thermophilic anaerobic digestion of source-sorted organic fraction of household municipal solid...
TRANSCRIPT
ARTICLE IN PRESS
Available at www.sciencedirect.com
WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 6 2 1 – 2 6 2 8
0043-1354/$ - see frodoi:10.1016/j.watres
�Corresponding auE-mail address:
journal homepage: www.elsevier.com/locate/watres
Thermophilic anaerobic digestion of source-sorted organicfraction of household municipal solid waste: Start-upprocedure for continuously stirred tank reactor
Irini Angelidakia,�, Xingxing Chena, Junbo Cuia, Prasad Kaparajua, Lars Ellegaardb
aInstitute of Environment and Resources -DTU, Building 113, Technical University of Denmark, DK-2800, Kgs. Lyngby, DenmarkbBurmeister and Wain Scandinavian Contractor A/S, Gydevang 35, DK-3450 Allerød, Denmark
a r t i c l e i n f o
Article history:
Received 15 February 2006
Received in revised form
11 May 2006
Accepted 17 May 2006
Keywords:
Anaerobic digestion
Batch
Full-scale
Lab-scale
Methane yield
SS-OFMSW
Start-up
Thermophilic
nt matter & 2006 Elsevie.2006.05.015
thor. Tel.: +45 45251429; [email protected] (I. Angelida
A B S T R A C T
Two feeding strategies for start-up of continuously stirred tank reactors (CSTR) treating
source-sorted organic fraction of household municipal solid waste (SS-OFMSW) at 55 1C
were evaluated. Two reactors were started up separately with a limited amount of initial
inoculum (i.e. 10% of the final volume of 3.5 l) and operated in a fed batch mode until the
reactors were filled (30 days). A reference reactor was filled up with 3.5 l of inoculum and
fed at a constant rate (11.4 g volatile solids (VS)/d). Loading at progressively increasing rate
(from 1.7 to 15 gVS/d), as calculated based on an activated biomass concept, showed
superior process performance compared to a fixed feed rate (5.7 gVS/d). Methane yield of
0.32 m3/kg VS was produced during the start-up in reactor filled at progressively increasing
rate and was comparable to the reference reactor. On the contrary, significant inhibition
due to volatile fatty acid (VFA) build-up, mainly due to butyrate, was noticed in the reactor
filled at constant rate. Thus, low initial and progressive increasing inoculum loading rate
could be used as a strategy for a successful start-up of CSTR treating SS-OFMSW as it
allowed a gradual acclimation of the biomass. Lab-scale results were further reaffirmed
from the start-up of a full-scale plant (7000 m3 total capacity) which was supplied with
inoculum corresponding to approx. 16% of final volume and operated in a fed batch mode
until the reactors were filled (58 days). Stable biogas production with low VFA (o3 g/L; based
on titration method) were noticed during the start-up period when fed at progressively
increasing rate. Thus, a controlled and reliable start-up procedure was found essential,
which could allow rapid process stabilization and time to focus on other technical aspects
of plant operation. In addition, the influence of substrate to inoculum amount (1.5–30% TS)
and temperature (5–65 1C) on anaerobic degradation and methane production of SS-OFMSW
was investigated in batch assays as a protocol for start-up procedure.
& 2006 Elsevier Ltd. All rights reserved.
1. Introduction
Anaerobic digestion of organic fraction of municipal solid
waste (OFMSW) has increasingly being studied and applied in
Europe, especially after the introduction of source-sorted (SS)
r Ltd. All rights reserved.
x: +45 45932850.ki).
collection (De Baere, 2000). Currently, more than one million
tonnes of organic wastes (wet weight) per year are digested in
dedicated industrial plants worldwide (Bolzonella et al.,
2003a). Earlier, full-scale plants treating OFMSW were oper-
ated only at mesophilic temperatures (Cecchi et al., 1993) and
ARTICLE IN PRESS
Table 1 – The characteristics of substrate (SS-OFMSW)before dilution
Parameter SS-OFMSW
TS (%w/w) 30
VS (%w/w) 24.4
pH 4.6
TotalVFA (g/kg-waste) 27.3
Acetate (g/kg-waste) 22.2
Propionate (g/kg-waste) 3.6
Iso-butyrate (g/kg-waste) 0.34
Butyrate (g/kg-waste) 0.50
Iso-valerate (g/kg-waste) 0.46
Valerate (g/kg-waste) 0.24
Total N (g-N/kg-waste) 6.5
NH4+-N (g-N/kg-waste) 1.5
WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 6 2 1 – 2 6 2 82622
reported to show good operational performance (Buhr and
Andrews, 1977). But with successful demonstration of anae-
robic treatment of OFMSW at thermophilic temperature both
in lab-scale (Wellinger et al., 1992) and full-scale (Cozzolino
et al., 1992), the thermophilic anaerobic process has gained
increased attention primarily due to its capacity for higher
loading rate and greater volumetric gas production potential
(Cecchi et al., 1991). In addition, the thermophilic process has
been established as a reliable and accepted mode of
fermentation as claimed by several plant operators (Mata-
Alvarez et al., 2000) because of enhanced hydrolysis, in-
creased organic solids destruction and better pathogen
control (Ahn and Forster, 2000; Lund et al., 1995). Despite
these advantages, application of the thermophilic process in
the past has been limited mainly due to difficulty in process
control, poor effluent quality and poor process stability
related to chronically high propionate concentrations (Kugel-
man and Guida, 1989). The latter problems were often
encountered in connection with or shortly after start-up.
