Water Research 36 (2002) 43694385

Comparative process stability and efciency of anaerobic

digestion; mesophilic vs. thermophilic
Moonil Kima,*, Young-Ho Ahnb, R.E. Speecea
a
Department of Civil & Environmental Engineering, Vanderbilt University, PO Box 1831, Station B, Nashville, TN 37235, USA
b
Department of Civil Engineering, Yeungnam University, Kyungsan 712-945, South Korea
Received 1 August 2001; received in revised form 1 March 2002; accepted 19 March 2002

Abstract

The comparative process stability and efciency of mesophilic (351C) and thermophilic anaerobic digestion (551C)
has been evaluated for four different reactor congurations, which are: daily batch-fed single-stage continuously stirred
tank reactor (CSTR), continuously fed single-stage CSTR, daily batch-fed two-phase CSTR, and daily batch-fed non-
mixed single-stage reactor. The results are discussed for three periods: (1) start-up, (2) steady state, and (3) organic
loading rate (OLR) increase until reactor failure (pH below 5.5). During the start-up, the single-stage CSTRs at both
temperatures showed the least stability, while the non-mixed single-stage reactors reached steady state in the shortest
time with relatively stable pH and low volatile fatty acid (VFA). In the case of the thermophilic non-mixed reactor,
efcient removal of propionate occurred but supplementation of nutrients (Ca, Fe, Ni, and Co) was required when
VFA increased. The results imply the importance of inorganic nutrients bioavailability. The comparative results of the
reactor performance at steady state clearly showed the superior performance of the thermophilic non-mixed reactor
with respect to lower VFA, higher gas production and volatile solids removal implying that microbial consortia
proximity can alleviate the problem of poor efuent quality in thermophilic system. During the OLR increase until
reactor failure, all thermophilic reactors except the thermophilic non-mixed reactor showed increases in propionate
concentrations as the OLR increased, while all mesophilic reactors except the mesophilic two-phase system showed little
increase in VFA concentrations. When all reactors had the same conditions with OLR increase, the continuously fed
reactors showed the lowest gas production, while the non-mixed reactors showed the highest gas production at both
temperatures. It is hypothesized that the non-mixing reactor conguration has closer microbial consortia proximity
than others. Therefore, the results in this study indicated the importance of microbial consortia proximity. A proposed
model for the effect of the distance between two syntrophic bacteria reasonably matched the data in this study. r 2002
Elsevier Science Ltd. All rights reserved.

Keywords: Mesophilic; Microbial consortia proximity; Nutrients; Reactor conguration; Thermophilic; Volatile fatty acid

1. Introduction destruction rate of organic solids, improved solids

Mesophilic temperature regimes (30401C) have long
been adopted for anaerobic digestion, showing good
operational performance. Thermophilic regimes (50
601C) have also been adopted for anaerobic digestion
showing several advantages, such as an increased
*Corresponding author. Tel.: +1-615-322-4348.
E-mail address: moonil.kim@vanderbilt.edu (M. Kim).
liquid separation, and increased destruction of patho-
genic organisms [13]. However, the use of thermophilic
anaerobic digestion has been limited, because of some
disadvantages like poor supernatant quality and poor
process stability related to chronically high propionate
concentrations [4].
It has been shown by previous studies that thermo-
philic microorganisms are characterized by their higher
substrate utilization and growth rates as well as higher

