Você está na página 1de 11

Chemical Engineering Science 61 (2006) 3951 – 3961

www.elsevier.com/locate/ces

A fed-batch design approach of struvite system in controlled supersaturation


Md. Imtiaj Ali ∗,1 , Philip Andrew Schneider
School of Engineering, James Cook University, Townsville, Queensland, QLD-4811, Australia

Received 25 November 2004; received in revised form 23 August 2005; accepted 20 January 2006
Available online 20 March 2006

Abstract
This paper focuses on struvite (MgNH4 PO4 · 6H2 O) crystallization in controlled supersaturation. Struvite can be used as a slow-release
fertilizer. Crystallization experiments were conducted using supersaturated solutions. The secondary focus of this paper is the design of a
struvite recovery system in fed-batch-controlled supersaturation mode. The design and commissioning of fed-batch struvite crystallization
included the determination of operating supersaturation of struvite crystallization, suitable seed materials and the composition of feed solution.
Determination of operating supersaturation of struvite crystallization was conducted by two steps including thermodynamic simulation using
gPROMS2 (process simulation software) along with a set of batch experiments. Investigation of suitable seed materials was also conducted
using set of batch experiments. Two types of seed materials including quartz sand and struvite seeds were used in the investigation of seed
materials. Composition of feed solution included the investigation of struvite solution chemistry using PHREEQC3 thermodynamic modeling
package. Based on the previously investigated design approach, struvite crystallization in fed-batch system was conducted using a 44-L of
reactor with 15-L of initial reactant volume.
䉷 2006 Elsevier Ltd. All rights reserved.

Keywords: Struvite; Control; Supersaturation; Reactor; Design; Thermodynamics

1. Introduction a need of slow-release fertilizer and has many uses in horticul-


ture such as in nurseries, golf course, etc. as boutique fertilizers
Magnesium ammonium phosphate hexa-hydrate (Mg (Nelson et al., 2000).
NH4 PO4 · 6H2 O), commonly known as struvite, is a threat for The basic technique of growth type crystallizer was derived
nutrient-rich wastewater systems when the nutrient composi- from mixed suspension mixed product removal (MSMPR). The
tion exceeds stable saturation limit (Muramatsu et al., 2000). key endeavor of this paper is to design a crystallization pro-
Crystalline deposits of struvite characteristically form in the cess capable of reducing spontaneous precipitation and allow-
wastewater treatment infrastructure in high turbulence zone. ing crystals to grow. Pilot scale-controlled crystallization, in the
The very sensitive zones of forming struvite are valves, bends presence of suspended seeds, has been investigated in this paper.
in pipe, separating screens, pumps, etc. The mass of crystalline Thermodynamically, the metastable zone is defined as the
deposits formed can be extensive and can lead to operational critical zone of supersaturation of solution where crystalliza-
failure by clogging water distribution pipes. Struvite satisfies tion is not governed by nucleation and thus avoids rapid and/or
spontaneous precipitation. Hirasawa (1996) documented an ex-
perimental approach to determine metastable zone width for
∗ Corresponding author. Tel.:+61 07 4781 43463; fax: +61 07 4775 1184. hydroxyapatite crystallization, relating to solution supersatura-
E-mail address: Md.Ali@jcu.edu.au (Md.I. Ali). tion, pH value, and minimum solubility limit of crystallization.
1 After December 2006, corresponding email is phil.schneider@jcu.edu.au
Crystallization in the metastable zone is heterogeneous. In in-
(Dr. Philip A. Schneider), Tel.: +61 07 4781 5427.
2 gPROMS process simulation software, Process System Enterprise, dustrial crystallization, the metastable zone technique is widely
London, UK. practiced due to smoother continuous operation in controlled
3 US Geological Survey, Hydrologic Analysis Software Support Program, supersaturation (Thaller et al., 1981; McPherson, 1988; Srini-
437 National Centre, Reston VA 20192, E-mail: h2osoft@usgs.gov. vasakannan et al., 2002).
0009-2509/$ - see front matter 䉷 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ces.2006.01.028
3952 Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961

2. Materials and methods Table 1


Thermodynamic equilibria of struvite solution chemistry (Childs, 1970; Smith
and Martell, 1976; Martell and Smith, 1989; Morel and Hering, 1993)
This paper describes the fed-batch struvite crystallization in
controlled supersaturation mode. Prior to the main experiment, Thermodynamic equations Value of K
preliminary experimental observations in batch scale along with
{MgOH+ } ⇔ {Mg2+ } + {OH− } 10−2.56
thermodynamic simulation of struvite were required, which in- {NH+ + 10−0.45
4 } ⇔ {NH3 (aq)} + {H }
cluded the following investigations: 2− +
{HPO4 } ⇔ {H } + {PO4 } 3−
10−12.35
{H2 PO− +
4 } ⇔ {H } + {HPO4 }
2−
10−7.20
• Determination of the metastable zone. {H3 PO4 } ⇔ {H+ } + {H2 PO− 10−2.15
4}
• Determination of the suitable seed materials. {MgH2 PO+ 4 } ⇔ {Mg 2+ } + {H PO− }
2 4 10−0.45
• Determination of suitable composition of feed solutions. {MgHPO4 } ⇔ {Mg2+ } + {HPO2− 4 } 10−2.91
{MgPO− 2+
4 } ⇔ {Mg } + {PO4 }
3−
10−4.80
2.1. Determination of metastable zone

