当前位置:文档之家 > Potential phosphorus recovery by struvite formation

Potential phosphorus recovery by struvite formation


Water Research 36 (2002) 18341842

Potential phosphorus recovery by struvite formation

Y. Jaffera, T.A. Clarkb , P. Pearceb, *, S.A. Parsonsa
a School of Water Sciences, Cranfield University, Cranfield, Bedfordshire, MK43 OAL, UK bSpencer House Laboratory, Thames Water Institute, Manor Farm Road, ReadingRG2 0JN, UK
Received 24 October 2000; received in revised form 20 July 2001; accepted 27 August 2001

Abstract
Formation of struvite (MgNH4PO46H2O) at sewage treatment works can cause operational problems and decrease
efficiency. Struvite has a commercial value and the controlled formation and recovery of it would be beneficial.
A mass balance was conducted at full scale across the whole sewage treatment plant in order to identify a stream to
conduct bench-scale struvite crystallisation studies. The most suitable stream was identified as the centrifuge liquors.
The average flow of the liquor stream was 393 m3d1and the composition was as follows: 167 mg L1phosphorus,
44 mg L1magnesium, 615 mg L1ammonium, 56 mg L1calcium and 2580 mg L1of alkalinity. The pH averaged at
7.6 and the stream had a predicted struvite precipitation potential of 140 mg L1 .
Struvite crystallisation occurred quickly during the trials, by raising the pH of the centrifuge liquors to 9.0 and dosing
with magnesium. Up to 97% phosphorus removal as struvite was achieved. Struvite formation occurred when the molar
ratio of magnesium: phosphorus was at least 1.05 : 1. Below this ratio phosphorus removals of 72% were observed, but
not exclusively as struvite. Annual yields of struvite were calculated to be 42100 tonnes a year, depending on the dose
regime. Revenue from the sale of produced struvite could be betweend8400 andd20,000 a year.r2002 Elsevier
Science Ltd. All rights reserved.

Keywords:Phosphorus recovery; Mass balance; Struvite; Magnesium ammonium phosphate

1. Introduction
Phosphorus pollution in surface waters can lead to
problems with eutrophication in the receiving water. A
number of legislative initiatives exist to overcome this
problem, one of which is the Urban Wastewater
Treatment Directive 91/271 (UWWTD) [1]. The wide-
spread implementation of the UWWTD has lead to an
expansion in secondary and tertiary wastewater treat-
ment in Europe to meet new discharge standards.
With more stringent standards imposed regarding
nutrient removal, processes have been developed to
remove compounds containing nitrogen and phos-
phorus. The result of removing greater concentrations
of nutrients from the wastewater is that the wasted

sludge has a greater concentration of phosphorus,
nitrogen and magnesium. The combination of these
ions found in sludge produced from nutrient
removal, specifically biological nutrient removal
(BNR) processes, can result in the formation of a
mineral called struvite.
Struvite is magnesium ammonium phosphate
(MgNH4PO4) and forms a hard crystalline deposit when
the molar ratio of Mg : NH4:PO 4 is

greater than 1 : 1 : 1.
Struvite is most likely to form in areas of increased
turbulence, as its solubility decreases with pH and its
formation is often associated with anaerobic and post-
digestion processes. Struvite in wastewater treatment
plants was identified as early as 1939. Whilst studying
digestion, Rawn et al. [2] identified struvite in the
digested sludge supernatant lines. Problems with struvite
formation date back to the 1960s when it was noticed at
the Hyperion treatment plant, Los Angeles, where the
digested sludge pipeline diameter had diminished from

*Corresponding author. Tel.: +44-118-923-6232.
E-mail address:pete.pearce@http://www.doczj.com/doc/31951129cfc789eb172dc89c.html
(P. Pearce).
0043-1354/02/$ - see front matterr2002 Elsevier Science Ltd. All rights reserved.
PII: S0043-1354(01)00391-8