0043-1354/02/$ - see front matter r 2002 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 4 3 - 1 3 5 4 ( 0 2 ) 0 0 1 4 7 - 1
4370 M. Kim et al. / Water Research 36 (2002) 43694385
decay rate compared to mesophilic bacteria. Studies
undertaken by several researchers [57] showed that
thermophilic systems were capable of treating higher
organic loadings and had higher specic growth rate as
compared to their mesophilic counterparts. The yield of
these microorganisms per unit amount of substrate is
lower. The lower growth yield of thermophilic bacteria
could be due to their increased decay rate, which is
double that of mesophilic cultures because the cells have
a tendency to lyse quickly under thermophilic conditions
[8], and may be due to their higher energy requirement
for maintenance or the specic molecular properties of
enzyme reactions at thermophilic temperatures [9].
Class A biosolids termed by EPA are essentially
pathogen free and have no restrictions on crop choice or
human access to land application sites used for disposal.
Mesophilic anaerobic digestion cannot meet the criteria
of Class A biosolids because of insufciently high
temperatures. Thermophilic anaerobic digestion is one
of the proven methods for producing Class A biosolids.
Even though thermophilic anaerobic digestion has
proven pathogen reduction potential, US EPA (technol-
ogy-based, 40 CFR 257, 1989) excluded thermophilic
digestion from processes to further reduce pathogens
(PFRP) and US EPA (performance-based, 40 CFR 503,
1992) listed aerobic thermophilic digestion as just one of
the PFRP alternatives, still not including single-stage
anaerobic thermophilic digestion because of instabil-
ity in operation and possibility of short circuiting
characteristics of single-stage reactor [14,15].
Thermophilic anaerobic digestion often manifests
chronically higher volatile fatty acid (VFA) concentra-
tion (especially propionate) than that for mesophilic.
Three hypotheses are put forth to explain this situation:
* Sinlge-stage continuously stirred tank reactor
(CSTR) configuration: In conventional, single-stage
CSTR, anaerobic digestor, all phases (i.e. hydrolysis,
acidogenesis, and methanogenesis) take place in a
dispersed common environment which is not ideal for
all members of consortia.
* Syntrophic reaction: A consortia of up to three
groups of bacteria are involved as shown in Fig. 1.
Propionate conversion to acetate/H2 and H2 conver-
sion to methane is thermodynamically possible only
Fig. 1. Three groups of bacteria involved in propionate degradation.
at H2 concentrations between 10
4 to 10
6 atm. This
is a severe limitation and an example of exquisite
microbial ecological partnership. Diffusion of dis-
solved H2 from the microorganisms that produce it
to the microorganisms which consume it is the main
mechanism for processes on this microscale and
therefore close microbial proximity favors low H2
concentrations [1012].
* Limit of nutrient bioavailability: Essential in micro-
and macroinorganic nutrients even though at present
it may not be sufciently bioavailable to maintain
efcient enzymatic processing.
The purpose of this research is to evaluate the
comparative process stability and process efciency of
thermophilic vs. mesophilic anaerobic digestion by
evaluating the key parameters of reactor conguration,
microbial consortia proximity, and nutrient supplemen-
tation. For the sake of comparison, all experiments in
this research were conducted at 551C or 351C so that
thermophilic operation can be directly compared with
the more commonly used mesophilic temperature.
2. Materials and methods
2.1. Seed biomass, substrate, and temperature
The thermophilic seed source was taken from a
thermophilic anaerobic sludge culture, which was
enriched on calcium acetate (10,000 mg/L) and calcium
propionate (4000 mg/L) for 3 months in a 17 L batch-fed
reactor at a temperature of 551C with an HRT of 30
days. This culture was originally taken from the
Murfreesboro, TN, municipal anaerobic sludge digester
which had been operated at mesophilic temperature. The
mesophilic seed source was also taken from the same
digester. The substrate was a simulated primary sludge.
Dog food was selected for the substrate due to its
similarity to municipal primary sludge. This material
had protein, fat, and ber contents of 21%, 8%, and
5%, respectively, and a volatile solid (VS) content of
90% of total solids (TS). It also contained all the micro-
and macronutrients necessary for animal growth. The
feed slurry was prepared by blending after mixing with
 M. Kim et al. / Water Research 36 (2002) 43694385
tap water containing 6 g/L of NaHCO3. Temperatures
of 551C and 351C were selected as the thermophilic and
mesophilic temperatures, respectively.
2.2. Apparatus
Four different reactor congurations, which are a
daily batch-fed single-stage CSTR, a continuously fed
single-stage CSTR, a daily batch-fed two-phase CSTR,
and a daily batch-fed non-mixed reactor, were studied at
551C and at 351C, respectively. Table 1 tabulates a
summary of operating conditions evaluated and Fig. 2
shows the schematic of these four different reactor
congurations. MB, MC, MTP2, MNMR, TB, TC,
TTP2, and TNMR utilized 1.1 L glass bottles (Wheaton
Scientic Co.) and the operating volume was 800 mL.
MB and TB were fed once per day and MC and TC were
fed continuously. MB, MC, MTP2, TB, TC, and TTP2
