Development and examination of a granular nitrogen-fixing wastewater treatment system Steven Pratt a, * , Michael Tan a , Daniel Gapes b , Andy Shilton a a Centre for Environmental Technology and Engineering, Massey University, Palmerston North, New Zealand b Scion, Rotorua, New Zealand Received 16 November 2006; received in revised form 12 February 2007; accepted 25 February 2007 Abstract This work presents the first success at aerobic granulation in a nitrogen deficient system. Two sequencing batch reactors (SBRs) were used to treat nitrogen deficient (the N-fix system) or nitrogen-sufficient (containing NH 4 Cl) synthetic wastewater (acetic acid as the sole carbon source). Granulation was observed in both systems, with particularly large granules (average diameter: 7 mm) grown in the N-fix system. We propose that the unique morphology of nitrogen-fixing granules is a consequence of the response of oxygen-sensitive diazotrophs to elevated oxygen concentrations. Both the nitrogen-fixing and nitrogen-supplemented systems were shown to be capable of removing all of the influent substrate carbon. Excellent biomass settleability characteristics were obtained, with the N-fix system having a final sludge volume index (SVI) of less than 100 mL g 1 and its granules having settling velocities of over 1.4 cm s 1 . However, moderately high solids discharges were recorded for both systems, which revealed a potential limitation of granular sludge processes that is not widely discussed in the literature. # 2007 Elsevier Ltd. All rights reserved. Keywords: Nitrogen fixation; Nitrogen deficient; Aerobic granules; Sludge settleability; Sequencing batch reactor 1. Introduction All wastewater treatment processes reliant on biological activity require the supply of adequate nitrogen for main- tenance of microbiological population growth. Often, the ratio of nitrogen to carbon in influent waste streams is more than sufficient to support adequate microbial growth. However, some industrial waste streams, such as discharges from pulp and paper mills, have an extremely low ratio of nitrogen to carbon. For biological treatment of these nitrogen deficient waste streams, novel processes that utilise nitrogen-fixing bacteria may be employed [1,2]. The diazotrophic organisms in these systems are able to directly fix nitrogen from the atmosphere, thus satisfying their cellular nitrogen require- ments, while maintaining extremely low nitrogen discharges in the final effluent [3]. Biological nitrogen fixation involves the reduction of atmospheric dinitrogen to ammonia, catalysed by the nitrogenase enzyme. Nitrogenase is an important cellular component of diazotrophs, comprising up to 20% of their total protein [4]. The component proteins of nitrogenase are extremely oxygen sensitive and thus aerobic bacteria have been found to possess varying mechanisms to survive oxygen inhibition via reducing the ambient oxygen environment with such means as respiratory protection and slime formation, or engaging in enzyme conformational protection [5–7]. Such mechanisms allow two apparently paradoxical processes to occur together within the same cell, namely aerobic respiratory metabolism, coupled with anaerobic nitrogenase activity. Separation of biomass from treated wastewater via gravity settling is a key requirement for full-scale implementation of activated sludge-type wastewater treatment systems. The suspended floc processes that are employed in activated sludge systems are known to be susceptible to bulking problems under conditions of nitrogen deficiency, attributable to proliferation of filamentous organisms and overproduction of extracellular polymeric substances [8]. While nitrogen-fixing (defined here as N-fix) activated sludge systems have been demonstrated to be highly effective in reducing carbon concentrations, biomass settleability has been identified as a potential problem [3]. www.elsevier.com/locate/procbio Process Biochemistry 42 (2007) 863–872 * Corresponding author. Tel.: +64 6 350 5085; fax: +64 6 350 3604. E-mail address: [email protected](S. Pratt). 1359-5113/$ – see front matter # 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.procbio.2007.02.009
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
www.elsevier.com/locate/procbio
Process Biochemistry 42 (2007) 863–872
Development and examination of a granular nitrogen-fixing
wastewater treatment system
Steven Pratt a,*, Michael Tan a, Daniel Gapes b, Andy Shilton a
a Centre for Environmental Technology and Engineering, Massey University, Palmerston North, New Zealandb Scion, Rotorua, New Zealand
Received 16 November 2006; received in revised form 12 February 2007; accepted 25 February 2007
Abstract
This work presents the first success at aerobic granulation in a nitrogen deficient system. Two sequencing batch reactors (SBRs) were used to
treat nitrogen deficient (the N-fix system) or nitrogen-sufficient (containing NH4Cl) synthetic wastewater (acetic acid as the sole carbon source).
Granulation was observed in both systems, with particularly large granules (average diameter: 7 mm) grown in the N-fix system. We propose that
the unique morphology of nitrogen-fixing granules is a consequence of the response of oxygen-sensitive diazotrophs to elevated oxygen
concentrations.