12 to 6 in [3]. Similar instances of pipe blockages have
been reported elsewhere [47].
The blockage of pipes leads to an increase in pumping
costs; as the diameter of the pipe is reduced, more energy
is required to move the sludge. Also, the time taken for
the sludge to be moved from one place to another
increases. Most plants which have a struvite problem,
incorporate a time consuming maintenance program
into the normal operation of the plant. For example, at
the Ponggol Pigwaste plant, Singapore, floating aerators
are regularly cleaned resulting in a loss of 810 man
hours each time [5].
The work presented in this paper focuses on Slough
sewage treatment works (STW). Since the commission-
ing of the new BNR plant at Slough, problems with the
formation of struvite have been detected [8]. Eight
months after the new plant was operational, the pipe
between the digesters and the digested sludge holding
tank became blocked with an accumulation of small
struvite crystals in a matrix of digested sludge solids.
One year after operating, the pipeline between digested
sludge holding tank and the centrifuge had become
restricted to such an extent that it was no longer possible
to transfer sludge to the centrifuge. The 100 mm
diameter pipe fittings close to the pump had been
reduced to 50 mm. The problem of struvite formation at
Slough STW still exists and a time consuming main-
tenance program has been incorporated to try and keep
the problem manageable. As yet no long-term solution
has been found to improve the situation.
This paper looks at the formation, control and
recovery of struvite at Slough STW.
2. Materials and methods
2.1. Mass balance of Slough sewage treatment works
Slough STW treats a population equivalent (PE) of
approximately 250,000, of which 114,000PE is industrial
euent. The works consist of two treatment streams,
one consisting of conventional activated sludge and the
other a Bardenpho BNR plant. The works have been
described in detail elsewhere [8,9].
2.1.1. Sampling
Samples were taken from selected sites around Slough
STW. A number of the samples obtained were
composite samples, as the samples were e

xpected to be
highly variable. Composite samples were obtained using
automatic samplers (American Sigma, model 900). All
other samples taken were grab samples.
2.1.2. Sample analysis
All samples were sent to Thames Water Quality
Centre Laboratories, which has NAMAS (National

Measurement Accreditation Service) accreditation. The
pH of each sample was measured on site, using a WTW
pH probe (model SenTix 41-3) and a WTW portable
unit (model pH197-S).
2.1.3. Calculation of the struvite precipitation potential
The struvite precipitation potential (SPP) for each
sampling point was then calculated using a computer
model called Struvite (Version 3.1). This model was
developed for the Water Research Commission, South
Africa [10].
2.1.4. Pilot plant
The bench-scale reactor was constructed using a
Water Research Council (WRC) porous pot apparatus
(Bird & Tole Ltd., UK). A diagram of the bench-scale
rig can be seen in Fig. 1. Two peristaltic pumps (Watson
Marlow, UK) were used to feed centrate liquors and
magnesium chloride into the reactor. The average
influent flow rate was 20 ml min1(of which 63% was
magnesium chloride) producing an HRT of 3 h. Aera-
tion was supplied at 220 ml min1 by an aquarium
aeration unit, which consisted of a pump and two
aeration stones.
2.1.5. Pilot plant feed
The reactor was fed with centrate liquors obtained
from Slough STW. The liquors were collected in 25 L
plastic containers and stored at room temperature. The
pH of the liquors were recorded using a Jenway 100 pH
probe and then adjusted to 9.0, using 20% sodium
hydroxide solution (Hays Chemicals, UK). The reactor
was also fed with magnesium chloride, which was
obtained from NedMag chemicals (Netherlands).
2.1.6. Analysis
The euent from the reactor was sampled on a daily
basis after at least three hydraulic retention times
(HRTs). A sample of euent was sent to Thames Water
Quality Centre Laboratories for analysis of total
kjeldahl nitrogen (TKN), ammoniacal nitrogen, total
and soluble phosphorus, total and soluble magnesium,
calcium, alkalinity and suspended solids. On the spot
analysis was carried out on the euent samples for pH,
ammoniacal nitrogen, soluble phosphorus, total phos-
phorus and total magnesium. Spot analyses were carried
out using Dr. Lange test kits (Dr. Lange, Basingstoke,
UK).
3. Results and discussion
3.1. Mass balance across Slough STW
A mass balance of total phosphorus, TKN and
magnesium through Slough STW is shown in Fig. 2

Y. Jaffer et al. / Water Research 36 (2002) 18341842 1835

and summarised in Table 1. The primary settlement tank
is abbreviated to PST, other abbreviations are explained
in Table 1. When trying to predict the formation of

struvite it is necessary to measure the levels of
phosphorus, magnesium and ammonia through the
works. The areas where the highest level of all three



Porous Pot

Effluent
MgCl2
Aeration

Centrate

Fig. 1. Schematic

of pilot plant.