Start-up is therefore an important step in establishing a
proper microbial community in thermophilic as well as other
biological treatment processes. Poor start-up in biological
treatment systems can lead to prolonged period of acclima-
tion (Wu et al., 2001) and ineffective removal of organic
matter (Griffin et al., 1998). Several researchers have dis-
cussed the general start-up procedure for biogas plants
treating solid waste (Bolzonella et al., 2003a, b; Fernandez et
al., 2001; Lepisto and Rintala, 1995). These studies reported
that successful start-up was related to a number of factors
e.g., seed sludge source, initial loading rate, hydraulic
retention time (HRT) and/or solids retention time (Fang and
Lau, 1996). Some of these studies also included a start-up
strategy such as use of seed material from plants treating
similar waste, gradual process acclimatization from meso-
philic to thermophilic and/or gradual organic load increase
and withholding feeding temporally after inoculation (e.g.,
Bolzonella et al., 2003a; Cecchi et al., 1993; Fernandez et al.,
2001; Fang and Lau, 1996; Van Lier et al., 1993).
For a successful start-up and to reach the design load
quickly, strategies such as the amounts of inoculum and the
initiation of the feeding should be designed to avoid
accumulation of anaerobic degradation intermediate pro-
ducts such as propionic acid and other volatile fatty acids
(VFA’s) and hydrogen, which could inhibit methanogenesis
and acetogenesis (Lepisto and Rintala, 1995). Several re-
searchers have considered and demonstrated that thermo-
philic anaerobic processes can be readily started with small
amounts of inocula from a thermophilic process or with any
mesophilic sludge having methanogenic activity (Cecchi
et al., 1993; Fang and Lau, 1996; Lepisto and Rintala, 1995;
Van Lier et al., 1992, 1993). However, inability to acquire large
quantities of thermophilic inocula as seed material and long
and unstable start-up of the thermophilic process when
inoculated with mesophilic seed in practice, has led to poor
reputation with respect to process stability. In this study,
feeding strategies for starting thermophilic reactors with only
an initial amount of 10% of the final volume were evaluated. A
new strategy was designed based on an ‘‘activated biomass’’
concept (see later section for definition of concept), where
loading is increased in relation to the content of ‘‘activated
biomass’’ in the reactor. In this way daily loading is applied in
proportion to the gradual build-up of microbial degradation
capacity, avoiding the risk of accumulation of VFA and other
intermediates. A reference reactor started with 100% inocu-
lum, was fed with the same amount of fresh substrate every
day. Furthermore, the influence of substrate to inoculum
amount (1.5–30% total solids, TS) and temperature (5–65 1C) on
anaerobic degradation and methane production of SS-
OFMSW were investigated in batch assays.
2. Material and methods
2.1. Origin of substrate and inoculum
SS-OFMSW from a full-scale biogas plant (Vaarst–Fjellerad
plant; Denmark) was used as substrate. At the biogas plant,
substrate was shredded and sorted in multiple steps to yield a
purified digester feed that was homogenized, enriched in
biodegradable organics and devoid of large pieces or plastic
stringers. In the lab, substrate was further blended and
shredded to smaller pieces to prevent clogging. The prepared
feed was then stored in plastic bags of approx. 2 kg at �20 1C
until use. Characteristics of the substrate used are presented
in Table 1.
Thermophilic digested manure from a full-scale biogas
plant (Vegger plant, Denmark) treating manure together with
industrial waste and OFMSW was used as inoculum.