were continuously mixed by a magnetic stirrer. MTP1
and TTP1 utilized 500 mL glass bottles and the
operating volume was 120 mL. MTP1, MNMR, TTP1
and TNMR were mixed manually just prior to wastage
and immediately after feeding and then allowed to settle
and remain unmixed until the next days feeding.
2.3. Analytical methods
Daily gas production in all reactors was measured
using wet-tip gas meters. pH in all reactors was
measured daily using a pH meter (Fisher Co. digital
pH meter 107). VFAs in all waste samples were
measured using a gas chromatograph (Shimadzu Model
GC-14A) equipped with a ame ionization detector and
a 1.7 m glass column packed with 0.3% GP Carbopack
C/0.3% Carbowax 20 M/0.1% H3PO4 (Supelco, Inc.).
The column temperature was 1401C and the injector/
detector temperature was 2001C. Nitrogen was the
carrier gas with a pressure of 1.75 kg/cm2. Data
integration was accomplished using a Varian 4270
Integrator. Samples were prepared by centrifuging
Table 1
Summary of reactor operating conditions
Conguration Temperature (1C)
TB: thermophilic batch-fed CSTR 55
TC: thermophilic continuously fed CSTR 55
TTP1: phase 1 reactor in thermophilic two-phase system 55
TTP2: phase 2 CSTR in thermophilic two-phase system 55
TNMR: thermophilic batch-fed non-mixed reactor 55
MB: mesophilic batch-fed CSTR 35
MC: mesophilic continuously fed CSTR 35
MTP1: phase 1 reactor in mesophilic two-phase system 35
MTP2: phase 2 CSTR in mesophilic two-phase system 35
MNMR: mesophilic batch-fed non-mixed reactor 35
4371
(Beckman Instruments, Model GP) at 4000 rpm for
30 min and ltering approximately 3 mL of the super-
natant through 0.45 mm lter paper (Micron Separations
Inc.) The ltered samples were then acidied with 10%
H3PO4 (volumetric basis) to lower the pH below 3 and
ensure that acids were unionized and able to volatilize.
VFAs were measured twice per week. TS were measured
according to section 2540 D and VS were measured
according to section 2540 E in Standard Methods,
respectively [13]. TS and VS were measured twice per
week.
3. Results and discussion
The results are discussed for three periods: (1) start-up
until steady state (Days 0200); (2) steady state with feed
solids concentration of 4% and HRT of 20 days (Days
201248); and (3) organic loading rate (OLR) increase
from 4% solids at steady state to reactor failure by
increasing the solids concentration in the feed and
decreasing the HRT (after Day 201).
3.1. Start-up (Days 0200)
Before starting operation of all reactors except the
TNMR, it was intended to compare reactor perfor-
mance at the feed solids content of 4% and HRT of 20
days. However, during the rst 4% feeding period all
reactors showed a pH drop as well as an increase of
VFA concentration. Therefore, the solids percent in the
feed was reduced to 0% or 2% and then increased
gradually up to 4% depending on the pH recovery of
each reactor. Stable reactor performance of all reactors
reached 4% of feed solids percent at different intervals
(Fig. 3), which were:
* MB at Day 178;
* TB at Day 178;
* MC at Day 63;
Starting HRT (days) Feed
20 Dog food
20 Dog food
2 Dog food
18 Efuent TTP1
30 Dog food
20 Dog food
20 Dog food
2 Dog food
18 Efuent MTP1
20 Dog food
4372 M. Kim et al. / Water Research 36 (2002) 43694385
Fig. 2. Schematic of four reactor congurations.
* TC at Day 63;
* MTP at Day 56;
* MNMR at Day 49; and
* TTP at Day 44.
TNMR started with an HRT of 30 days at the feed
solids content of 4% and the HRT was changed to 20
days at Day 120. During the period before adding
nutrients, both MB and TB were not only less stable
than other reactors but also reached 4% feed solids the
latest (Fig. 3). This result appears to indicate that, for
the start-up period:
* two-phase had an advantage over single-stage from
the favorable results of MTP2 and TTP2;
* continuous feed had an advantage over batch feed
from the favorable results of MC and TC; and
* non-mixing had an advantage over CSTR from the
favorable results of MNMR.
Measured VFA compounds and organic acids were
acetate, propionate, iso-butyrate, n-butyrate, 2-methyl-
butyrate, iso-valeric, n-valeric, and lactate. Among them
the main acids were acetate, propionate, and n-butyrate.
With respect to the comparison of temperature effect, all
reactor congurations at thermophilic temperature
except the two-phase system (i.e. all single-stage
reactors) showed higher acetate concentrations (Fig. 3).
Even though individual specic microorganisms (i.e.
acidogens and methanogens) at thermophilic tempera-
ture have a higher substrate utilization rate, these results
implied that slower growing methanogens in a single-
stage were more affected by the higher temperature,
probably due to an unbalanced reaction rate between
acidogens and methanogens. On the other hand, for the
two-phase reactor conguration, no difference between
the two temperatures was observed until supplementing
nutrients, probably due to the separation of the two
kinds of microorganisms. Decreasing the HRT of
TNMR from 30 to 20 days did not affect the reactor
performance (Figs. 3, 4 and 6).
Fig. 5 shows the results of start-up for both phase 1
reactors in the two-phase system (MTP1 and TTP1) with
respect to VFA and pH vs. time and average results
from Days 64 to 200 (i.e. HRT of 2 days and feed
solids of 4%). Both phase 1 reactors had high concen-
trations of:
 M. Kim et al. / Water Research 36 (2002) 43694385 4373

