Hydrolysis and fermentation of activated sludge to enhance biological phosphorus removal

doi:10.2166/wst.2006.406Hydrolysis and fermentation of activated sludge to enhance biological phosphorus removal
J. Vollertsen*, G. Petersen** and V.R. Borregaard***
*Aalborg University, Section of Environmental Engineering, Sohngaardsholmsvej 57, 9000 Aalborg,
Denmark (E-mail: [email protected])
***Aalborg Kommune, Kloakforsyningen, Stigsborg Brygge 5, Postboks 222, 9400 Nørresundby, Denmark
Abstract The conventional mainstream enhanced biological phosphorus removal (EBPR) process depends
on the quality of the raw incoming wastewater. An alternative sidestream EBPR process is presented, where
the substrates for storage by the polyphosphate accumulating organisms (PAOs) instead come from
hydrolysis of the return activated sludge. This process is studied in full-scale at two treatment plants and
quantified by means of phosphorus release rates and readily biodegradable COD (RBCOD) accumulation
rates. It was seen that not only was a significant amount of RBCOD stored by PAOs but an approximately
equal amount was accumulated in the sidestream hydrolysis tank and made available for the subsequent
nitrogen removal process. The phosphorus release of the sludge with and without addition of different
substrates was furthermore studied in laboratory scale. The study showed that the process is promising and
in a number of cases will have significant advantages compared with the conventional mainstream EBPR
design.
Introduction
The activated sludge enhanced biological phosphorus removal (EBPR) process has
become widely implemented, as the process has economic and environmental advantages
compared with chemical precipitation. EBPR reduces the cost for chemicals and, more
importantly, it reduces the amount of surplus sludge produced, resulting in reduced
sludge management costs and reduced environmental impacts from sludge disposal.
Avoiding or reducing the chemical sludge produced by phosphorus precipitation, EBPR
also allows a higher organic matter content of the activated sludge.
The EBPR process is dependent on redox conditions alternating between anaerobic and
aerobic/anoxic conditions. During anaerobic conditions, readily biodegradable COD
(RBCOD) is stored within the cells and under aerobic/anoxic conditions this storage is uti-
lised. The microorganisms store energy-rich compounds under aerobic/anoxic conditions so
that they possess an energy reserve to fuel storage of RBCOD under anaerobic conditions.
Some bacteria – polyphosphate-accumulating microorganisms (PAOs) – store phos-
phorus, while other bacteria – glycogen-accumulating microorganisms (GAOs) – store
glycogen (e.g. Wentzel et al., 1986; Liu et al., 1996; Carucci et al., 1999a,b).
The RBCOD in the form of VFAs can be taken up and stored as polyhydroxyalkano-
ates (PHAs) (e.g. Wentzel et al., 1986; Satoh et al., 1996). In addition, Carucci et al.
(1999a,b) show that glucose and other carbohydrates can be directly taken up during
anaerobic conditions and stored as glycogen. The organisms involved can use polyphos-
phate hydrolysis or fermentation as the energy source. They observed that carbohydrates
could be stored without involving PAO-like or GAO-like organisms, and theorise that the
competition between the different groups of microorganisms capable of storing RBCOD
W ater
S cience
echno lo
g y V o l 5 3 N o 1 2 p p 5 5 – 6 4 Q
IW A
P ub
lishin g
2 0
0 6
55doi: 10.2166/wst.2006.406
can explain the high variability in P-removal rates often encountered in full-scale
treatment plants. Also, for example, Shintani et al. (2000) describe a bacterium capable
of anaerobic glucose uptake and glycogen synthesis.
The conventional mainstream plant layout for EBPR contains an anaerobic tank to
which all incoming raw wastewater is fed, together with the activated sludge returned
from the clarifier (Figure 1). Biomass capable of anaerobic substrate storage is favoured
as it takes up the RBCOD of the incoming wastewater, i.e. PAOs, GAOs and other types
of substrate storing microorganisms gain a competitive advantage. With respect to EBPR,
only organisms that use stored polyphosphate as the energy source are desirable, as they
accumulate phosphorus and thereby allow the biological phosphorus removal process to
take place. Other types of RBCOD storing organisms introduce an unwelcome loss of
RBCOD, and their presence results in an increased ratio of RBCOD to phosphorus
needed to achieve biological phosphorus removal.