2.2. Batch experiments
The effect of substrate to inoculum ratios by varying substrate
concentration on anaerobic degradation of SS-OFMSW was
performed in 550 ml glass serum bottles. To each assay, 150 ml
of inoculum and 100 ml of a mixture of SS-OFMSW and tap
water corresponding to different initial TS contents of 0%,
1.5%, 3%, 4.5%, 6%, 7.5%, 9% and 30% were added. Anaerobic
conditions in the assays were established by flushing the
headspace with nitrogen gas for 3 min. Assays were sealed
immediately with butyl rubber stoppers and aluminum
crimps. All experiments were conducted in duplicate and
incubated statically at 55 1C in a temperature controlled
ARTICLE IN PRESS
Table 2 – Percentage conversion of fresh feed to ‘‘activebiomass’’ based on feed activation concept for loading ata progressive increasing rate
Feed activation concept
Days ai Conversion (%)
1 a1 25
2 a2 20
3 a3 20
4 a4 15
5 a5 10
6 a6 5
7 a7 5
WAT E R R E S E A R C H 40 (2006) 2621– 2628 2623
chamber. Assays containing inoculum and water were used
as controls. Methane produced from inoculum alone was
subtracted from the methane produced in other serum
bottles for comparison of methane production from fresh
substrate. Experiments were terminated after 40 days when
no significant gas production was noticed.
In another series of batch assays, the effect of temperature
on methane production potential of SS-OFMSW with thermo-
philic inoculum was evaluated. To each assays, 150 ml of
inoculum and 100 ml of SS-OFMSW diluted to 4.5% TS with
tap water were added as described above and incubated at
temperatures of 5, 15, 20, 30, 37, 45, 55 and 65 1C. The
experiments were conducted in duplicate and incubated for
75 days.
Total 100Note: ai is the conversion percentage of substrate during the first 7
days after feeding.
2.3. Start-up procedure of CSTR treating SS-OFMSW
Three separate 4.5 litres (l) CSTRs referred to as R1, R2 and R3
with liquid volume of 3.5 l were operated at 55 1C with heating
water pumped through the reactor water jackets. Reactor
contents were mixed at 40 s/min by mechanical mixers. On
day 1, 3.5 l of inoculum (i.e. 100% inoculation) was introduced
into reactor R1 and operated continuously with 15 d HRT
(reference reactor). In reactors, R2 and R3, 350 ml of inoculum
(i.e. 10% inoculation relative to final volume) was introduced
prior to substrate addition. The substrate was diluted with tap
water in weight ratio of 1:4 to achieve 6% TS (w/w) feed. The
reactors were fed twice every day by peristaltic pumps.
Feeding strategies were loading at a progressive increasing
rate (from 1.7 to 15 gVS/d) in R2 and at constant rate of 5.7 gVS/
d in R3 while the reference reactor R1 was fed at a constant
rate of 11.4 gVS/d. Thus both R2 and R3 were filled up to final
volume 3.5 l in 30 d.
2.4. Calculation of loading rate for reactor fed at aprogressive increasing rate (R2)
Loading rate for R2 was calculated based on a feed ‘‘activation
concept’’. This concept is based on constant loading rate (10%)
of calculated ‘‘activated biomass’’ content. Initial inoculum
added to the reactor is considered as ‘‘activated biomass’’ and
subsequent addition of fresh biomass is assumed to be
converted gradually to ‘‘activated biomass’’ over a period of
7 days after feeding at a conversion rate as shown in Table 2.
During the first day, 25% of the feed is assumed converted into
‘‘activated biomass’’ and on the second day 20% and so on. In
a continuous loading situation the amount of ‘‘activated
biomass’’ on any given day is the amount of ‘‘activated
biomass’’ present on the previous day plus 25% of the fresh
feed from the previous day plus 20% of the fresh feed 2 days
before etc. This concept can easily be implemented in a work
sheet starting with initial inoculum at day 0. In a formula the
principle can be expressed as
ABn ¼ ABn�1 þX7
i¼1
Fn�iai,
where AB ¼ activated biomass; F ¼ daily feed; n ¼ day num-
ber; ai ¼ conversion percentage day i ¼ 1–7 (Table 2).
This concept results roughly in an exponential increasing
loading designed to follow the gradual formation of microbial
capacity. Relative loading rate can be adjusted during the
start-up phase based on VFA analysis.