Fig. 3. Monitoring start-up of reactors for pH and acetate.

4374 M. Kim et al. / Water Research 36 (2002) 43694385

Fig. 4. Monitoring start-up of reactors for propionate and n-butyrate.

 M. Kim et al. / Water Research 36 (2002) 43694385
Fig. 5. Monitoring start-up of MTP1 and TTP1 for VFA and pH. (a) VFA and pH vs. time. (b) Average from Days 64 to 200. (Error
bar represents 2 standard deviations.)
* acetate (mesophilic: 20005000 mg/L; thermophilic:
10003000 mg/L);
* n-butyrate (mesophilic: 100010,000 mg/L; thermo-
philic: 10007000 mg/L); and
* lactate (mesophilic: 200011,000 mg/L; thermophilic:
10007000 mg/L).
But propionate was barely produced (o100 mg/L) at
both temperatures. On the other hand, the phase 2
reactors produced high levels of propionate (mesophilic:
before adding nutrients (02000 mg/L), after adding
nutrients (20006000 mg/L); thermophilic: no difference
between before and after adding nutrients (1000
2000 mg/L)). The pH of TTP1 was maintained at a
slightly higher range (between 5 and 5.5) than that of
MTP1 (between 4.5 and 5) (Fig. 5).
In order to observe the effect of inorganic nutrient
supplementations, a cocktail of Ca, Fe, Co, and Ni,
which were mostly considered as stimulating nutrients,
4375
was added to all reactors having manifested chronically
elevated VFA concentrations (i.e. all reactors except
MNMR) on Day 69.
Added inorganic nutrients consisted of CaCl2, FeCl2,
CoCl2, and NiCl2, at the concentration of 500, 10, 1,
and 1 mg/L respectively. Soon after adding nutrients
daily, the gas production of those reactors increased
signicantly followed by a slight decrease and, there-
after increased gradually until steady state (Fig. 6).
Even though the substrate (dog food) had enough
nutrients for animals, the positive results from nutri-
ent supplementation show that these reactors were
decient in bioavailability of those nutrients at
both temperatures. The feed contained more than
ample nutrients but its bioavailability was denitely
limiting probably due to precipitation reactions
in the reactor, in all probability. Therefore, the results
imply the importance of inorganic nutrients bioavail-
ability.
4376 M. Kim et al. / Water Research 36 (2002) 43694385