In the plant design sketched in Figure 1, the RBCOD for the EBPR is supplied by the
incoming wastewater, and the design becomes dependent on the quality of this waste-
water. Especially during extended periods with high precipitation, the wastewater quality
can deteriorate and the residence time in the anaerobic zone of the treatment plant can
become crucially low, resulting in a poor efficiency of the biological phosphorus removal
process. When wastewater quality is poor in terms of RBCOD, substrate can be added as,
for example, molasses. For treatment plants with primary treatment, primary sludge can
be hydrolysed and dosed as extra substrate for the EBPR process (e.g. Brinch et al.,
1994; Canziani et al., 1995). On the other hand, when primary treatment is combined
with sludge digestion to produce biogas, hydrolysis of primary sludge will diminish the
biogas yield.
In Denmark, typical levels of VFAs in raw wastewater are rather low, and EBPR must
rely on other forms of RBCOD and on fermentation in the anaerobic tank. For example,
for raw wastewater entering one of the treatment plants studied (WWTP West), the
median RBCOD was found to be 150 g COD m23, whereas VFA was as low as 1 g m23
(Vollertsen, 2002). As an alternative to addition of substrate and to primary sludge
hydrolysis, nearly 30 Danish treatment plants have, since 1996, implemented anaerobic
hydrolysis of return activated sludge, taking a sidestream of the return sludge from the
clarifier and letting it undergo anaerobic hydrolysis (Figure 2) (Petersen, 2002, 2003).
The first such design was reported by Andreasen et al. (1997). Later experiences have
shown that the optimum configuration of the sidestream EBPR design is achieved by
taking 4–7% of the return flow into the sidestream hydrolysis tank for a retention time
Inlet An- aerobic Anoxic/aerobic
Inlet
Excess sludge
Figure 2 A sidestream EBPR design with anaerobic hydrolysis of the return activated sludge
J.V o
llertsen et
al.
56
of 30–40 h (Petersen, 2002, 2003). If necessary, the process rates in the hydrolysis tank
can be boosted by the addition of small amounts of primary sludge, molasses, septic
sludge or similar COD sources. Furthermore, the sidestream hydrolysis tank can be used
for equalisation of, for example, nitrogen-rich reject water from the dewatering of
digested sludge.
It is the objective of this study to investigate the operation of sidestream EBPR with
anaerobic hydrolysis of the return activated sludge at two Danish treatment plants. The
activity of PAOs under different substrate conditions is studied, as is the formation of
RBCOD not stored but accumulated in the sidestream hydrolysis tanks.
Methods
Study sites
Experiments were conducted at two activated sludge treatment plants in Aalborg,
Denmark. Both treatment plants apply nitrogen removal and biological phosphorus
removal and are equipped with thermophilic sludge digesters. Wastewater treatment plant
west (WWTP West) has a capacity of 330,000 person equivalents and is equipped with
primary settlers. Wastewater treatment plant east (WWTP East) has a capacity of
100,000 person equivalents and is without primary settlers. At both treatment plants, one
process tank has been converted from conventional operation (Figure 1) to sidestream
EBPR operation (Figure 2). Both tanks have a volume of 1,800 m3. Experiments were
conducted at WWTP West from November 2002 to August 2003 and at WWTP East
from March to June 2004.
Phosphorus release rate
The activity of the PAOs was determined by the ability of the sludge under anaerobic
conditions to release phosphorus stored under aerobic conditions (Figure 3). The activity
was determined in two parallel operated glass reactors of 2.3 L. The reactors were
continuously stirred by magnetic stirrers and kept at room temperature (approximately
21 8C). At the start of an activity measurement, 30 gPO4-P m23 was added and the batch
kept aerobic until the main part of the phosphate was stored by the PAOs. The reactor
was then closed and kept anaerobic. pH was not controlled during the experiments.