2.5. Start-up of full-scale plant based on the ‘‘activatedbiomass’’ concept
The ‘‘activated biomass’’ start-up concept has also been
applied in the start-up of a full-scale plant (Lemvig Biogas
plant, Denmark). Lemvig biogas plant is a centralised biogas
plant treating manure in co-digestion with various types of
organic industrial waste. The plant consists of 3 reactors with
a total digester volume of approx. 7000 m3. An initial amount
of 1100 m3 of thermophilic inoculum (i.e. approx. 16% of final
volume) was supplied to the plant during the first 10 days
after the initial start-up from another thermophilic biogas
plants in the region. Feed batch operation was continued for
58 days, after which all 3 reactors were full and effluent
pumping was started. Intended daily loading was 5% of the
calculated ‘‘activated biomass’’ content according to a con-
cept as described above. However, actual influent could not
always follow the intended feed rate due to technical
disturbances (pump break downs etc.). The three reactors
were for practical reasons (mixing and heating) started-up
sequentially. R1 was the first to be filled and when full, R1
content was divided into R1+R2 and finally the R1+R2 contents
were divided into R1+R3. Therefore, the VFA analyses for R2
and R3 were available only after some time.
2.6. Analytical methods, laboratory experiments
pH was measured with Metrohm 744 pH meter immediately
after each sampling. Methane content in the biogas was
analysed using a gas chromatography (GC-14A) equipped with
TC detection. The biogas produced in CSTRs was measured
through gas meters (Angelidaki et al., 1992). VFA concentra-
tions were determined using GC5890-series II equipped with
flame ionisation detection (FID). Total nitrogen (sum of
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 6 2 1 – 2 6 2 82624
ammonia nitrogen and organic bound nitrogen), TS and VS
were determined according to Standard Methods (APHA,
1998).
3. Results and discussion
3.1. Effect of substrate-inoculum ratios by varyingsubstrate concentration (1.5–30% TS) on anaerobic digestionof SS-OFMSW at 55 1C
Table 3 shows the specific methane yields for SS-OFMSW
incubated with inoculum at varying substrate concentration.
After an initial lag phase (5–7 d), a slow initial methane
production was noticed in all assays (data not shown). This
phase was followed by a sharp increase in the methane
production due to degradation of accumulated intermediate
compounds. A similar pattern was reported in a study using
spent brewery grain and anaerobic sludge incubated at 37 1C
in batches at different inoculum-substrate ratios of 7%, 13%
and 20% (Fernandez et al., 2001).
Specific methane yields after 40 days of incubation, in
general, increased with decrease in SS-OFMSW solids content
(Table 3). The maximum specific methane yield of 0.41 m3/
kgVS was obtained at 1.5% and 3% TS, while 63–83% of this
value were achieved at 4.5–9% TS. At 30% TS, the process was
completely inhibited due to overloading, with a methane
production lower than that obtained from control. The slow
rates and prolonged methane production at TS contents
higher than 1.5% indicate inhibition. Jungersen and Ahring
(1994) also reported that degradation of liquefied manure was
most efficient at relatively low concentrations, whereas a
significant inhibition occurred at higher concentrations. The
high specific methane yield at 1.5% TS was attributed to the
fact that complete anaerobic degradation could be achieved
in those assays with high inoculum to substrate ratio (low
substrate concentration with constant inoculum volume),
without VFA accumulation reaching inhibiting levels. The
present results thus suggest that during start-up of a
conventional digester like CSTR, where the acid and methane
producing phases are together, it is necessary to optimise for
Table 3 – Specific methane yields of varying TS content ofSS-OFMSW after 40 days at 55 1C
TS content (%) Specificmethane yield(m3/kgVS added
waste)
Relativemethane yieldcompared to
1.5%TS
1.5 0.41 (1.3) 100
3 0.41 (1.5) 99.7
4.5 0.35 (1.5) 83.9
6 0.26 (1.6) 63.5
7.5 0.26 (1.7) 62.2
9 0.30 (2) 73.5
30 Inhibition Inhibition
Note: Values in parentheses are standard deviation.
a low initial amount of fresh organic material relative to
inoculum amount. In the present study, the initial amount of
fresh organic material was varied by dilution of the fresh
substrate. The same ratio between fresh organic material and
inoculum could also be achieved by applying a reduced
amount of undiluted fresh material, however without the
potential benefit of general dilution of potential inhibitors
and minor accumulation of intermediates. Upon successful
start-up, a gradual increase in TS of fresh feed can be
attempted.
3.2. Effect of temperature (5–65 1C) on anaerobic digestionof SS-OFMSW
The specific methane yields from assays incubated with
thermophilic inoculum at various temperatures for 75 days
are shown in Table 4. Methane production rates (data not
shown) and the final yields were generally higher at
temperature closer to that of original inoculum. The low
methane yields at 30 1C and below were probably due to the
fact that the bacterial culture was not adapted to the low
temperatures and a long adaptation period is required for
such a non-adapted thermophilic culture (Nozhevnikova
et al., 2002).