The determination of the metastable zone includes the ther- pH Controller


modynamic modeling along with a set of experimental obser-
vations. The determination of the metastable zone incorporates
(i) determination of minimum struvite solubility limit (satu-
ration) by thermodynamic modeling, (ii) determination of the
minimum limit of spontaneous precipitation using a set of batch
experiments.
pH probe

2.1.1. Determination of minimum struvite solubility limit


The constituent chemicals of struvite are magnesium, am- Laser scattering device
monium and phosphate. Solution chemistry plays an impor- Sharp red line
tant role in struvite formation. In supersaturated solution, stru-
vite forms by chemical reaction of Mg2+ , NH4+ and PO3− 4 as
demonstrated
Mg2+ + NH+ 3−
4 + PO4 + 6H2 O ⇔ MgNH4 PO4 · 6H2 O. Fig. 1. Schematic of experimental setup.
(1)
The key parameters involved in struvite solution chemistry are the absence of seed to identify the minimum pH of spontaneous
solution supersaturation, pH, and concentration of reactants. precipitation at given concentrations. The schematic diagram
Solution, consists of Mg2+ , NH4+ and PO3− 4 , remains in com- of experimental setup is demonstrated in Fig. 1.
plex forms of Mg2+ , MgOH+ , MgH2 PO+ 4 , MgHPO4 , H3 PO4 , Each experiment was conducted in a dark room using a laser
H2 PO− 4 , HPO 2−
4 , PO 3−
4 , MgPO −
4 , NH 3 (Bouropoulos and Kout- light scattering device (laser pointer of helium neon laser light
soukos, 2000). The basic thermodynamic relations of chemi- source) in which red color laser light passed through reactive
cal complexes are listed in Table 1. The detailed description of solution of volume 1 l. The reactive solution, contained in 1.5 l
struvite chemistry is previously published by Ali and Schneider clean glass beaker, was agitated using a rotational mechanical
(2005). agitator of agitator blade diameter 2.5 in. A uniform speed of
Coding of thermodynamic equilibria in gPROMS determines 35 rpm was employed throughout all experiments to provide
the minimum pH limit of struvite saturation at the given range of sufficient mixing of solution. The initial pH of reactive solu-
concentrations of magnesium, ammonium and phosphate. The tion (about 5.38) was slowly brought up to saturation level (in-
result of gPROMS thermodynamic simulation was then vali- vestigated by thermodynamic modeling) using 0.5M of NaOH,
dated by the simple thermodynamic modeling using PHREEQC and afterward 0.1M of NaOH was slowly added until the first
and the derived data from the Ohlinger (Ohlinger, 1999) solu- appearance of crystal cloud in solution. In each experiment, a
bility limit curve. 30 min interval was allowed after each drop addition of NaOH
to observe any alteration of reactive solution. By careful ob-
2.1.2. Determination of minimum limit of spontaneous servation, the first appearance of crystal cloud was detected by
precipitation laser light passing through the reactive solution. Scattering of
The set of experiments were conducted using synthetic so- laser light was applied parallel to the bottom and perpendicu-
lution of 3, 4, 5 and 7 mM solution. Synthetic solution made lar to the surface of reactive solution container. To avoid any
up of de-ionized water was used in this experiment. Analyti- interference, laser light in solution must not touch the agitator
cal grade MgCl2 and NH4 H2 PO4 was used to make up solu- blade.
tion maintaining molar ratio of Mg : NH4 : PO4 as 1:1:1. The In the absence of any solid particles, undersaturated solu-
metastable zone experiment was conducted in batch-scale in tion did not show any distinct redline of laser light, since
Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961 3953

illumination of light depended on the reflection of light onto media and enhances crystal growth by layering of newly born
suspended particles present in solution. A distinct sharp red crystallites onto the surface of the seeds.
line in reactive supersaturated solution was formed due to the Sets of preliminary batch experiments were conducted us-
formation of spontaneous precipitation, which was visible from ing 3, 4 and 7mM concentration of solution with size classified
few meters distance in a dark room. This limit of pH is the (45–63 m) quartz sand and struvite seeds. Each experiment
apparent minimum pH for spontaneous precipitation. was conducted at previously investigated optimal supersatura-
tion (close to saturation as previously investigated by thermody-
2.2. Preparation of seed materials namic simulation) with 12 h operational time. Sieving of seeds
was conducted using ASTM standard sieve. The crystal size
This paper incorporates two types of experiments, which in- distribution (CSD) of seeds as well as growing struvite crystals
clude batch scale and fed-batch pilot scale experiments. Batch was measured using Malvern particle-sizer.
experiments (in the presence of seeds) were conducted to de-
termine the suitable seed materials for struvite crystallization. 2.4. Determination of feed composition
Fed-batch scale crystallization was conducted in controlled su-
persaturation mode in the presence of previously investigated To determine the concentration of different species of mag-
suitable seed materials. nesium, ammonium and phosphate, concentration of requisite
species along with the range of pH values were utilized as
2.2.1. Preparation of seed materials for batch experiments model input. All PHREEQC calculations were performed at
Preparation of seed (for batch experiment) was conducted 25 ◦ C temperature. Using the default database file, PHREEQC
in batch scale homogeneous struvite crystallization. In this cir- model calculated the concentrations of different complexes in-
cumstance, equimolar initial solution of 0.007 M of magnesium, cluding free Mg2+ , NH+ 4 and PO4 .
3−

