D.J., Xie, C.M., Jiang, J.Q., 2011. Extracellular biological organic matters in sewage sluge during mesophilic digestion at reduced hydraulic retention time. Wa ter Research 45, 1472 80. Williams, S., 1999. Struvite precipitation in the sludge stream at Slough wastewater treatment plant and opportunities for phosphorus recovery. Env ironmental Technology 20, 743 7. Wilsenach, J.A., Schuurbiers, C.A.H., van Loosdrecht, M.C.M., 2007. Phosphate and potassium recovery from source separated urine through struvite precipi tation. Water Research 41, 458 66. Wilsenach, J.A., van Loosdrecht, M.C.M., 2006. Integration of processes to treat wastewater and source separat ed urine. Journal of Environme ntal Engineering Asce 132, 331 41.
94 Young, K., Morse, G., Scrimshaw, M., Kinniburgh, J., Macleod, C., Lester, J., 1999. The relation between phosphorus and eutrophication in the Thames catchment, UK. The Science of the Total Environment. 228, 157 83. Zeng, H., Fisher, B., Giammar, D.E., 2008. Individual and competitive adsorption of arsenate and phosphate to a high surface area iron oxide based sorbent. Environmental Science & Technology 42, 147 52. Zhao, D., Sengupta, A.K., 1 998. Ultimate removal of phosphate from wastewater using a new class of polymeric ion exchan gers. Water Research 32, 1613 25. Zhu, X.P., Jyo, A., 2005. Column mode phosphate removal by a novel highly selective adsorbent. Water Research 39, 2301 8. Zong, E. Wei, D., Wan, H., Zheng, S., Xu, Z., Zhu, D., 2013. Adsorptive removal of phosphate ions from aqueous solutions using zirconia functionalized graphite oxide. Chemical Engineering Journal 221, 193 203.
95 BIOGRAPHICAL SKETCH the University of Florida, Gainesvill e, FL in May 2011 with a BS in agricultural and biological e ngineering. In January 201 2 he began studies to attain and ME in environmental e ngineering s ciences at the University o f Florida. Upon completion of his ME h e plans to practice environmental engineering for a consulting firm.