The pH of all samples was 7.0 ^ 0.2. The phosphorus release was measured for about
24 h, but only the first 5–7 h were used to calculate phosphorus release rates.
The different substrates added to the phosphorus release experiments were:
† sodium acetate to a concentration of 100 g m23 of acetate;
† molasses to a concentration of 1 mL L21 of molasses;
† supernatant of activated sludge from the sidestream hydrolysis tanks. The supernatant
was obtained by filtration and added in concentrations from 370 to 430 mL L21.
Addition of substrate
P
Figure 3 Determination of phosphorus release activity in a batch reactor
J.V o
llertsen et
al.
57
† Primary sludge in concentrations from 7 to 11 mL L21. Initial experiments on six
samples had shown that the primary sludge RBCOD content on average was
4,600 gCOD m23, of which 1,600 g m23 were VFAs. Of the measured VFAs, acetate,
propionate and iso-butyrate were present in significant concentrations, whereas lactate
and n-butyrate were absent.
† Reject water from dewatering of thermopilic digested sludge was added to concen-
trations of 100 mL L21, 250 mL L21 and 500 mL L21. Of the measured VFAs, acetate,
propionate, n-butyrate and iso-butyrate were present in varying concentrations,
whereas lactate was absent.
Determination of wastewater biodegradability
Oxygen uptake rate (OUR) determination on the supernatant of hydrolysed activated
sludge was carried out to identify the amount of RBCOD leaving the sidestream hydroly-
sis tanks. The supernatant was obtained by filtration. The OUR was interpreted by
means of a concept describing aerobic transformation processes in wastewater (Hvitved-
Jacobsen et al., 1998). The concept used is based on the COD transformation part of the
activated sludge model no. 1 and 2 (Henze et al., 1987, 1995). In contrast to these
models, the concept applied does not take decay of biomass into account, as this process
is believed to be of minor importance during conditions with high concentrations of
RBCOD. Instead, maintenance energy requirement of the biomass is accounted for.
To obtain a better characterisation of the hydrolysable substrate, two substrate fractions
with different process kinetics were applied according to Sollfrank (1988).
The mathematical formulation of the concept is presented in Table 1. Model
parameters and components are listed in Table 2. The COD fractions and most of
Table 1 The model for characterisation of COD in hydrolysed return sludge (Hvitved-Jacobsen et al., 1998)
Component Process SS XS,fast XS,slow XBw 2SO Process rate
Aerobic growth 21 YH
KX ;fastþXS;fast =XBw XBw
KX ;slowþXS;slow =XBw XBw
*If SS is not present in sufficient concentration XBw is used to supply the remaining COD (endogenous respiration)
Table 2 Parameters in the model for characterisation of COD in hydrolysed return sludge, cf. Table 1
Symbol Description Unit Obtained as
mH Maximum specific growth rate constant of XBw
d21 Simulation result. If no SS then 6.0
kh,fast Hydrolysis rate constant of XS,fast d21 Simulation result kh,slow Hydrolysis rate constant of XS,slow d21 Simulation result KS Saturation constant of SS gCOD m23 Fixed at 1.0 KX,fast Saturation constant, hydrolysis of XS,fast gCOD gCOD21 Simulation result KX,slow Saturation constant, hydrolysis of XS,slow gCOD gCOD21 Simulation result qm Maintenance energy requirement
rate constant d21 Fixed at 1.0
SS Readily biodegradable substrate gCOD m23 Simulation result XBw Heterotrophic biomass gCOD m23 Simulation result XS,fast Fast hydrolysable substrate gCOD m23 Simulation result XS,slow Slowly hydrolysable substrate gCOD m23 XS,slow ¼ CODtot 2
SS 2 XBw 2 XS,fast
YH Yield constant of XBw gCOD gCOD21 Fixed at 0.6
J.V o
llertsen et
al.
58
the model parameters are obtained by simulation of the OUR curves. For those
parameters where this was not feasible, parameters were obtained from Vollertsen and
Hvitved-Jacobsen (1999).