Effluent A

PST
Clarifier PST Humus
Tank

PFT

Lagoon

Crude



Legend
Tot.P (Kg d-1)
TKN (Kg d-1)
Mg (Kg d-1)
SPP (mgL-1)
Flow (m3 d-1)

708
2108
644
-341
66023

386
1608
415
-352
51202

67
358
333
-440
49985

112
377
144
-276
20423

447
1108
92
-320
1186

375
627
83
-21
165

82
204
23
NA
163

214
52
113
NA
553

139
190
13
-336
222

102
NA
100
-1075
331

29
481
12
--342
1022

603
1474
103
NA
538

108
448
116
-
14268

610
1495
93
198
596

57
547
19
140
393

352
667
56
420
64

470
1152
71
541
456

188
1028
131
NA
1746

520
1080
513
NA
64277

Cake
Effluent C

SAS Tank Belt Thickener Digester Centrifuge

AS
Plant
Nitrifying
Filters

BNR Plant Clarifier

Imports

140
343
22
NA
140

Fig. 2. Mass balance of phosphorus, ammonia and magnesium at Slough STW.

Y. Jaffer et al. / Water Research 36 (2002) 183418421836

components occurs, should correspond to the area
which has the most potential to form struvite.
The mass balance conducted at Slough STW demon-
strated that 26% of the phosphorus entering the works
was due to the phosphorus feedback, i.e. phosphorus in
the return liquors. This compares well with the 20%
detected by Pitman et al. [6], and the 2050% range
given by Munch and Barr [11]. Other authors have
found phosphorus feedback as high as 40% [12]. It
should be noted that Jardin and Popel抯 figure was
determined during a pilot study and the recycle
percentage range given by Munch and Barr was based
on an estimation, which assumed sludge flow to be 1%
of the total flow to treatment. Magnesium in the return
liquors adds 20% to the magnesium load into the works.
The majority of the phosphorus and magnesium is
found in the centrifuge cake.
The potential for struvite formation (SPP) is also
included in the mass balance. The SPP is calculated
by comparing the ionic product of a solution
([Mg2+ ][PO4
3 ][NH4
+ ]) with the solubility product of
struvite (pKsp12:6). If the ionic product exceeds theKsp
then precipitation occurs, if it is less, dissolution occurs.
The SPP is then calculated by applying or removing a
known dosage of struvite to the solution and calculating
the new ionic product. This is repeated until the ionic

product equals the solubility product [10]. A positive
SPP indicates that there is a potential to form struvite
(in mg L1 ), whereas a negative SPP denotes no
potential for struvite to form. The data used to obtain
the SPP results are shown in Table 1 and Fig. 3.
The digested sludge, centrifuge liquor and centrifuge
cake are identified by the model as streams which have
the potential to form struvite. These streams all have
high concentrations of soluble phosphorus. The centri-
fuge cake has the highest SPP, as it has the highest
magnesium concentration of any of

the streams. The
digested sludge also has a high SPP which could be due
to the high concentration of phosphorus, alkalinity and
ammonia. The centrifuge liquor also has a positive SPP,
though at 140 mg L1 , it is not as high as the other two
streams. The magnesium concentration in the centrifuge
liquors is more than 50% less than the magnesium
concentration in the other two SPP positive streams.
The thickened SAS has the highest magnesium
concentration of all the streams, but has a negative
SPP. The pH of the thickened SAS is less than 7.0, so the
potential for this stream to form struvite is low. If the
pH of the thickened SAS stream increased to 7.5, the
SPP would increase from21 to 51 mg L1 . The
blended SAS also has a high soluble phosphorus and
magnesium concentration but a low SPP. The pH of the