ammonium and phosphate was used along with constant oper- To determine the saturation state of feed solution, concentra-
ating pH of 7.75. Generated struvite crystals were then used tion of the feed solution along with the range of pH value was
as seed to perform another experiment in heterogeneous crys- utilized as model input. The default database of PHREEQC
tallization mode (in the presence of seeds) until the desirable does not contain the equilibrium solid phase data of struvite;
growth of struvite. This technique of seed preparation is called hence, it was necessary to code the equilibrium state of stru-
as serial seeding. Produced struvite crystal was then collected vite together with the equilibrium constant. The thermodynamic
and sieved using 45–63 m of ASTM standard sieve. data of struvite are available in the literature (Taylor et al.,
1963; Ohlinger, 1999), and is shown in Eqs. (2)–(4). The de-
tailed outcome of the PHREEQC thermodynamic modeling is
2.2.2. Preparation of seed materials for fed-batch experiments
illustrated in Section 3.2. It is worthwhile pointing out that the
Seed crystals for use in pilot scale experiments were prepared
input data of heat transfer due to the reaction of struvite for-
in fed-batch pilot scale crystallization using serial seeding tech-
mation (H ) is used zero, since no such data are available in
nique (Section 2.2.1), as described in the previous paragraph.
the available literature
Size classified struvite seeds were prepared using 63–150 m
of ASTM standard sieve. MgNH4 PO4 · 6H2 O ⇔ 6H2 O + Mg2+ + NH+ 3−
4 + PO4 ,
(2)
2.2.3. Sieving technique of crystals Log K = −13.27, (3)
Wet sieving was employed to prepare size classified struvite
seeds. Subsequent to the separation of crystals from crystal H = 0. (4)
slurry, further separation of fines was employed using mild The saturation index (SI) identifies the saturation stage of so-
alkaline water (pH = 8.5) to avoid any dissolution of crystal lution, which is defined as
(struvite) during sieving. Size classified crystal, obtained from
wet sieving, was then air dried for 7-days to remove surface SI = Log(IAP) − Log Ks , (5)
moisture. It is recommended to use the drying temperature of where Ks is the temperature corrected solubility constant. IAP
struvite in the range of 40–50 ◦ C. Experimental observations the ion activity product of struvite.
show that over drying (above 50 ◦ C) of struvite produces fragile If SI > 0, supersaturation is indicated. When SI < 0, the so-
crystals. Fragile seeds used in the fed-batch experiment may lution is undersaturated with respect to the solid. When SI = 0,
cause breakage of crystals due to recirculation of parent solution there is an apparent equilibrium with respect to the solid.
during crystallization.
3. Results and discussion for design and commissioning of
2.3. Experiments on suitable seed materials struvite reactor

As part of the design of the struvite reactor, it is necessary 3.1. Identification of metastable zone
to determine suitable seed materials, which have positive influ-
ence on struvite growth. It is worthwhile pointing out that the Thermodynamic equilibria of struvite chemistry were simu-
presence of suitable seeds in reactive solution acts as diffusive lated using gPROMS and the simulated response was verified
3954 Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961

10.0
Labile supersaturation
9.5
9.0
8.5
Metastable Zone
8.0
Solution pH

7.5 Undersaturation

7.0
pH for minimum solubility by coded model
6.5 pH for minimum solubility using PHREEQC
6.0 pH for minimum solubility using Ohlinger's solubility curve
pH for rapid nucleation
5.5
5.0
0.001 0.002 0.003 0.004 0.005 0.006 0.007
Concentration of Magnesium, Ammonium and Phosphate (moles)

Fig. 2. Identification of the metastable zone for struvite crystallization.

with thermodynamic modeling using the PHREEQC thermody-


Reactive Solution
namics modeling package and the derived data of the Ohlinger
(1999) solubility limit curve. Together with concentrations, the (Concentration = x molar)
1
graphical presentation of experimental pH (Section 2.1.2) along
with the simulated pH (Section 2.1.1) are presented in Fig. 2.
The range of maximum and minimum pH limits (Fig. 2) is
known as the operating pH range of struvite crystallization. Titrant-1 Titrant- 2
Crystallization within that range is heterogeneous. Crystalliza-
tion beyond this range is homogeneous, influenced by labile
supersaturation. Fig. 3. Schematic diagram of feed addition.
This operating zone of the crystallization is known as
metastable zone. Crystallization, operating close to saturation,
Table 2
may intensify crystal growth by limiting nucleation. Crystal Possible composition of feed solution
growth in this zone may be governed by surface diffusion
(layering) of newly born crystallites onto seed particles. Feed Type Titrant-1 pH Titrant-2 pH