OUR was measured in two completely filled and independent batch reactors. The
reactors were stirred and kept at 20 ^ 0.2 8C. The wastewater was aerated until 75% of
oxygen saturation was exceeded and the following oxygen consumption, time and tem-
perature logged. Aeration started when the oxygen concentration dropped below 30%.
The reactors were equipped with expansion chambers that were partly filled with stagnant
water, efficiently inhibiting reaeration during the measurement of oxygen uptake.
The reactors were made of stainless steel of food processing quality. OUR was calculated
from the slope of the measured DO, finding a moving average between adjacent DO
values.
Two different OUR-based methodologies were used to find the biodegradability of the
COD. The first method was used when relatively high concentrations of readily biode-
gradable substrates were present. In this case, the supernatant was added to an ongoing
OUR measurement on return activated sludge (Figure 4). The sludge was diluted to a
suspended solids concentration of either approximately 600 or 1,200 gSS m23. The second
method was used when the sample contained comparatively little RBCOD. In this case,
the supernatant was seeded with approximately 200 gSS m23 return activated sludge and
the OUR subsequently measured (Figure 5). The RBCOD was calculated as the sum of
the two COD fractions SS and XS, fast (Hvitved-Jacobsen, 2002). RBCOD on primary
sludge was found applying the first method (Figure 4).
The RBCOD on supernatant from the hydrolysis tank at WWTP West was measured
during winter 2002/03. Average values during this period were: temperature, 10.5 8C;
retention time in the sidestream hydrolysis tank, 35 h ; SS concentration, 10.9 kg m23.
At WWTP East, RBCOD was measured in March 2004 and average values during
Addition of supernatant
Background oxygen uptake of the activated sludge
Time
Figure 4 Addition of supernatant to an ongoing OUR measurement on activated sludge
O xy
ge n
up ta
ke r
at e
(O U
Background oxygen uptake of the activated sludge
Time
J.V o
llertsen et
al.
59
the period were: temperature, 11.3 8C; retention time in the sidestream hydrolysis tank,
30 h ; SS concentration, 12.2 kg m23.
Chemical parameters
COD was measured using a LANGE COD cuvette test and phosphate was measured
using a LANGE orthophosphate cuvette test. Samples for analysis of soluble COD were
filtered through a 0.45mm filter. SS was determined according to APHA et al. (1998).
VFAs (lactate, acetate, propionate, n-butyrate and iso-butyrate) were analysed on a
Dionex ionchromatograph with a suppressed conductivity detector, 0.2 mM NaOH as the
mobile phase and an IonPac AS11 column.
Results and discussion
Phosphorus release activity
The phosphorus release rate is governed by the amount and activity of organisms present
and the availability of suitable substrates. When no substrate was added to the batch
experiments on return activated sludge, or when primary sludge or supernatant from the
sidestream hydrolysis tank was added, phosphate release rates were significantly lower
compared to adding molasses or acetate (Tables 3 and 4). The reduced rates consequently
reflected the formation of substrates suitable for storage by the PAOs.
Analysing the data of Johansson et al. (1996), they obtained similar rates for phos-
phorus release from hydrolysis of return activated sludge without the addition of
substrates, namely approximately 0.2 gP kgSS21 h21. When they added acetate, that rate
increased to approximately 7 gP kgSS21 h21, i.e. significantly above what was seen for
the activated sludge from WWTP West and WWTP East.
Readily biodegradable COD
The supernatant from both sidestream hydrolysis tanks contained significant amounts of
RBCOD; however, only an insignificant fraction were VFAs (Table 5). The main part of
the VFAs at WWTP East was acetate (97%) and the remaining 3% was lactate, whereas
Table 3 Phosphorus release rates determined on return activated sludge from WWTP West
Substrate addition N Median [gPkgSS21 h21] 25th and 75th percentile
[gPkgSS21 h21]
None 12 0.19 0.17 and 0.21 Supernatant from the sidestream hydrolysis tank
2 0 –
Primary sludge 8 0.27 0.24 and 0.29 Molasses 4 1.6 0.91 and 2.09
Table 4 Phosphorus release rates determined on return activated sludge from WWTP East
Substrate addition N Median [gPkgSS21 h21] 25th and 75th percentile
[gPkgSS21 h21]
None 4 0.09 0.05 and 0.14 Supernatant from the sidestream hydrolysis tank
4 0.07 0.05 and 0.12
Acetate 4 1.15 0.99 and 1.33 Acetate plus reject water from sludge dewatering
3 1.53 1.24 and 1.85
Molasses 1 2.22 –
al.