Table 1
Summary of mass balance dataa

Stream Location
number
P-PO4
(mg L1 )
NH4
(mg L1 )
Mg
(mg L1 )
Ca
(mg L1 )
Alk
(mg L1 )
pH Temp
mg L1 SPP
mg L1

PFT feed 1 NA 220 33 644 39.5 6.9 18.4 NA
PFT sludge 2 32.2b 123 61 982 63.8 5.9 19.5 336
PFT liquor 3 32.2 129 13 185 478.8 6.3 18.8 1075
Blended SAS 4 65c 90 9 NA 10.3 7 18.3 320
Thickened SAS 5 14.2d 389 482 NA 50.7 6.9 18.3 21
Belt liquor 6 14.2 1.6 11 121 285.8 7.3 17.2 342
Imports 7 NA 441 134 NA 133 5.7 18.4 NA
Digester feed 8 NA 1166 153 NA 98.8 7.3 26.8 NA
Digested sludge 9 154e 1166 153 NA 98.8 7.3 26.8 198
Centrifuge liquor 10 94.9 615 44 56 2580 7.6 24.1 140
Centrifuge cake 11 94.9f 4477 1049 NA 482 7.0 NA 420
Crude sewage 12 5.7 16.1 8.9 118 360 7.9 14 341
Settled sewage 13 5.7 23.9 8 111 356 7.7 14 352
Euent C 14 5.6 0.8 8.1 109 199 8.2 13.4 276
Euent A 15 0.6 1.4 6.7 112 268 7.6 13 440
aPFT: Picket fence thickener; SAS: surplus activated sludge; NA: data not available. bUsed P-PO4 concentration of PFT liquor, should be the same. cP-PO4concentration of this stream could not be measured by laboratories. Data base on one sample filtered and analysed using Dr.
Lange kit LCK 350. dUsed P-PO4 concentration of Belt Thickener Liquor, should be the same. eP-PO4concentration of this stream could not be measured by laboratories. Data base on one sample filtered and analysed using Dr.
Lange kit LCK 350. fUsed P-PO4 concentration of centrifuge liquor, should be the same.

Y. Jaffer et al. / Water Research 36 (2002) 18341842 1837

blended SAS is relatively low at 7.0, but the pH would
have to increase to 8.5 for the SPP to increase from
320 to 16 mg L1 .
3.2. Options for controllingstruvite formation
In 1999, Booker et al. [13] suggested dosing digested
sludge with acid to decrease the pH to below 7.5,
reducing the potential for struvite to form. The pH of
the digested sludge at Slough STW, would have to be
dropped to 6.2 to reduce the struvite precipitation
potential to a negative, in accordance with the model
devised by Loewenthal et al. [10].
Another possible solution to the formation of struvite
at Slough STW would

be to precipitate the phosphorus
from the centrifuge liquor as struvite, in a side-stream
process. This would lower the amounts of phosphorus
recycled around the works and would reduce the
amount of struvite formed upstream of the process.
This is in accordance with a number of authors who
have used digested sludge supernatant as feed, during
pilot studies into struvite crystallisation [14,15]. This
option was investigated further in a series of bench-scale
experiments.
3.3. Bench-scale struvite precipitation usingcentrifuge
liquors
Centrifuge liquor was collected from Slough and
stored at room temperature. The pH of the centrifuge
liquor was raised to 9.0, using sodium hydroxide. A pH
of 9.0 was found to be optimum for struvite precipita-
tion by Siegrist et al. [16]. Centrifuge liquor samples
were analysed before and after pH adjustment.
Levels of soluble phosphorus, magnesium and cal-
cium dropped after raising the pH to 9.0. Raising the pH
must have caused the magnesium and calcium in the
sample to react with phosphorus, forming precipitates.