The present results demonstrate the importance of obtain-
ing inoculum adapted to the same temperature level as
intended for the process to be started-up or at least to seek
inoculum adapted to a temperature within 710 1C of the new
process. Otherwise long start-up periods are required for the
development and retention of high concentrations of active
and well balanced biomass inside the reactor. This is
especially important for thermophilic processes because only
9% thermophiles and 1% obligate thermophiles were reported
to be present in mesophilic sludge (Chen, 1983). The fact that
the highest specific methane production was achieve at 45 1C,
somewhat below the process temperature at the source of
inoculum could be due to temperature dependent ammonia
concentration.
Table 4 – Mean-specific methane yields from SS-OFMSW(4.5% TS) incubated in batch assays at different tem-peratures (5–65 1C)
Temperature(1C)
Specificmethane (m3/
kgVS)
Relativemethane
compared to1.5%TS
5 0.001 (0) 0.2
15 0.003 (0.01) 0.5
20 0.13 (0.5) 22.3
30 0.30 (1.2) 51.6
45 0.58 (2.2) 100
55 0.42 (1.5) 71.8
65 0.52 (0.46) 85.5
Note: Values in parentheses are standard deviation.
ARTICLE IN PRESS
R1 (100% inoculum)
0
3
6
9
VF
A c
ompo
nent
(g/L
)
R2 (Progressive increasing loading rate)
0
3
6
9
VF
A c
ompo
nent
(g/L
)
R3 (Constant loading rate)
0
3
6
9
0 10 20 30 40
Time (day)
VF
A c
ompo
nent
(g/L
)
Fig. 2 – Monitoring the VFA components during the start-up
of anaerobic digestion SS-OFMSW at 55 1C in CSTR: J,
acetate; K, propionate; &, butyrate; ’, iso-butyrate; n,
valerate; m, iso-valerate.
0
150
300
450M
etha
ne y
ield
(mL/
gVS
)
0
5
10
15
0 10 20 30 40
Time (day)
Load
ing
rate
(gV
S/d
)
Fig. 1 – Loading rate (below) and methane production (above)
during the start-up of thermophilic anaerobic digestion of
SS-OFMSW (6% TS) in CSTR: R1 (J, constant rate with 10%
inoculum), R2 (&, progressive increasing rate with 10%
inoculum) and R3 (n, constant loading rate with 100%
inoculum).
WAT E R R E S E A R C H 40 (2006) 2621– 2628 2625
3.3. Start-up procedure of continuously stirred tankreactor treating SS-OFMSW
Start-up process performance of the CSTRs is illustrated in
Fig. 1. Reactor R2 responded well to the loading strategy of a
progressively increasing rate, although R2 was inoculated at
10% volume compared to that of R1. The initial low-feeding
strategy approach in R2 has apparently (as intended) allowed
a gradual balance between the various degradation steps to
develop sufficient microbial degradation capacity thereby
avoiding any possible VFA build-up. Total VFA content in R2
dropped steadily from 5.2 to 1.3 g/L within 12 d and remained
under 0.2 g/L thereafter (Fig. 2). These VFA values indicate a
stable start-up process and were in accord with those values
of 0.4 g/L reported earlier during start-up process of mechani-
cally sorted (MS) OFMSW (Cecchi et al., 1993). Correspond-
ingly, the specific methane production in R2 increased at first
slowly (first 3 days) producing only 0.06 m3/kgVS, increasing
sharply on day 4 (0.11 m3/kgVS) and then steadily thereafter
to produce 0.32 m3/kgVS at the end (Fig. 1). A similar
development in biogas production was noticed in the
reference reactor R1 fed at constant rate. Specific methane
yields of 0.31–0.32 m3/kgVS produced in R1 and R2 respectively
were comparable to other results obtained during thermo-
philic treatment of MS-OFMSW. For instance, Pavan et al.
(2000) reported a methane yields of 0.32 m3/kgVS obtained
during thermophilic semi-dry digestion of OFMSW (25%TS)
fed at loading rate of 9.7 kgVS/m3 d. Feeding R3 (also inocu-
lated with 10% of final volume as R2) at a constant rate on the
contrary resulted in a sharp increase in VFA concentration,
especially on day 7 (6.7 g/L), and resulted in a very low specific
methane production (0.10 m3/kgVS) which even remained low
throughout the start-up period (Fig. 2). This low specific
methane yield in R3 was apparently due to process inhibition
caused by VFA accumulation (Bolzonella et al., 2003a). After
day 15, methane production in R3 increased steadily to reach
0.20 m3/kgVS in the end but remained much lower than that
produced in R1 and R2.