Fig. 6. Monitoring start-up of reactors for gas production and VS destruction (gas production: 4-day moving average).
 M. Kim et al. / Water Research 36 (2002) 43694385
Soon after supplementing inorganic nutrients, the
pH of all reactors at both temperatures increased
for a short time and then decreased below the pH
before adding nutrients. Normally increased gas produc-
tion accompanies increasing pH because methano-
genesis reduces VFA and produces alkalinity. However,
the results in this experiment show a pH reduction
with increased gas production. This result may be
explained by one hypothesis, which is that nutrients
increased the rate of acidogenesis as well as methano-
genesis. The stimulated acidogenesis produced more
VFA and H+ ions, then also stimulated methanogen-
esis, which removed more VFA, but did not produce
enough alkalinity to neutralize the H+ ion. These
phenomena occurred after the supplementation of
nutrients and continued until reactor failure. The
average pH values are compared later in this
study.
Adding inorganic nutrients signicantly removed
almost all VFA, except propionate, in all reactors except
the non-mixed reactor. Residual propionate concentra-
tion was little changed or even increased after supple-
menting inorganic nutrients. However, in the TNMR,
propionate was removed to below the detection limit at
the HRT of 30 days and even 20 days (Fig. 4). The
complete reduction of propionate soon after supple-
menting nutrients can be explained by three parameters,
which are longer HRT, inorganic nutrient supplementa-
tions, and closer microbial consortia proximity. At the
HRT of 30 days, it is hard to say which parameter or
what combination of these three parameters caused this
result. However, no increase of propionate concentra-
tion after 3 HRT of operation with the HRT of 20 days
implies that closer microbial consortia proximity
strongly favored the removal of propionate. In addition,
nil propionate concentration in the MNMR also
explains the same reason.
In order to stimulate propionate metabolism, a
number of assays were conducted, in which yeast
extract, NH4Cl, (NH4)2HPO4, Na2SO4, MgSO4, KCl,
ZnCl2, Na2WO4, Na2MoO4, Na2SeO3, MnCl2KI,
NH4VO3, AlCl3, H3BO3, and CuCl2, were assayed in
aliquots taken from reactors showing chronically
elevated propionate concentrations (i.e. MB, MTP2,
TB and TTP2). However, no stimulation in any of the
assays for the degradation of propionate was observed
at both temperatures (data not shown).
Comparing all reactor congurations at both tem-
peratures, the MNMR and TNMR appeared to show
markedly superior performance during the start-up
period. This was most probably related to closer
microbial consortia proximity, which minimizes the
distance between H2 producing and H2 consuming
microbes, removing the H2 diffusion bottleneck of the
biotransformation and thus facilitating the scavenging
of H2 to low concentrations which enhanced the
4377
thermodynamics of the propionate conversion. Propio-
nate concentrations below 50 mg/L were also the most
signicant characteristic of MNMR and TNMR. With
respect to the poor efuent quality at the thermophilic
temperature, it turned out that only close microbial
consortia proximity could solve this problem during
start-up period.
3.2. Steady state with feed solids concentration of 4% and
HRT of 20 days
For the comparison of the various reactor perfor-
mances at the steady state with a feed solids concentra-
tion of 4% and an HRT of 20 days, all data were
calculated for the average value with 1 standard
deviation from Days 200 to 248. During this period,
the MNMR was not supplemented by inorganic
nutrients because unlike the other three reactor cong-
urations, MNMR manifested no elevated VFA. The
calculated averages are tabulated in Table 2.
The comparisons of average values are shown in
Fig. 7. Fig. 7(a) shows that most VFAs consisted of
propionate in all reactors. Thermophilic single-stage
CSTRs produced much higher VFA concentrations than
their mesophilic counterparts. The two-phase reactor
conguration had much reduced performance over the
single-stage reactor type at mesophilic temperature while
two-phase had a slight benet over the single-stage
reactor type at thermophilic temperatures. On the other
hand, the thermophilic non-mixed reactor showed much
lower VFA concentrations among all thermophilic
reactors as it also did during the start-up period. This
result implies that the problem of the poor efuent
quality at thermophilic temperature can be at least
partly solved by non-mixing (i.e. closer microbial
consortia proximity).
Fig. 7(b) shows the comparisons of VS and VSS
removal percentage. TB showed higher VSS removal
ability than MB while no difference was observed for VS
removal between these two reactors. This result implies
that more solubilized compounds (i.e. VFAs) at the
thermophilic temperature were not converted to CH4 as
efciently. Therefore, the results clearly show the higher
rate of hydrolysis/acidogenesis and lower rate of
methanogenesis at the thermophilic single-stage batch-
fed CSTRs. On the other hand, TNMR showed higher
VS removal than the thermophilic single-stage CSTRs
(TB and TC), which implies that the solubilized
compounds (i.e. VFA) were converted to CH4 ef-
ciently. Therefore, this result implies that in order to
raise the rate of methanogenesis at thermophilic
temperature, closer microbial consortia proximity is
necessary.
Fig. 7(c) shows the comparison with respect to pH.
Only MNMR continued to show a higher pH range
among all reactors. However, even though this higher
4378 M. Kim et al. / Water Research 36 (2002) 43694385