The ratio of calcium : magnesium in the centrifuge liquor
was roughly 2 : 1. This decreased to a 1 : 1 ratio after the
pH had been raised. This reduction in the ratio indicates
that more calcium is reacting with the phosphorus in the
sample than magnesium.
The magnesium levels in the centrifuge liquor samples
were too low to remove the remaining phosphorus as
struvite. The pilot plant was therefore dosed with
magnesium chloride (MgCl2) to provide a source of
magnesium ions and maximise struvite production.
3.4. Struvite crystallisation
After 24 h of operation, using centrifuge liquors and a
dose of 252 mg L1of MgCl2, crystals were seen on the
surface of the porous pot. The crystals produced
underwent X-ray diffraction (XRD) and were confirmed
to be struvite.
3.5. Optimum magnesium dose
Once it was established that struvite could be formed
from the centrifuge liquors at Slough STW, the
magnesium dose to the pilot plant was altered to
determine an optimum dosing regime. The removal of
soluble phosphorus and ammonia increased with in-
creased magnesium dose, up to 97% at a dose of
3.46 mM, Fig. 4. The relationship between magnesium
dose and phosphorus removal is shown in Fig. 5 and
indicates that at high dosages of magnesium, the
phosphorus is probably removed as struvite. If struvite
production is occurring, phosphorus removal and
magnesium usage will be similar, this trend can be seen
at higher doses of magnesium. At magnesium doses
below 3.4 mM L1 , phosphorus is still removed, but not
solely as struvite. The molar removal of ammonia
exceeds the molar removals of phosphorus and the
molar usage of magnesium and was greater than the

0
100
200
300
400
500
600
700
800
900
2 3 4 5 6 9 10 12 13 14 15
Location number

Concentration (mg/l)

0
1
2
3
4
5
6
7
8
9

pH

P-PO4
N-NH4
Mg
pH

Fig. 3. Concentration of m

agnesium, soluble phosphorus and ammonia through Slough STW.

Y. Jaffer et al. / Water Research 36 (2002) 183418421838

requirement for struvite. The surplus ammonium was
probably being removed from the reactor by air
stripping.
It was found that 95% of the total phosphorus could
be removed from the centrifuge supernatant as struvite,
by the addition of at least a 1.05 : 1 molar ratio of
magnesium to phosphorus, i.e. a magnesium dose of
about 83 mg L1 . The ratio of 1.05 : 1 was also found by
Fujimoto et al. [17] whereas Siegrist et al. [16] found that
a higher ratio of 1.3 : 1 was required to guarantee
phosphorus removal as struvite. The lower ratio
necessary for struvite precipitation during this series of
experiments was probably due to the lack of any
competing reactions. When the pH of the centrifuge
liquor was raised to 9.0 in the storage containers, 77%
of the calcium was removed, before entering the reactor.
On average 6 mg L1of calcium entered the reactor. At
a dosing regime of 83 mg L1 , this equates to a 0.04 : 1
molar ratio of calcium : magnesium. Hwang and Choi

[18] found that for effective struvite formation, the ratio
of calcium to magnesium should be less than 1. Mustovo
et al. [19] found that the ratio of magnesium : calcium
should be greater than 0.6. In this case, the ratio of
magnesium : calcium is 23 : 1. Before the addition of
sodium hydroxide to raise the pH, the molar ratio of
calcium : magnesium was 1.2 : 1. A soluble phosphorus
removal of 7% was observed on raising the pH, before
the centrifuge liquor was fed into the reactor. Most of
this phosphorus was probably removed as calcium
phosphates.
At full scale, the presence of calcium should be taken
into consideration when determining an ideal magne-
sium dose. The centrifuge liquors were batch pH
adjusted prior to being fed into the reactor. At full scale
it is more likely that pH correction will occur within the
reactor, so the calcium ions will not have been removed
and will compete with magnesium ions. A ratio of 1.3 : 1
(magnesium : phosphorus) is more suitable at full scale.

0
10
20
30
40
50
60
70
80
90
100
0.00 1.65 2.58 3.46 5.67 10.52
Magnesium Addition (mM/l)

% Removal

Total-P
P-PO4
N-NH4

Fig. 4. Percentage removal of phosphorus and ammonia with increasing magnesium dose.