60
propionate, n-butyrate and iso-butyrate were not found. In WWTP West the picture was
less clear, and all VFAs but propionate were found – however, never in the same sample.
Addition of supernatant from the hydrolysis tanks to phosphorus release tests did not
result in an increase in the observed release rates (Tables 3 and 4). This indicates that
the RBCOD found by OUR measurement was not suitable for uptake by the PAOs in the
treatment plants. This conclusion is supported by the full-scale measurements. During the
period where OUR was measured, the phosphate concentration in the hydrolysed return
activated sludge of WWTP West and WWTP East was on average 22 gP m23 and
31 gP m23, respectively. The batch tests on the other hand showed that a total of
40–50 gP m23 could be released within a day. Furthermore, during warmer periods, the
phosphate content in the hydrolysis tank became some 50% higher compared with what
was observed during winter periods, i.e. the PAOs were potentially capable of taking up
additional substrate, but could not utilise the RBCOD present.
Full-scale operation of the sidestream hydrolysis tanks
The full-scale measurements of the operation of the sidestream hydrolysis tanks indicated
that phosphorus release was temperature dependent at both treatment plants (Figure 6).
Especially the hydrolysis tank at WWTP East showed a clear tendency towards increased
rates with increasing temperature. As discussed above, the increase in phosphorus release
rate reflects an increase in the hydrolysis rate, as the potential phosphorus uptake rate of
the PAOs is significantly higher compared with what could be observed in the sidestream
hydrolysis tanks.
Comparing full-scale results (Figure 6) with laboratory-scale results (Tables 3 and 4),
reasonable agreement between observed phosphate release rates is found. For the side-
stream hydrolysis tank of WWTP East, the median of the full-scale rates was
0.10 gP kgSS21 h21, and laboratory experiments had a median of 0.09 gP kgSS21 h21.
Similarly, for WWTP West, the median of the rates for the sidestream hydrolysis tank
was 0.07 gP kgSS21 h21, and laboratory experiments had a median of 0.19 gP kgSS21 h21.
Table 5 RBCOD in the studied sidestream hydrolysis tanks
Treatment plant RBCOD VFAs
percentile [gVFAm23]
WWTP West 8 86 76 and 102 10 1.3 0.04 and 2.7 WWTP East 6 47 42 and 51 17 2.7 1.8 and 5.0
*Double determinations
10 12 14 16 18 20
Temperature [C] 8 10 12 14 16 18 20
Figure 6 Release of phosphorus in the sidestream hydrolysis tanks
J.V o
llertsen et
al.
61
The release rates depicted in Figure 6 resulted in average phosphorus concentrations
in the sidestream hydrolysis tanks of WWTP East and WWTP West of 41 and
28 gP m23, respectively; i.e. even though the rates as such seem rather low, both the
concentration of activated sludge and the retention time are high. Together these two
factors compensate for the low rate, and allow the PAOs to gain the competition advan-
tage needed to operate an EBPR process successfully.
Mass balance on the RBCOD formed by hydrolysis of return activated sludge
The hydrolysis of the return activated sludge resulted in the formation of RBCOD. Part of
this was stored by PAOs, GAOs and probably also by other substrate storing organisms
(e.g. Liu et al., 1996; Carucci et al., 1999a,b), and part was accumulated in the tanks.
The accumulated fraction was measured by the OUR method. However, the fraction
stored by substrate storing organisms could not be directly measured. Instead, it had to be
estimated based on the released phosphorus.
Some of the RBCOD produced by hydrolysis was fermented and stored by PAOs.
This fraction can, with good accuracy, be estimated from the phosphorus release together
with reported ratios of phosphorus released to VFAs stored. A number of studies have
investigated this ratio, and…