M1 MgCl2 + NaOH ≈ 10–11 NH4 H2 PO4 ≈ 3.8


M2 NH4 H2 PO4 + NaOH ≈ 9.0 MgCl2 ≈ 7.0
3.2. Composition of feed solution M3 NH4 H2 PO4 + MgCl2 ≈ 5.3 NaOH ≈ 12.5

Experimental control depends on correct stoichiometry of


feed solution based on the fundamental reaction of struvite crys- (Bouropoulos and Koutsoukos, 2000)
tallization. The struvite reactor, operated at constant pH and
constant reactant concentration was employed in this experi- x2 = Cx 1 + 2x1 , (6)
mental control. Assume that the reactor is consisted of V l of x3 = 2Cx 1 − 2x1 , (7)
reactive solution of x1 equimolar concentration of magnesium,
ammonium and phosphate. The schematic diagram of the feed where C is an arbitrary constant and 2x1 is considered for the
addition system is shown in Fig. 3. It is worthwhile pointing dilution effect. Based on the feed addition rate in a trial fed-
out that the feed solution must be split into two streams, since batch experiment, C = 10 was considered to maintain longer
the presence of magnesium, ammonium and phosphate in the duration of experimental control.
same stream may cause the formation of struvite crystal. For- The combination of feed solution (titrant) has significant ef-
mation of struvite crystal in feed stream can adversely affect fect on robust and long-term experimental control of struvite
struvite process, leading to poor control. crystallization. Possible compositions of feed solution using
Fed-batch action of the process was maintained by the ad- MgCl2 , NH4 H2 PO4 and NaOH are listed in Table 2. To evalu-
dition of two titrant streams, which comprised of x2 molar ate the detailed solution chemistry of the various compositions
reactant concentration and x3 molar concentration of NaOH. (Table 2) of feed solution, thermodynamic modeling was con-
Concentration of feed solution was maintained based on ducted using PHREEQC thermodynamic modeling package.
the stoichiometric formulation described in Eqs. (6) and (7) The model input was x2 = 0.06 M (MgCl2 and/or NH4 H2 PO4
Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961 3955

concentration), x3 =0.11 M (NaOH concentration), as described 100

% Free Magnesium ion (Mg )


in Eqs. (6) and (7). It is worthwhile noting that the feed con-

2+
centration of NaOH and MgCl2 and/or NH4 H2 PO4 was cal- 80
Free Free
culated considering the average reactive solution concentration Magnesium ion Magnesium ion
of 0.006 M (x1 ) as employed in the fed-batch experiment (viz. 60 at pH value 5.3 at pH value 11
is = 100% (M3: is = 86.06%
Section 4). Titrant-2 of each feed type (Table 2) consists of Titrant-1) (M : Titrant-1)
single component, whereas titrant-1 of each feed type is multi- 40 1

components phase. Hence, there is high possibility of forming


various complexes in titrant-1 of each feed type, leading to al- 20
teration of desired feed concentration. Therefore, the detailed
solution chemistry of titrant-1 is described in the next para- 0
4 6 8 10 12 14
graphs, since it is accountable for poorer experimental control.
pH value
Eq. (8) shows the formation of the possible chemical compo-
nents when MgCl2 and NaOH solution are mixed to make up Fig. 4. Free Mg2+ concentration in feed-type M1 and M3 of titrant-1.
titrant-1 (feed type 1 of composition types M1 and M3 ). Pos-
sible aqueous phases form in the feed solution are OH− , H+ ,
H2 O, Cl− , H2 , MgOH+ , Mg2+ , Na, NaOH and O2 , whereas 100 Free Ammonium
the possible solid phases are Brucite [Mg(OH)2 ] and Periclase ion at pH value

% Free Ammonium (NH4+ )


Free Ammonium 9.0 is = 72.91%
(MgO) 80 ion at pH value (M2: Titrant-1)
5.4 is = 100%
4MgCl2 + 8NaOH = OH− + H+ + H2 O + 8Cl− + Mg2+ 60
(M3: Titrant-1)