Among the measured VFA components, acetate, propionate
and n-butyrate were the predominant acids. Specific distribu-
tion of VFAs in the three reactors indicated that with time,
acetate in R1 decreased gradually (from 6.8 to o0.06 g/L)
whilst propionate decreased with some fluctuation. Never-
theless, total VFA concentrations in R1 remained at low levels.
In R2, a sharp decrease in acetate was noticed, the concentra-
tions of propionate and butyrate however increased during
the initial 12 and 5 days, respectively. On the contrary, acetate
concentration in R3 remained much higher than the values of
propionate and butyrate. As a simultaneous response to
inhibition, the butyrate concentration in R3 was observed to
increase from the initial 0.17 to 4.9 g/L within the first 7 d.
Thereafter, the concentration decreased with time. The
acetate concentration on the other hand fluctuated from 2.4
to 3.6 g/L during the first 5 d and later decreased promptly to
very low levels within 2 days. The increase in propionate
concentration was found to be around day 5, which increased
sharply and reached 1.0 g/L on day 12 but later decreased
slowly with time. Thus in this study, the butyrate rather than
the propionate or acetate, was found to be the most
important parameter for process monitoring. These results
were not in accord with previous studies, where the first sign
of reactor overload was reported to be either due to a
chronical build-up of propionic acid in the CSTR process
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 6 2 1 – 2 6 2 82626
(Marchaim and Krause, 1993), or acetic acid accumulation in
both CSTRs and reactors with attached growth (Bolzonella et
al., 2003b; Cobb and Hill, 1991). Nevertheless, this observation
reaffirms the suggestion made by Ahring et al. (1995), that
VFA concentration during the start-up process is the best and
most reliable parameter to evaluate process stability.
The present results thus suggests that thermophilic CSTR
for treating SS-OFMSW and other high-strength substrates
can be started up successfully and relatively fast provided
thermophilic inoculum can be supplied and a careful loading
procedure is used, such as the ‘‘activated biomass’’ concept
used for R2 above. A similar start-up strategy such as use of
seed material from plants treating similar waste and gradual
organic load increase were reported (e.g., Bolzonella et al.,
2003a; Fernandez et al., 2001; Van Lier et al., 1993). Based on
the ‘‘activated biomass’’ concept, substrate could be added
progressively in a controlled manner until the designed
working volume was reached. Once the designed working
volume is attained, daily feeding and extraction can be
initiated and final design loading rate should be possible
shortly after. In practice (start-up of full-scale plants), the
‘‘activated biomass’’ concept can be adapted to handle also a
gradual supply of inoculum during the initial operating
period, as inoculum may only be available in limited daily
batches (normal daily extraction amount) from other plants
and without disturbing its operation. The controlled feeding
according to the ‘‘activated biomass’’ concept, designed to
avoid accumulation of intermediates, is also ideal to allow for
0
2
4
6
0 15 30 45
Time (d
TV
FA
(g/
L)
0
3
6
9
Bio
gasP
rod.
(100
0 m
3 /d
)
0.0
0.3
0.5
0.8
1.0
Sp.
bio
gas
Pro
d.(m
3 /k
gVS
)
Fig. 3 – Biogas production (J) and filling volume (’), specific bio
acids (TVFA: titration method; J reactor 1, ’ reactor 2 and K re
centralized biogas plant, Denmark) with daily loading at 5% of ca
concept’’.
less frequent VFA sampling/monitoring, as the loading
philosophy in itself should ensure against major imbalance
occurring between VFA analyses, while periodic VFA analyses
are sufficient to adjust the relative loading rate according to
actual process capacity.
3.4. Start-up of full-scale plant based on the ‘‘activatedbiomass’’ concept.
The start-up process performance of a thermophilic full-scale
plant is illustrated in Fig. 3. Reactor R1 (the first reactor to be
started-up) showed to respond well to the loading strategy
based on the feed activation concept. Following the batch-fed
mode filling up period (58 days), stable biogas production was
noticed with a pH around 8 and VFA content o3 g/L (based on
titration). The low VFA content during and after the start-up
period reaffirmed that the initial low-feeding strategy ap-
proach has avoided possible VFA build-up through a gradual
acclimation of the biomass. Filling the other two reactors
(R2 and R3) by splitting the R1 effluent did not affect the start-
up in either R2 or R3 and VFA content in the respective
reactors remained o4 g/L (Fig. 3). Throughout the start-up
period, pH remained in the region of 8 (70.2) and the
ammonia/ammonium concentration only increased slightly
from an initial level of approx. 3 g/L to a stable level of approx.