Both the nitrogen-fixing and nitrogen-supplemented systems were shown to be capable of removing all of the influent substrate carbon.
Excellent biomass settleability characteristics were obtained, with the N-fix system having a final sludge volume index (SVI) of less than
100 mL g�1 and its granules having settling velocities of over 1.4 cm s�1. However, moderately high solids discharges were recorded for both
systems, which revealed a potential limitation of granular sludge processes that is not widely discussed in the literature.
diffusion coefficient (DAc) of 4.4 � 10�10 m2 s�1 and half-saturation constant
(KSAc) of 26 g m�3 were selected from data presented in Wu and Hickey [23].
The maximum oxygen consumption rate ðqO2 ;maxÞ was calculated from the
stoichiometry of carbon oxidation. The effective diffusion coefficient ðDO2Þ of
1.58 � 10�9 m2 s�1, half-saturation constant ðKSO2Þ of 1.9 g m�3 and biomass
concentration (X) of 4750 gVSS m�3 were selected from data presented in Su
and Yu [24].
For nitrogen-supplemented carbon conversion, the growth yield (Y) can be
assumed to be 0.47 gVSS gO2�1 [25]. However, for diazotrophs the growth yield
has been shown to vary as a function of dissolved oxygen (DO). At elevated DO
diazotrophs exhibit extremely high maintenance rates, which result in signifi-
cant yield reductions [26,27]. For example, the biomass yield of Azotobacter
vinelandii, a representative diazotroph, is severely reduced at elevated oxygen
concentration [27]. In this work, a relationship for the yield of diazotrophs
(Eq. (2)) was developed from data presented in Kuhla and Oelze [27].
nitrogen fixing system : Y
�gVSS
gO2
�¼ 0:47� 0:47� SO2
SO2þ y
(2)
where y is an empirical constant: 0.7 gO2 m�3 (from data presented in Nagai
and Aiba [28]).
The rate of biomass growth (f [gVSS s�1]) through the granules was
calculated as a function of biomass yield and substrate utilisation at various
distances from the centre of the granule (r):
fr ¼ Y � qO2max
SO2
KSO2þ SO2
SAc
KSAcþ SAc
X (3)
For the purpose of comparing the biomass activity between the nitrogen-
supplemented and N-fix systems, the effective growth rates (hg) of granules of
various volume (VS [m3]) were determined:
hg ¼PR
r¼0 rgrð1=VSXÞmmax
(4)
MATLAB (Mathworks Inc.) was used to solve the model presented in (1)
along with the algebraic relationships presented in (2)–(4). The modelling
experiments were designed to represent the system shortly after carbon addi-
tion: bulk acetic acid concentration was assumed to be 500 gAc m�3 and bulk
oxygen concentration was assumed to be 6 gO2 m�3.
3. Results
3.1. Confirmation of nitrogen fixation
As expected, only the N-fix system yielded a positive
response to the acetylene reduction assay (data not shown),
confirming the presence of nitrogen fixation within this reactor.
A nitrogen balance, which consistently showed elevated total
nitrogen in the effluent from the system with the nitrogen
deficient feed, was further evidence of nitrogen fixation in the
N-fix system.
Fig. 1. Digital camera images of (A) the reactor set-up and (B) samples from the N-fix and nitrogen-supplemented reactors.
S. Pratt et al. / Process Biochemistry 42 (2007) 863–872866
3.2. General observations
The physical characteristics of the granules in the N-fix
system were markedly different to those in the nitrogen-
supplemented system (Fig. 1), with the former producing very
large granular structures, up to approximately 10 mm in
diameter. Clearly, the severe nitrogen deficiency in the feed,
and subsequent modification of the bacterial microbiota to
Fig. 2. Population diversity in reactors based on terminal restriction fragments, norm
total (A) N-fix system and (B) nitrogen-supplemented system.
allow metabolism of atmospheric nitrogen, resulted in a
different granule morphology.
The differences in measured microbial community between
the two reactors are highlighted by the T-RFLP profiles (Fig. 2).
The T-RFLP profile for the nitrogen-supplemented system
indicates the dominance of one microbial group in that system,
while the more varied array of terminal fragments in the
measured microbial population associated with the nitrogen
alised to the sum of fragments present at abundances greater than 0.5% of the
Fig. 3. Organic carbon removal (^ = N-fix system and � = nitrogen-supple-
mented system).Fig. 5. Nitrogen in effluent (^ = N-fix system and� = nitrogen-supplemented
system) (solid line shows TKN; dotted line shows dissolved KN).
S. Pratt et al. / Process Biochemistry 42 (2007) 863–872 867
deficient system implies a greater diversity in microbial
population in that system. The T-RFLP profiles also show that
the communities in the two systems did not change markedly
with time, while the observation of equal-length fragments in
the two reactors does indicate the presence of some similar
bacterial populations capable of proliferating under the two
nutrient environments.