0

0.5

1

1.5

2

2.5

3
0 1.65 2.58 3.46 5.67 10.52
Magnesium Addition (mM/l)

Removal (mM/L)

P-PO4
Mg

Fig. 5. Molar removal of phosphorus and molar usage of magnesium against molar addition of magnesium.

Y. Jaffer et al. / Water Research 36 (2002) 18341842 1839

Also, by batch treating the centrifuge liquors it was
possible to settle out the suspended solids in the liquor,
prior to the pilot plant. If the suspended solids were
allowed to enter the pilot plant they may have caused
clogging of tubing and increased the organic content of
the struvite produced. It may be beneficia

l to add a unit
process prior to the reactor at full scale to remove
suspended solids.
3.6. Costs
The cost of a pilot plant, with a capacity to treat all of
the average centrifuge liquor flow, i.e. 400 m3d1was
estimated. The construction costs are as follows
(Thames Water Engineering, personal communication):
Precipitation reactor d17,000 Civil
Blowers and diffusers d15,000 Civil
d18,000 Mechanical and
electrical
Desludge pump d5000 Civil
d25,000 Mechanical and
electrical
The cost of electricity was estimated atd1400 per year.
The annual chemical cost of struvite precipitation is
shown in Fig. 6. The calculations assume chemical costs
of Mg (as MgCl2) of d90 tonne1 and NaOH of
d59 tonne1(prices quoted for year 2000).
As the percentage phosphorus removal increases, the
cost of chemical addition also increases. Between a
magnesium dose of 2.610.5 mM L1 (62253 mg L1 )
phosphorus removal is about 97%. There is no benefit to
be gained by increasing the magnesium dose above
62 mg L1 . However, to ensure exclusive struvite pro-
duction, the magnesium dose needs to be around
83 mg L1 (3.5 mM L1 Mg). The increase in dose,
increases the cost fromd51,500 tod52,000 per annum.

The cost, in terms of increasing magnesium dose is
small. The majority of the chemical costs can be
attributed to the addition of sodium hydroxide to raise
the pH. The cost of sodium hydroxide was based on the
amount required to raise the pH of the centrifuge liquors
from 7.5 to 9.0, during the bench-scale work. Assuming
the cost of sodium hydroxide isd59 tonne1 , the daily
cost of sodium hydroxide addition would bed139 or
d50,735 per year.
Calculations show that 97% of the total chemical cost
was due to the addition of sodium hydroxide. The
alkalinity in wastewater is generally high, so there is a
large buffering capacity which needs to be overcome. If a
full-scale plant is to be viable, the cost required for pH
correction needs to be addressed.
Some authors have used magnesium hydroxide as a
source of magnesium ions and to raise the pH. Salutsky
et al. [20] achieved 90% phosphorus recovery with
magnesium hydroxide addition, but at a temperature of
251C. Munch and Barr [11] also used magnesium
hydroxide as a dual function chemical and obtained an
average of 94% phosphorus removal as struvite.
However, using magnesium hydroxide to serve both
functions means that the magnesium dose or the pH
cannot be optimised independent of each other. The
Phosnix process uses magnesium hydroxide, but also has
sodium hydroxide addition to control the pH. The
sodium hydroxide requirement is less with magnesium
hydroxide, than with magnesium chloride [21]. An
advantage of using magnesium chloride over magnesium
hydroxide is that magnesium chloride disassociates
faster than magnesium hydroxide, resulting in shorter
reaction times. A shorter reaction time means that a
smaller full-scale reactor can be constructed as the
hydraulic retention t

ime can be reduced.
Some authors have managed to crystallise phosphorus
from wastewater without the addition of any chemicals.
Battistoni et al. [22] removed 80% of phosphorus from

0
10
20
30
40
50
60
70
80
90
100
0.08 1.65 2.58 3.46 5.67 10.52
Magnesium Addition (mM/L)

% Removal

50

51

52

53

54

55

Total-P
P-PO4
Cost (1000/Y)

Fig. 6. Estimated annual chemical cost of phosphorus removal by struvite precipitation.