4 g/L at the end of the start-up period. These results indicate
that the inoculum, apart from being a thermophilic, had
also similar chemical/ionic properties ensuring a smooth
60 75 90
ays)
0
3
6
9
12
15
Fill
ing
Vol
(100
0 m
3 )
0.0
0.3
0.6
0.9
Load
ing
rate
(m3
/m3 )
gas production (J) and loading rate (–) and total volatile fatty
actor 3) during the start-up of full-scale biogas plant (Limvig
lculated active biomass according to the ‘‘activated biomass
ARTICLE IN PRESS
WAT E R R E S E A R C H 40 (2006) 2621– 2628 2627
adaptation to the fresh feed supplied to the new plant. This
example on the application of a controlled start-up loading
procedure for a full-scale thermophilic process, which has
been repeated with success on later plants, demonstrates that
process stabilization can be smooth allowing full attention to
be focused on other technical aspects of starting up a new
plant with the aim of reaching full design production
(income) as fast as possible.
4. Conclusion
Results from batch experiments showed that maximum
methane yields (0.43 m3/kgVSadded waste) and best degradation
rates for SS-OFMSW could be achieved at relatively low fresh
TS content (1.5%) and under thermophilic conditions. At
higher TS (3–30%), the process would be inhibited initially and
the magnitude of the inhibition would increase with increase
in TS.
The study also demonstrated that full-scale plants to treat
SS-OFHSW can be started-up successfully and relatively fast
provided at least 10–15% of final inoculum volume was
supplied and a careful loading procedure such as the
‘‘activated biomass’’ concept was adopted. Low and progres-
sively increasing loading (1.7–15 gVS/d) was found to be an
ideal start-up strategy compared to high constant loading rate
(5.7 gVS/d). Thus, a controlled start-up loading procedure
would not only allow rapid process stabilization but also
provide time to focus on other technical aspects of new plant
operation with the aim of reaching full design production
(income) as fast as possible.
Acknowledgement
Danish Research Council (FTP) is greatly acknowledged for
partial funding of this project (33031-0029/EFP-05).
R E F E R E N C E S
Ahn, J.H., Forster, C.F., 2000. A comparison of mesophilic andthermophilic anaerobic upflow filters. Bioresour. Technol. 73(3), 201–205.
Ahring, B.K., Sandberg, M., Angelidaki, I., 1995. Volatile fatty acidsas indication of process imbalance in anaerobic digesters.Appl. Microbiol. BioTechnol. 43, 559–565.
Angelidaki, I., Ellegaard, L., Ahring, B.K., 1992. Compact auto-mated displacement gas metering system for measuring lowgas rates from laboratory fermentors. Biotechnol. Bioeng. 39,351–353.
APHA/AWA/WPCF, 1998. Standard Methods for the Examina-tion of Water and Wastewater, 20th ed, Washington, DC.
Bolzonella, D., Battistoni, P., Mata-Alvarez, J., Cecchi, F.,2003a. Anaerobic digestion of organic solid wastes: processbehaviour in transient conditions. Water Sci. Technol. 48 (4),1–8.
Bolzonella, D., Innocenti, L., Pavan, P., Traverso, P., Cecchi, F.,2003b. Semi-dry thermophilic anaerobic digestion ofthe organic fraction of municipal solid waste: focusingon the start-up phase. Bioresour. Technol. 86 (2),123–129.
Buhr, H.O., Andrews, J.F., 1977. Thermophilic anaerobic digestionprocess. Water Sci. 11 (2), 129–143.
Cecchi, F., Pavan, P., Mata-Alvarez, J., Bassetti, A., Cozzolino, C.,1991. Anaerobic-digestion of municipal solid-waste—thermo-philic vs mesophilic performance at high solids. WasteManage. Res. 9 (4), 305–315.
Cecchi, F., Pavan, P., Musacco, A., Mata-Alvarez, J., Sans, C.,Defaveri, D., 1993. Monitoring a fast thermophilic re-start-upof a digester treating the organic fraction of municipal solid-waste. Enviorn. Technol. 14 (6), 517–530.
Chen, M., 1983. Adaptation of mesophilic anaerobic sewagefermentor population to thermophilic temperatures. Appl.Enviorn. Microbiol. 45 (4), 1271–1276.