Table 2
Results of reactor performance at steady state operation

Reactor pH Gas production Propionate VFA VS VSS

conguration (mL/g VS (COD mg/L) (COD mg/L) removal removal
added day) (%) (%)

MB Ave. 6.84 494 56 142 68 67

S.D. 0.10 32 25 62 1 1
MC Ave. 6.99 482 46 129 76 76
S.D. 0.04 36 21 58 1 1
MTP Ave. 6.93 531 4495 5105 71 82
S.D. 0.21 57 1184 1316 3 1
MNMR Ave. 7.55 556 56 161 74 74
S.D. 0.08 37 41 90 1 1
TB Ave. 6.76 391 2113 4307 66 76
S.D. 0.06 37 239 835 1 1
TC Ave. 6.89 461 2995 4133 71 80
S.D. 0.06 29 239 546 1 2
TTP Ave. 6.95 607 1859 2444 77 81
S.D. 0.05 37 249 747 2 1
TNMR Ave. 6.94 716 53 162 77 77
S.D. 0.11 42 34 111 1 1

Operating period: from Day 200 to Day 248.

No. of observation: pH (49), Gas production (49), Propionate (7), VFA (7), VS (6), VSS (6).
VS removal %: (VS CODin
VS CODout)/VS CODin.
VSS removal %: (VSS CODin
VSS CODout)/VSS CODin=(VSS CODin
(VS CODout
VFA CODout))/VSS CODin assuming that
most soluble compounds consisted of VFAs.

Fig. 7. Comparison of mesophilic and thermophilic reactors for (a) VFAs, (b) VS and VSS destruction, (c) pH, and (d) gas production.
(Average values during Days 200248, steady state operation with Feed solids %=4 and HRT=20 days, except no nutrients for
MNMR; error bars stand for 2 positive and negative standard deviations.)
 M. Kim et al. / Water Research 36 (2002) 43694385 4379

Fig. 8. Long-term monitoring for the failure pattern of all reactors by pH drop.

range of pH may give a good environment for
methanogens, it turned out that inorganic nutrient
supplementation was perhaps necessary at the higher
OLR because MNMR eventually failed at a lower
OLR than three other reactor congurations in spite of
the advantage of non-mixing (i.e. closer microbial
consortia proximity) (Fig. 8). Fig. 7(d) shows the
comparison with respect to the average gas production.
Among all reactors, TNMR produced the highest
unit volume of gas while TB produced the least. Among
all mesophilic reactors, no difference in gas production
was observed while a severe difference among thermo-
philic reactors was observed. These results imply that
even though thermophilic bacteria have a higher
activity, the reactor performance strongly depends on
the reactor conguration. All reactors had CH4 compo-
sitions of 5560%. No statistical difference in methane
content was observed among all reactors at both
temperatures.
3.3. Long-term operation with OLR increase
(after Day 201)
The operation of all reactors was then continued until
reactor failure (pHo5.5) in the third phase of this study.
After a feed solids concentration of 9% was achieved,
further increase in OLR was achieved by a reduction of
HRT, while maintaining the feed solids at 9%. Table 3
shows the OLR at reactor failure and possible positive
and negative reasons for the failure.
Fig. 8 shows the failure pattern of all reactors as
manifested by pH drop. As the OLR increased, all
reactors eventually failed. TB failed the rst while MC,
TC and TNMR failed the last at the highest OLR. The
4380 M. Kim et al. / Water Research 36 (2002) 43694385

Table 3
The comparison of reactor conguration for failure with OLR increase

Reactor conguration The order of failure OLR (g VS/L day) at failure Possible positive reasons Possible negative reasons

TB 1st 7.4 (4) (1), (2), (3)

MNMR 2nd 7.4 (1), (2) (3), (4)
TTP 3rd 7.4 (1), (4) (2), (3), (5)
MB 4th 10.1 (2), (4) (1), (3)
MTP 4th 10.1 (1), (2), (4) (3), (5)
MC 6th 20.3 (2), (3), (4), (5) (1)
TC 6th 20.3 (3), (4), (5) (1), (2)
TNMR 6th 20.3 (1), (4) (2), (3)