3.3. Reactor performance
As shown in Fig. 3 both reactors demonstrated excellent
treatment performance, reducing the incoming organic carbon
from 500 mg L�1 to less than 25 mg L�1.
The solids concentrations within both reactors showed some
significant fluctuations (Fig. 4A), but despite the short-settling
times, even in the early development stages critical loss of
biomass was not observed. The extent of fluctuation was most
pronounced in the N-fix system, in which the biomass
concentration dropped during the early development stages
(days 7–20) and then significantly increased during the later
development stages (post day 20). Although the amount of
solids within the nitrogen-supplemented reactor was at some
stages higher than that within the N-fix system, at the end of the
study there was approximately twice as much biomass within
the N-fix system.
The concentration of suspended solids in the effluent is
shown in Fig. 4B. It can be seen that this discharge from both
systems was relatively stable throughout the study. The solids
concentration in the effluent from the N-fix system was
approximately half of that observed in the effluent of the
nitrogen-supplemented reactor. This reduced washout from the
Fig. 4. Solids concentration in (A) the mixed liquor and (B) the efflu
N-fix system contributed to this system’s relatively higher
solids concentration.
Fig. 5A shows that during the later development stages (post
day 20) the dissolved nitrogen content in the effluent of the N-
fix system (average 5 mg L�1) was consistently lower than that
in the nitrogen-supplemented system (average 11 mg L�1).
During these stages the TKN of the effluent of both systems was
high (averaging 27 mg L�1 for the N-fix system and 45 mg L�1
for the nitrogen-supplemented system), directly attributed to
the presence of biomass (as shown in Fig. 4B). The TKN in the
effluent of the nitrogen-supplemented system was only
marginally different to the feed nitrogen concentration
(56 mg L�1). However, the TKN of the effluent from the N-
fix system was considerably higher that the feed concentration
(1 mg L�1), providing further confirmation of nitrogen fixation
in the N-fix system. Negligible oxidized nitrogen (NOx) was
observed in either system with the maximum observed NOx
being 0.04 mgN L�1.
3.4. Granule morphology
Stereomicroscope images (Fig. 6) of the individual entities
show that the granules in the N-fix system were spherical with
small fibrous surface features. The granules in the nitrogen-
supplemented system appeared less regular in shape, and were
devoid of any obvious surface features.
The granule characteristics are summarized in Table 3, along
with data from a review of biogranulation [13] and from a study
of granulation for the treatment of industrial wastes [29]. The
image analysis confirms the significant difference in granule
ent (^ = N-fix system and � = nitrogen-supplemented system).
Fig. 6. Stereomicroscope images of individual granules in (A) N-fix system and (B) nitrogen-supplemented system (day 75).
S. Pratt et al. / Process Biochemistry 42 (2007) 863–872868
size between the two systems. The average equivalent diameter
of the granules in the nitrogen-supplemented system was 2 mm,
similar to the dimensions typically reported in the literature.
However, the average equivalent diameter of the granules in the
N-fix system (7 mm) was significantly higher than those
reported in the literature.
Interestingly, the results of the image analysis suggest that
the granules in the nitrogen-supplemented reactor are actually
more circular than those found in the nitrogen-fixing system.
This contradicts the visual observations made based on the
stereomicroscope images. The discrepancy was possibly
caused by the fibrous surface features of the granules in the
N-fix system, which, when processed, resulted in a reduced
area-to-perimeter ratio and consequently a reduced circularity.
The particle size distributions were also determined, using
both a ‘number’ and ‘equivalent volume’ basis. Fig. 7 shows
that, on a number basis, the majority of particles in both systems
were small (88%<1 mm diameter in the N-fix system and 70%
<1 mm diameter in the nitrogen-supplemented system), whilst
most of the material (volumetric basis) was associated with the
large distinct granules (over 90% of the volume of material in
both systems was associated with granules with >1 mm
diameter). The major difference between the two systems was
that the size distribution of the N-fix system was wider than that
of the nitrogen-supplemented system.
Table 3
Granular sludge characteristics (standard deviation in brackets)
Day N-fix Nitrogen-sup
Individual granules
61
Average diameter (mm) 7 (1.1) 2 (0.3)
Area (mm2) 39 (12) 3 (1)
Circularity 0.21 (0.8) 0.61 (0.12
Aspect ratio – –
75
Settling velocity (cm s�1) 1.4 (0.2) 0.9 (0.2)
Specific gravity 1.002 (1) 1.019 (10
Sludge
75
SVI (mL g�1) 55 (15) 106 (43)
a Review paper.