Cobb, S.A., Hill, D.T., 1991. Volatile fatty-acid relationships inattached growth anaerobic fermenters. Trans. ASAE 34 (6),2564–2572.
Cozzolino, C., Bassetti, A., Rondelli, R., 1992. Industrial applicationof semi-dry digestion process of organic solid waste. In:Cecchi, F., Mata-Alvarez, J., Pohland, F.G. (Eds.), Proceedings ofthe International Symposium on Anaerobic Digestion of Solid Waste.Venice, pp. 551–555.
De Baere, L., 2000. Anaerobic digestion of solid waste: state-of-the-art. Water Sci. Technol. 41 (3), 283–290.
Fang, H.H.P., Lau, I.W.C., 1996. Startup of thermophilic (55 1C)UASB reactors using different mesophilic seed sludges. WaterSci. Technol. 34 (5-6), 445–452.
Fernandez, B., Porrier, P., Chamy, R., 2001. Effect of inoculum-substrate ratio on the start-up of solid waste anaerobicdigesters. Water Sci. Technol. 44 (4), 103–108.
Griffin, M.E., McMahon, K.D., Mackie, R.I., Raskin, L., 1998.Methanogenic population dynamics during start-up of anae-robic digesters treating municipal solid waste and biosolids.Biotechnol. Bioeng. 57 (3), 342–355.
Jungersen, G., Ahring, B.K., 1994. Anaerobic-digestion of liquefiedcow manure pretreated by catalytic liquefaction. Water Sci.Technol. 30 (12), 385–394.
Lepisto, S.S., Rintala, J.A., 1995. Thermophilic anaerobic-digestionof the organic fraction of municipal solid-waste—start-upwith digested material from a mesophilic process. Enviorn.Technol. 16 (2), 157–164.
Lund, B., Benedixen, H.J., Have, P., Ahring, B. (Eds.), 1995.Reduction of Pathogenic Bacteria and Viruses by AnaerobicDigestion. Management of Urban Biodegradable Wastes. Jamesand James Ltd., London.
Kugelman, I.J., Guida, V.G., 1989. Comparative evaluation ofmesophilic and thermophilic anaerobic digestion. EPA/600/S2-89/001, August 1989.
Nozhevnikova, A.N., Nekrasova, V.K., Parshina, S.N., Shatilova,K.A., Simankova, M.V., Kotsyurbenko, O.R., 2002. Anaerobicmicroorganisms and communuties in low temperature wastetreatment. In: Kalyuzhnyi, S. (Ed.), Proceedings of the 7th FAO/SREN-Workshop on Anaerobic Digestion for Sustainability inWaste(water) Treatment and Re-use. Moscow, Russia, pp.24–32.
Marchaim, U., Krause, C., 1993. Propionic to acetic-acid ratios inoverloaded anaerobic-digestion. Bioresour. Technol. 43 (3),195–203.
Mata-Alvarez, J., Mace, S., Llabres, P., 2000. Anaerobic digestion oforganic solid wastes. An overview of research achievementsand perspectives. Bioresour. Technol. 74 (1), 3–16.
Pavan, P., Battistoni, P., Mata-Alvarez, J., Cecchi, F., 2000. Perfor-mance of thermophilic semi-dry anaerobic digestion processchanging the feed biodegradability. Water Sci. Technol. 41 (3),75–81.
Van Lier, J.B., Grolle, K.F.C., Stams, A.J.M., de Maccario, E.C.,Lettinga, G., 1992. Start-up of a thermophilic upflow anaerobicsludge bed (UASB) reactor with mesophilic granular sludge.Appl. Microbiol Biotechnol. 37, 130–135.
ARTICLE IN PRESS
WAT E R R E S E A R C H 4 0 ( 2 0 0 6 ) 2 6 2 1 – 2 6 2 82628
Van Lier, J.B., Grolle, K.C.F., Frijters, C.T.M.J., Stams, A.J.M.,Lettinga, G., 1993. Effect of acetate, propionate and butyrateon the thermophilic anaerobic degradation of propionate inmethanogenic sludge and defined cultures. Appl. Enviorn.Microbiol. 57, 1003–1011.
Wellinger, A., Baserga, U., Egger, K., 1992. New systems for thedigestion of solid-wastes. Water Sci. Technol. 25 (7), 319–326.
Wu, J.H., Liu, W.T., Tseng, I.C., Cheng, S.S., 2001. Characterizationof a 4-methylbenzoate-degrading methanogenic consortiumas determined by small-subunit rDNA sequence analysis.J. Biosci. Bioeng. 91 (5), 449–455.