+ MgOH+ + 7Na+ + NaOH + O2


+ Mg(OH)2 + MgO. (8) 40

A thermodynamic modeling using PHREEQC shows that both 20


Brucite [Mg(OH)2 ] and Periclase (MgO) remains in supersatu-
rated state in solution at the specified solution concentration of 0
0.06 M of MgCl2 and 0.11 M of NaOH. It is worthwhile point- 4 6 8 10 12 14
ing out that the saturation indexes (SI) for Brucite are 23.09 pH value
and 37.02 at pH value of 4 and 11, respectively. Whereas, the Fig. 5. Free NH+
4 and NH3 concentration in feed-type M2 and M3 of titrant-1.
saturation indexes (SI) for Periclase are 27.88 and 41.82 at
pH 4 and 11, respectively, for the specified feed solution con-
centration described above. Moreover, magnesium may form
some other complexes including MgOH+ and Mg(OH)2 (ac- Mixing of NH4 H2 PO4 and NaOH solution to make up titrant-
tive), as computed by PHREEQC model (Eq. (8)). MgOH+ 1 (feed type M2 ) led to imbalance control due to transformation
remains in dissolved form and not accountable for magnesium of NH+ +
4 to volatile NH3 . The transformation of NH4 to NH3
loss. Mg(OH)2 (active) is a solid phase, which remains in un- caused significant loss of NH+ 4 in the form of NH3 (probably
dersaturated form (SI = −40 at pH 4; SI = −22.50 at pH 14 at less than 27.09%) from the feed solution (viz. Fig. 5 at pH
the specified feed concentration, described above). Therefore, 9.0), leading to an alteration of the desired supersaturation in
Mg(OH)2 (active) also not accountable for magnesium loss in the reactive solution.
feed solution. Hence, the loss of reactive magnesium is actually Mixing of NH4 H2 PO4 and MgCl2 to make up titrant-1 of
caused by the solid formation (flock) of Brucite [Mg(OH)2 ] feed type M3 could easily form struvite in feed solution, since
and Periclase (MgO). feed solution (titrant-1 of feed type M3 ) remained strongly con-
Thermodynamic prediction, using PHREEQC, demonstrates centrated by the feed magnesium, ammonium and phosphate.
the trend of Mg2+ in titrant-1 when composition type M1 and A trial experiment conducted with 0.006 M equi-molar concen-
M3 is maintained (Fig. 4). As described in Fig. 4, about 14% tration of reactive magnesium, ammonium and phosphate along
reduction (loss due to precipitation+reduction of concentration with the equimolar feed concentration of 0.06 M of (Eqs. (6)
due to complex formation) of free magnesium occurs at pH and (7)) magnesium, ammonium and phosphate caused struvite
11, due to the formation of dissolved MgOH+ , as well as solid precipitation in feed solution at pH value of 5.52. Thermody-
formation of Brucite [Mg(OH)2 ] and Periclase (MgO). namic modeling using PHREEQC indicated supersaturation of
A trial experiment showed that the precipitation of flock in solution in this circumstance. Precipitation of struvite in feed
the feed solution did not take part in feed addition since it had solution caused the alteration of feed concentration very sub-
settled at the bottom in the static condition. At the make up stantially leading to very weak experimental control. In this
pH of titrant-1 of feed type M1 , this precipitant caused loss circumstance, reactive solution achieved undersaturation at the
of soluble magnesium in the form of Mg(OH)2 (and MgO) initial stage of experiment startup. Undersaturation of solu-
precipitate in the feed (less than 14%), leading to imbalance in tion occurred due to the reduced feed concentration. Therefore,
reactant concentration in the system. titrant-1 of feed type M3 caused significant impact on faulty
3956 Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961

Fig. 6. Scanning electron microscopic view of: (A) struvite seeds, (B) growing struvite.

Fig. 7. Scanning electron microscopic view of: (A) quartz sand seeds, (B) growing struvite with quartz sand.

experimental control leading to the change of supersaturated 3.3. Determination of suitable seeds (batch scale experiment)
solution to the undersaturated.
It is therefore imperative to use feed solutions according to Figs. 6 and 7 describe the scanning electron microscopic
M1 to minimize loss of reactants due to the formation of struvite (SEM) view of struvite growth, using struvite seeds and quartz
and NH3 in feed solution. Continuous stirring of titrant 1 (feed sand seeds, respectively. The SEM view in Fig. 6 demonstrates
type M1 ) set the feed solution in motion, thus avoiding the the similarity of typical orthorhombic shape of struvite seeds
settlement of Mg(OH)2 precipitate. This well-mixed Mg(OH)2 and growing struvite due to crystallization. Noticeable increase
in feed solution would then redissolve on entry to the reactor. of size of struvite seeds was observed after crystallization
The ionic strength of the reactive solution was altered main- (Fig. 6A and B). Quartz sand seeds did not take part in growth
taining 0.1 M of NaCl. Considering the dilution effect of the (Fig. 7); however, some struvite crystals formed due to the
combined feed addition, 0.2 M of NaCl was required in titrant-2 nucleation and subsequent growth of stable crystallites. The
of feed type M1 to maintain 0.1 M concentration of NaCl in the magnified view described in Fig. 8 shows the clear image of
reactive solution. The addition of 0.2 M of NaCl reduced the pH struvite growth in the presence of quartz sand seeds. Tiny
value of titrant-2 (feed type M1 ) to 3.60. In this circumstance dendrite-shaped struvite crystals form in the presence of quartz
“acid–base neutralization” occurred in reactive solution when seeds and they remain as separate crystal instead of layering
both titrants were added at the same flow rate to maintain con- onto the surface of quartz sand.
stant stoichiometry. A trial experiment showed that acid–base Particle size analysis using Malvern particle-sizer confirmed
neutralization caused excessive flow rate of feed solution, lead- growth of struvite of about 14 and 4 m of growth for struvite
ing to the occurrence of extreme supersaturation. Adjustment seeds and quartz sand seeds, respectively (Figs. 9 and 10). The
of pH of titrant-2 (feed type M1 ) was thus required to mini- reason is probably the similarity of lattice structure between
mize the acid–base neutralization effect. Experimental investi- struvite seeds and newly born struvite crystallites enhances the
gations showed that the pH difference between titrant-2 (type diffusion integration process (Mullin, 1993). It is worthwhile
M1 ) and reactive solution 1.0–1.25 maintained reasonably con- pointing out that the crystallite is a domain of solid-state mat-
sistent control of the process. ter that has the same structure as single crystal. However, the
Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961 3957