3.5. Settleability
A feature of granular systems is the efficiency with which
sludge can be separated from the treated liquid effluent. This is
an important aspect of sludge quality, and can be quantified in
terms of the sludge volume index (SVI), e.g. [30]. With a SVI of
less than 100 mL g�1 being considered desirable [31], Fig. 8
shows that this aspect of sludge quality of both the systems was
good. Importantly, during the later stages of the study (post day
20) the average SVI (55 mL g�1) of the N-fix system was
significantly lower than that of flocculent N-fix systems, which
have been reported as displaying SVIs of greater than
200 mL g�1 [3].
As well as a low SVI, the nitrogen-fixing granular sludge had
granules that exhibited very high settling velocities (average
1.4 cm s�1); higher than those from the nitrogen-supplemented
reactor (average 0.88 cm s�1, a similar level to velocities
reported by Su and Yu [29]) and much higher than that of
activated sludge flocs (0.17–0.42 cm s�1, Li and Yuan [33]).
The specific gravity of the N-fix system was found to be
lower than that of the nitrogen-supplemented system, with both
values being within the range observed from other work
(Table 3). From Stokes law, both the density and diameter
impact on the settling velocity, and clearly the elevated size of
the N-fix granules compensated for the lowered density levels.
plemented Liu and Tay (2004)a Su and Yu (2005)
0.6–1.3 0.9–1.1
– –
) – �0.55–0.63
0.73–0.79 0.74
Up to 0.8–1.9 1.02 � 0.24
) 1.004–1.065 1.017
30.8 � 5.3
Fig. 7. Cumulative particle size distribution for (A) the N-fix and (B) the nitrogen-supplemented systems (� = number of entities and & = equivalent volume of
entities).
S. Pratt et al. / Process Biochemistry 42 (2007) 863–872 869
3.6. Substrate penetration and biomass growth rates in
granules
The simulations shown in Fig. 9 provide descriptions of
acetic acid and oxygen penetration and biomass growth rates in
granules. For the representative conditions tested (see Wu and
Hickey [23] and Su and Yu [24] for details) the simulations
showed that substrate penetration would be very similar in both
systems (penetration for the nitrogen-supplemented system is
shown). Acetic acid can be expected to fully penetrate all
granules and oxygen can be expected to penetrate to the centre
of granules with radii of less than 1.0 mm. For larger granules,
significant environments with low/no dissolved oxygen are
predicted. So, for large-size aerobic granules, acetic acid
(organic substrate) is not a limiting factor, rather the whole
microbial process would be dominated by the availability of
DO [32].
Fig. 9 shows the effective biomass growth in granules
predicted in the nitrogen-fixing (Fig. 9C) and nitrogen-
supplemented (Fig. 9D) systems. The predictions are derived
by considering microbial kinetics in granules of various sizes. It
can be seen that growth in a nitrogen-supplemented system is
greatest in small granules as the substrates (acetic acid and
Fig. 8. SVIs of the aerobic granular systems (^ = N-fix system and � = nitro-
n-supplemented system) [3pt median average included].
oxygen) are most readily available. However, the simulations
indicate that for a nitrogen-fixing system, effective biomass
growth is actually retarded in small granules, as the presence of
oxygen affects the growth yield.
4. Discussion
This work has demonstrated for the first time an effective
aerobic granular sludge system for the treatment of severely
nitrogen-limited influent. The system described in this work
functioned as a result of the establishment of a stable population
of nitrogen-fixing microorganisms, capable of abstracting their
nitrogen requirement from atmospheric dinitrogen. Dennis
et al. [3] summarized that the significant advantages of
nitrogen-fixing systems are:
(i) s
elf-regulation of nitrogen requirements, allowing for
substantially less operator intervention and monitoring, and
(ii) im
proved environmental performance, as N2-fixation
eliminates the potential for excess supplementation and
consequent discharge of nitrogen with the effluent.
This paper demonstrates that these advantages are also
relevant for granular N-fix systems. Indeed, as shown in
Fig. 5A, a granular N-fix system can result in discharges of
dissolved nitrogen approximately half of those released from a
nitrogen-supplemented granular system loaded at conventional
COD to nitrogen ratios.
With regard to carbon removal, the granular N-fix system
performed extremely well. Fig. 3 confirms that the granular N-
fix system is suitable for effective oxidation of carbonaceous
inputs, displaying comparable removal with that of the
conventional granular system having supplemental nitrogen
addition, and with acetate-based granular systems reported in
the literature [12].
Importantly, the performance of the granular N-fix system
did not appear to be compromised by the high dissolved oxygen
concentration (ranging from 5.5 to 7.5 mg L�1), achieved with
Fig. 9. Simulation of substrate penetration (nitrogen-supplemented system and effective biomass growth (hg) for 10 granules with various radii (R). (A) Acetic acid