Possible positive reasons: (1) good microbial consortia proximity; (2) faster yield of methanogens due to low temperature; (3) constant
pH due to continuous feeding; (4) nutrients supplementation; (5) waste of non-acidied VS in feed.
Possible negative reasons: (1) poor microbial consortia proximity; (2) slower yield of methanogens due to high temperature;
(3) difculty of pH recovery due to batch feeding; (4) limited bioavailability of nutrients; (5) 2 days shorter HRT/STR.

continuously fed reactors had a relatively constant pH at
steady state. However, the pH of daily batch-fed
reactors dropped soon after feeding and then recovered
gradually until the next feeding. Therefore, with an OLR
increase, daily batch-fed reactors experienced wider
daily pH extremes. This is probably the reason why
the continuously fed reactors failed later than the batch-
fed reactors. The anomaly of the superior performance
of the MNMR during the start-up period and the failure
of MNMR at the lowest OLR is probably due to the
batch feeding induced pH drops and/or the lack of
inorganic nutrient supplementation because only the
MNMR was not supplemented with nutrients (Figs. 3
and 8). All other reactors were supplemented with
nutrients since Day 69. The MNMR was never
supplemented, nor appeared to need supplementation
with nutrients up to Day 350, when propionate started
to accumulate. This can also be explained by the failure
of TNMR at the highest OLR because TNMR having
the same reactor conguration as MNMR was supple-
mented by inorganic nutrients early in the study (Day
69). Therefore, inorganic nutrients appear to be more
important at high OLR.
Fig. 9 shows the long-term monitoring of all reactors
for propionate. After the start-up period, all thermo-
philic reactors except TNMR showed an increase in
propionate concentrations as the OLR increased. On the
other hand, all mesophilic reactors except MTP2 showed
little increase of propionate concentrations. All reactors
showed no other VFA concentration accumulation until
reactor failure. The propionate concentration of TNMR
did not show up until reactor failure except for a short
time (from Days 290 to 330). Therefore, these results
imply that the problem of the poor efuent quality in
thermophilic processes can be solved by non-mixing (i.e.
closer microbial consortia proximity).
Calculating average VS destruction percentage and
cumulative gas production from Day 239 (after 3 HRT
at 4% feed solids) to Day 345 (1 day before increasing
the OLR to 7.4 g VS/L day, when the rst reactor failed),
which ensures all reactors had the same conditions, no
statistical difference was observed among all reactors.
On the other hand, the continuously fed reactors showed
the lowest gas production, while the non-mixed reactors
showed the highest gas production at both temperatures
implying the importance of closer microbial consortia
proximity (Fig. 10).
3.4. A kinetic model for describing propionate
degradation by syntrophic reaction
McCarty and Smith [10] have graphically illustrated
the thermodynamics of propionate and H2 metabolism
as a function of the H2 concentrations. Propionate
conversion to acetate and H2 is thermodynamically
possible only at H2 concentrations below 10
4 atm. H2
conversion to methane is possible only at H2 concentra-
tions above 10
6 atm. Therefore, H2 concentrations in
the bulk liquid must be in the range of 10
410
6 atm.
The extent of propionate degradation is a function of
propionate concentration and can be modeled by the
Monod equation
kX Pr
dPr f1 Pr dt; 1
KS Pr
where Pr is the propionate concentration in bulk liquid
(mg/L COD), k the maximum specic propionate
removal rate (day
1), X the biomass (mg/L VSS), KS
the half-velocity substrate concentration (mg/L COD).
McCarty and Smith [10] also showed that higher
hydrogen partial pressures tend to supply more energy
to the hydrogen-consuming species, whereas lower
partial pressures result in more energy going to the
hydrogen-producing bacteria. Therefore, the extent of
propionate degradation is hypothesized to be a function
of negative free energy change. If it is assumed that the
 M. Kim et al. / Water Research 36 (2002) 43694385 4381

Fig. 9. Long-term monitoring of all reactors for propionate.