likelihood of diffusion integration process for quartz seeds is


less effective, leading to re-dissolving of unstable crystallites. It
is worthwhile noting that the experiments relating to this section
were conducted very close to saturation region, as investigated
previously by thermodynamic modeling using PHREEQC.

4. Fed-batch-controlled crystallization

Fed-batch struvite crystallization requires some pre-requisite


observation for reasonably accurate control of supersaturation
in fed-batch crystallization.

4.1. Summary of the design and commissioning

Fig. 8. Magnified scanning electron microscopic view of struvite growth when Based on the design and commissioning of fed-batch-
quartz sand is used as seeds. controlled struvite crystallization, following conclusions can
18

16
Struvite Seeds
14
Growing struvite (0.002M)

12 Growing struvite (0.003M)


Volume %

Growing struvite (0.004M)


10
Growing struvite (0.005M)
8

0
0 50 100 150 200
Particle size in microns (1.2 to 600)

Fig. 9. Crystal size distributions (CSD) of struvite seeds and growing struvite.

25

Quartz seeds
20
Growing struvite (0.002M)

Growing struvite (0.003M)

15
Volume %

Growing struvite (0.004M)

Growing struvite (0.005M)

10

0
0 50 100 150 200
Particle size in microns (1.2 to 600)

Fig. 10. Crystal size distributions (CSD) of quartz sand seeds and growing struvite with quartz sand.
3958 Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961

Data Logger
pH Controller

Automatic
temperature Control
System

Feed titrant-1 Feed titrant-2

Dosing Pump Dosing Pump

Recirculation PVC
pipe

Cooling Coil

Recirculation Pump

Fig. 11. Schematic of controlled struvite crystallization.

be made: 4.2. Design of struvite reactor

• Fed-batch experiment is required to operate within previously Unfortunately, no standard design approach exists for stru-
investigated stable metastable zone to minimize spontaneous vite crystallization. However, some common techniques such
precipitation (Section 3.1). as suspension bed, seedings, feed addition are widely practiced
• The fed-batch system is required to maintain using two feed (Bouropoulos and Koutsoukos, 2000; van der Houwen and Val-
solutions. The first is a solution of NH4 H2 PO4 (the ammonia sami-Jones, 2001; Kofina and Koutsoukos, 2003, 2005; Adnan
and phosphate source) and NaCl (to maintain ionic strength), et al., 2004). A schematic diagram of struvite fed-batch system
pH adjusted to 6.0. The second is composed of NaOH (for is shown in Fig. 11.
pH maintenance) and MgCl2 (the magnesium source). Com- A struvite reactor of 44 L volume, made of clear Perspex,
position factors of the feed solutions are 12x1 (Mg2+ , NH+ 4 was used in this study (Fig. 12). The reactor was operated in
and PO3−4 ) and 18x1 (NaOH), where x1 is the equimolar re-
suspended bed condition, allowing seeds and crystals to keep
active solution concentration (Section 3.2). in uniform and full suspension. A pH controller (alpha 2000W)
• Struvite crystal is required to use for struvite crystallization controlled the desired pH of solution and triggered feed ad-
to enhance struvite growth rate (Section 3.3). dition when solution pH dropped below set point. Two dos-
ing pumps (Grundfos DME-12) were operated for titrant (feed
Fed-batch experiment was conducted based on the previous solution) additions based on two separate output signals sent
investigation. The detailed design of fed-batch experiment is via the pH controller. The pH controller was operated in pulse
described in Section 4.1. frequency control mode together with a proportional integral
Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961 3959

Fig. 12. Photographic view of struvite reactor.