Y. Jaffer et al. / Water Research 36 (2002) 183418421840

belt press liquors, from a treatment plant that had
nitrification, denitrification and anaerobic digestion.
The liquors were aged and air stripped to remove
carbon dioxide. The pH was raised from 7.9 to 8.38.6
and the phosphorus was removed as struvite. The liquor
had an Mg : P ratio of 3.7 : 1, so did not require
magnesium addition. However, when the work was
repeated with centrate liquor from a biological nutrient
removal plant, phosphorus removal was achieved as a
mixture of struvite and hydroxyapatite. The Mg : P ratio
had decreased to 0.22 : 1 and was no longer sufficient to
exclusively form struvite [14].
The amount of struvite that could be produced at full
scale at each of the dosing regimes used during the
bench-scale work is summarised in Table 2. The
percentage removals of phosphorus are also included.
The amount of struvite produced is based on the amount
of magnesium used in the reactor. It is assumed that all
the magnesium used in the reactor is due to the
formation of struvite. Daily struvite production rates
were based on average centrifuge liquor flows.
If we assume that the optimum dose to remove over
90% of the phosphorus in the centrifuge liquor (as
struvite) is 83 mg L1 , the annual chemical cost of a full-
scale plant would be aboutd52,000. The amount of
struvite that was formed at bench-scale during this
dosing regime was 2.4 mM L1 . This would equate to
231 Kg d1of struvite at full scale. On a yearly basis,
84 tonne of struvite could be recovered. The sale of
struvite could bring in an yearly revenue of approxi-
matelyd17,000 (struvite from the Phosnix process is sold
for approximatelyd200 per tonne).
The revenue that could be generated from the sale of
struvite is one third of the cost of chemical addition.
However, the sale of struvite is not the only factor under
consideration. Costs of production have to be offset
against the revenue lost through increased pumping
costs, lost man hours, expensive pipe replacements,
possible excavation work if pipes are located under-
ground and STW downtime due to blockages. These
factors are difficult to quantify financially.

4. Conclusions

* The mass balance has shown that the majority of
phosphorus is bound in the centrifuge cake and that
26% of the phosphorus load to treatment is due to
phosphorus feedback.
* The centrifuge liquor was identified as the most
suitable stream for struvite recovery.
* At benc

h-scale, struvite can be formed rapidly
by raising the pH to 9.0 and magnesium addition
of at least 1.05 : 1 magnesium : phosphorus. Magne-
sium addition lower than the 1.05 : 1 ratio will
result in a mixture of struvite and hydroxyapatite.
At full scale a magnesium dose of 1.3 : 1 is
recommended.
* 97% phosphorus removal as struvite can be achieved.
* Chemical costs of a full-scale struvite crystallisation
plant will be at leastd50,000 per annum. The
addition of sodium hydroxide to raise the pH,
accounts for 97% of this cost. A revenue of
d16,00020,000 can be generated from selling the
struvite produced.
Acknowledgements
The authors would like to thank Thames Water
Utilities Ltd. and Thames Water Provinces Wastewater
Treatment staff for their help and cooperation during
the surveys, especially the staff at Slough STW. Also, a
special thanks to James Doyle, at Cranfield University
for conducting the XRD experiments.
References

[1] Council of the European Communities. Directive concern-
ing the collection, treatment and discharge of urban
wastewater and the discharge of wastewater from certain
industrial sectors (91/271/EEC). Off J Eur Communities
Ser L 1991;135/40.

Table 2
Potential production of struvite at full scale
Mg dose
(mg L1 )
Bench-scale
production of
struvite (mM L1 )

Full-scale
production of
struvite (kg d1 )

Full-scale
production of
struvite (ton yr1 )

Possible income
from struvite
(dyr1 )a

TotalP
removal (%)
0 0 0 0 0 19
36 1.2 116 42 8400 88
62 1.8 173 63 12,600 95
83 2.4 231 84 16,800 95
136 2.6 250 91 18,200 97
252 2.8 272 99 19,800 96
aBased on the price ofd200 per tonne.