control strategy. Based on the pH sensor output, the dosing atures, enabling effective supersaturation control. A plastic-
pump adjusted the flow rate of feed solution. coated cooling coil (Figs. 11 and 12) assembled with control
The pH controller monitored and controlled the desired solu- module was required to avoid corrosion of exposed copper and
tion pH by two automatic relays (relay A and relay B) system, to maintain constant experimental temperature. The fed-batch
mounted in it. This system made the experimental control fully experiment was conducted using the desired reactive concentra-
automatic. Therefore, the feed solution must be split into two tion 0.0060 M along with the operating pH of 7.22. Previously
parts as described in Section 3.2. It is worthwhile pointing out generated fresh struvite seeds of mean diameter 127.77 m of
that the split of feed solution into three parts is not adaptable weight 30 g were used in the fed-batch experiment.
with a single pH controller due to the absence of third relay.
Nonetheless, using more than one pH controller would be one 5. Results and discussion from fed-batch experiment
way to maintain experimental control, which however could
make the control system more complicated and more expensive Followed by the experimental setup, as described in Figs. 11
(due to addition pH controller, dosing pump). and 12, the conducted experiment in controlled supersaturating
The mixing of the reactive solution was carried out by a re- showed the stability of control all through the experiment. Dur-
circulation pump (Onga pump: model 413) together with solu- ing the fed-batch experiment seed crystals took part in growth.
tion recirculation loop, composed of PVC pipe of 1-in diameter. The experimental control of pH and reactant concentrations is
A variable transformer (variac) was used to maintain trouble shown in Section 5.1 and the growth of struvite crystals during
free controlled flow rate of the recirculation pump. The top por- crystallization is shown in Section 5.2.
tion of PVC pipe was connected with an adjustable and flexible
recirculation system (Fig. 12) to avoid short circuiting of recir- 5.1. Control of supersaturation in fed-batch experiment
culated streams. The reactor was seeded with 30 g of previously
generated struvite crystal of size range 63–150 m. Samples Fig. 13 illustrates the control profile of struvite system at dif-
of crystal suspension were collected through pump outlet and ferent process conditions. Fig. 13(A) shows the profile of exper-
were filtered using 0.45 m filter paper. imental control relating to the concentration of magnesium and
Continuous operation of the recirculation pump led to ex- phosphate as well as experimental pH during the experiment.
cessive temperature rise in the reactor. Significant increases of Reasonably acceptable control of reactive solution composi-
solution temperature caused offset (drift) of pH value. There- tion and the operating pH of reactive solution were achieved
fore, it was necessary to maintain constant operating temper- in the fed-batch pilot scale experiment. Fig. 13(B) shows the
3960 Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961

0.010 8
0.009 7
0.008

pH value of Solution
Concentration (M)

6
0.007
0.006 5
0.005 4
0.004 Mg Conc (M)
Average Controlled Mg (M)
3
0.003
PO4 Conc (M) 2
0.002 Average Controlled PO4 (M)
0.001 pH value 1
0.000 0
0 2 4 6 8 10
(A) Time in hours

30 8
Fig. 14. SEM view of produced struvite.
7
25

pH value of solution
6
Volume in Litre

20 25
5
15 4
Volume of reactive 3 20
10 solution (L)
pH value 2 Percent Volume
5
1 15
0 0 Seeds
0 2 4 6 8 10 10.04-hrs
10
(B) Time in hours

Fig. 13. (A) Process control of the system of expected reactive concentration 5
0.0045 M and pH 7.40. (B) Volume of reactive solution during the process
control.
0
0 50 100 150 200 250 300 350 400 450
increase of reactive volume due to the fed-batch action of Particle Size Range (microns)
experiment.
Fig. 15. CSD curve of struvite seeds and growing struvite.
After fed-batch experiment, generated struvite was separated
from mother liquor using ASTM standard sieve of aperture
63 m. Collected struvite was air dried for seven days. Fol- harvested crystal was 161.63 m. Hence, the growth of crystal
lowed by air drying, dry sieving was implemented to separate was 33.86 m. XRD analysis of the produced crystals indicates
fines of mean size smaller than 63 m. A recovery of 65–70% the crystals as struvite (Fig. 16).
of the expected struvite mass was found in this experiment. For
the perfect diffusion integration mechanism, it was expected 6. Conclusions
that 100% of the generated struvite would be larger than 63 m
(considering the lower size range of struvite seed). However, 1. The strategy described here has potential to cope with stru-
experimental observation showed that 30–35% of the generated vite crystallization at constant supersaturation, since pH and
struvite showed smaller size than was expected. Fractional gen- reactant concentration can be kept constant.
eration of smaller struvite was due to the production of fines. 2. The fed-batch system must be maintained using two feed
Some possible reasons of fines generation may be due to the solutions. The first is a solution of NH4 H2 PO4 (the ammonia
imperfect diffusion of newly born crystals, breaking of grow- and phosphate source) and NaCl (to maintain ionic strength),
ing crystals/seeds due to the impeller action of recirculation pH adjusted to about 6.0. The second is composed of NaOH
system, etc. (for pH maintenance) and MgCl2 (the magnesium source).
Composition factor of the feed solutions are 12x1 (Mg2+ ,
5.2. Produced struvite crystal in fed-batch crystallization NH+ 3−
4 and PO4 ) and 18x1 (NaOH), where x1 is the equi-
molar reactive solution concentration.
The controlled fed-batch experiment, described above, pro- 3. It is possible to identify the metastable zone when the mini-
duced struvite crystal of orthorhombic shape (Fig. 14). CSD mum struvite solubility limit (saturation limit) and the min-
curve of seeds and growing struvite crystals, during the op- imum limit of spontaneous precipitation are known. Laser
eration of pilot scale reactor operation, is shown in Fig. 15. light scattering into the reactor can detect the minimum limit
The mean diameter of struvite seeds employed in the fed- of pH for spontaneous precipitation. Minimum struvite sol-
batch experiment was 127.77 m and the mean diameter of ubility limit is identified by thermodynamic simulation and
Md.I. Ali, P.A. Schneider / Chemical Engineering Science 61 (2006) 3951 – 3961 3961