Y. Jaffer et al. / Water Research 36 (2002) 18341842 1841

[2] Rawn AM, Banta AP, Pomeroy R. Multiple stage sewage
digestion. Am Soc Civil Eng Trans 1939;105:93132.
[3] Borgerding J. Phosphate deposits in digestion systems.
J Water Pollut Control Fed 1972;44:8139.
[4] Mamais D, Pitt PA, Cheng YW, Loiacono J, Jenkins D.
Determination of ferric chloride dose to control struvite
precipitation in anaerobic sludge digesters. Water Environ
Res 1994;66:9128.
[5] Mohajit KK, Bhaattarai E, Taiganides P, Yap BC.
Struvite deposits in pipes and aerators. Biol Wastes
1989;30:13347.
[6] Pitman AR, Deacon SL, Alexander WV. The thickening
and treatment of sewage sludges to minimise phosphorus
release. Water Res 1991;25:128594.
[7] Webb KM, Ho GE. Struvite (MgNH4PO46H2O) solubi-
lity and its application to a piggery euent problem. Water
Sci Technol 1992;26(911):222932.
[8] Williams S. Struvite precipitation in the sludge stream at
Slough wastewater treatment plant and opportunities for
phosphorus recovery. Environ Technol 1999;20:7437.
[9] Williams SC, Beresford J. The effect of anaerobic mixing
on the performance of a three stage Bardenpho plant.
Water Sci Technol 1998;38(1):5562.
[10] Loewenthal RE, Kornmuller URC, van Heerden EP.
Modelling struvite precipitation in anaerobic

treatment
systems. Water Sci Technol 1994;30:10716.
[11] Munch EV, Barr K. Controlled struvite crystallisation for
removing phosphorus from anaerobic digester sidestreams.
Water Res 2000;34:186880.
[12] Jardin N, Popel J. Phosphate release of sludges from
enhanced biological P-removal during digestion. Water Sci
Technol 1994;30:28192.
[13] Booker NA, Priestley AJ, Fraser IH. Struvite formation in
wastewater treatment plants: opportunities for nutrient
recovery. Environ Technol 1999;20:77782.

[14] Battistoni P, Pavan P, Cecchi F, Mata-Alverez J.
Phosphate removal in anaerobic supernatants: modelling
and performance of a fluidized bed reactor. Water Sci
Technol 1998;38:27583.
[15] Momberg GA, Oellermann RA. The removal of phosphate
by hydroxyapatite and struvite crystallisation in South
Africa. Water Sci Technol 1992;26:98796.
[16] Siegrist H, Gajcy D, Sulzer S, Roeleveld P, Oschwald R,
Frischknecht H, Pfund D, Morgeli B, Hugerbuhler E.
Nitrogen elimination from digester supernatant with
magnesium朼mmonium杙hosphate. Proceedings of the
Fifth Gothenburg Symposium, September 2830. Berlin:
Springer, 1992.
[17] Fujimoto N, Mizuochi T, Togami Y. Phosphorus
fixation in the sludge treatment system of a biological
phosphorus removal process. Water Sci Technol
1991;23:63540.
[18] Hwang HJ, Choi E. Nutrient control with other sludges in
anaerobic digestion of BPR sludge. Water Sci Technol
1998;38:295302.
[19] Mustovo EV, Wentzel MC, Ekama GA. Integrated
chemical杙hysical processes modellingFII. Simulating
aeration treatment of anaerobic digester supernatants.
Water Res 2000;34:186880.
[20] Salutsky ML, Dunseth MG, Ries KM, Shapiro JJ.
Ultimate disposal of phosphate from wastewater by
recovery as fertiliser. Euent Water Treat J 1972;22:
50919.
[21] Taruya T, Ueno Y, Fujii M. Development of phosphorus
resource recycling process from sewage. IWA World
Congress, Paris, Poster Session, 2000.
[22] Battistoni P, Fava G, Pavan P, Musacco A, Cecchi F.
Phosphate removal in anaerobic liquors by struvite
crystallisation without addition of chemicals: preliminary
results. Water Res 1997;31:29259.

Y. Jaffer et al. / Water Research 36 (2002) 183418421842

下载Word文档免费下载:

Potential phosphorus recovery by struvite formation下载

(共11页)