900

800
struvite - File: 8057-01p general.raw
Operations: X Offset -0.058 | Import
700
Struvite, syn - NH4MgPO4·6H2O
Dittmarite, syn - NH4MgPO4·H2O
600
Lin (Counts)

500

400

300

200

100

0
2 10 20 30 40
2-Theta - Scale

Fig. 16. Analysis of struvite by XRD analysis.

was validated precisely by PHREEQC thermodynamic mod- Kofina, A.N., Koutsoukos, P.G., 2005. Spontaneous precipitation of struvite
eling package and the available solubility limit curve, pro- from synthetic wastewater solutions. Crystal Growth and Design 5 (2),
posed by Ohlinger (1999). 489–496.
Martell, A.E., Smith, J.C., 1989. Critical Stability Constants.
4. Uses of mother crystals (struvite as seeds) in struvite crys- McPherson, A., 1988. The use of heterogeneous and epitaxial nucleants to
tallization provide positive influence on struvite growth. promote the growth of protein crystals. Journal of Crystal Growth 90
Mother crystals provide support of efficient diffusion inte- (1–3), 47–50.
gration process, leading to faster growth of struvite during Morel, F.M.M., Hering, J.G., 1993. Principles and Applications of Aquatic
crystallization. Chemistry. Wiley, New York.
Mullin, J.W., 1993. Crystallization. third ed. Butterworth-Heinemann
5. Produced struvite crystals show the typical orthorhombic Publications, Ipswich, UK.
shape with sharp edge. Muramatsu, K., Yasui, A., Suzuki, T., Kiuchi, K., 2000. Factors involved in
struvite formation by Natto bacilli. Biocontrol Science 51 (1), 57–60.
References Nelson, N.R., Mikkelson, R.L., Hesterberg, D.L., 2000. Struvite formation to
remove phosphorus from anaerobic swine lagoon. In: Eighth International
Adnan, A., Dastur, M., Mavinic, S.D., Koch, F.A., 2004. Preliminary Symposium on Animal, Agriculture and Food Processing Wastes. Des
investigation into factors affecting controlled struvite crystallization at Moines, USA, pp. 18–26.
the bench scale. Journal of Environmental Engineering Science 3, Ohlinger, K.N., 1999. Kinetics effects on preferential struvite accumulation in
195–202. wastewater, Ph.D. Thesis, School of Science and Engineering, California
Ali, M.I., Schneider, P.A., 2005. Crystallization of struvite from metastable State University.
region with different types of seed crystal. Journal of Non-Equilibrium Smith, R.M., Martell, A.E., 1976. Critical Stability Constants. Plenum
Thermodynamics 30 (2), 95–113. Publishers, New York.
Bouropoulos, C.Ch., Koutsoukos, P.G., 2000. Spontaneous precipitation of Srinivasakannan, C., Vasanthakumar, R., Iyappan, K., Rao, P.G., 2002. A study
struvite from aqueous solutions. Journal of Crystal Growth 213, 381–388. on crystallization of oxalic acid in batch cooling crystallizer. Chemical
Childs, C.W., 1970. A potentiometric study of equilibria in aqueous divalent and Biochemical Engineering Quarterly 16 (3), 125–129.
metal orthophosphate solutions. Journal of Inorganic Chemistry 9 (11), Taylor, A.W., Frazier, A.W., Gurney, E.L., 1963. Solubility product of
2465–2469. magnesium ammonium. Transaction Faraday Society 59, 1580–1589.
Hirasawa, I., 1996. Study on the recovery of ions in wastewater by Thaller, C., Weaver, L., Eichele, G., Wilson, E., Karlsson, R., Jansonius,
crystallization. Memoirs of the School of Science and Engineering 60, J., 1981. Repeated seeding technique for growing large single crystal
97–119. structures of proteins. Journal of Molecular Biology 147, 465–469.
Kofina, A.N., Koutsoukos, P.G., 2003. Nucleation and Crystal Growth of van der Houwen, J.A.M., Valsami-Jones, E., 2001. The application of calcium
Struvite in Aqueous Media: New Prospective in Phosphorus Recovery. phosphate precipitation chemistry to phosphorus recovery: the influence
WASIC, Istanbul, Turkey. of organic ligands. Environmental Technology 22, 1325–1335.

Você também pode gostar