relationship between the extent of propionate degrada- If it is assumed that dPr f1 Prf2
DGPr ;
tion and negative free energy change is linear in the  
kX Pr 
reaction of propionate conversion to acetate and dPr f1 f2 dt A
1:25 logpH2
5 : 4
KS Pr
hydrogen (Pr-Ac+H2), the following equation can
be obtained: Assuming A 1 kcal
1, k 6 day
1, KS 200 mg/L,
and X 1000 mg/L, the extent of propionate degrada-
dPr f2
DGPr A
DGPr ; 2 tion can be expressed as
 
where A is the propionate removal coefcient (kcal
1). 6000 Pr 
dPr dt
1:25 logpH2
5 ; 5
From their illustrations [10],
DGPr is a function of 200 Pr
pH2 and can be expressed as
where log (pH2)o
4.

DGPr
1:25 logpH2
5; 3 Using Excel spreadsheet software, the extent of
propionate degradation is severely retarded with an
where pH2=H2 partial pressure in bulk liquid. increase of H2 partial pressure, as shown in Fig. 11.
4382 M. Kim et al. / Water Research 36 (2002) 43694385

Fig. 10. Average VS destruction and cumulative gas production (from Days 239 to 345, error bars represent 2 standard deviations).

substrate concentration (mg/L COD), where

6olog(pH2).
Assuming B 1 kcal
1, k 6 day
1, KS 0:0005
mg/L, X 1000 mg/L, the extent of H2 degradation
can be expressed as
 
6000H2 
dH2 dt 1:33 logpH2 8 : 7
0:0005 H2
Using Excel spreadsheet software, Fig. 12 shows the
extent of H2 degradation with H2 partial pressure.
At steady state (1 mol propionate (=112 g COD)
produces 3 mol hydrogen (=16 g COD)),
112
dPr dH2 2:3 dH2 ; 8
16  3
where propionate and H2 are in mg/L as COD.
Fig. 11. Effect of H2 partial pressure on the extent of The relationship between H2 partial pressure (pH2)
propionate degradation (for the interval of 1 day). and H2 molar concentration is given by Henrys law,
which at 351C is as follows: [H2](g)(atm)/[H2](aq)(mol/
Likewise, the extent of H2 degradation can be L)=1330 atm L/mol. For pH2=4  10
5 atm, the corre-
hypothesized by the Monod equation and McCarty sponding H2 solution concentration would be 3  10
5
and Smith [10]: mol/m3. Then, 3  10
5 mol/m3=3  10
5  16 g COD/
  L=4.8  10
4 mg/L as COD. Therefore, pH2=0.083H 2.
kX H2 
dH2 dt B 1:33 logpH2 8 ; 6 If the term of H2 is replaced by the term of C1
K S H2
(=corresponding H2 solution concentration of pH2 at
where B is the hydrogen removal coefcient (kcal
1), k the surface of H2 producer, pH2=0.083 C1 mg/L as
the maximum specic hydrogen removal rate (day
1), COD) in Eq. (5) and C2 (=corresponding H2 solution
H2 the H2 concentration in bulk liquid (mg/L COD), X concentration of pH2 at the surface of H2 consumer,
the biomass (mg/L VSS) and KS the half-velocity pH2=0.083 C2 mg/L as COD) in Eq. (7), from Eqs. (5)
 M. Kim et al. / Water Research 36 (2002) 43694385 4383

Fig. 12. Effect of H2 partial pressure on the extent of H2

degradation (for the interval of 1 day).

and (7)(8), the following equation can be obtained:

 
6000Pr
dtf
1:25 log0:083C1
5g
200 Pr
 
6000C2
2:3 dtf1:33 log0:083C2 8g: 9
0:0005 C2
McCarty and Smith [10] calculated the distance between Fig. 13. Effect of the distance between two syntrophic bacteria
two syntrophic bacteria as the OLR varies at steady and OLR on propionate removal.
state, which is shown in the following equation:
 
2D C1
C2 1=2
DZ ; 10
Rt C1 C2
where DZ is the distance between two syntrophic
bacteria (mm), C1 the H2 concentration at the surface
of propionate oxidizer (mg/L COD), C2 the H2
concentration at the surface of H2 consumer (mg/L
COD), Rt the turnover rate per second and D the
molecular diffusivity of H2 in water=5.85  10
5 cm2/s,
Rt can be calculated as
Rt OLR=H2 ; 11
where OLR is the organic loading rate as H2 generated
(H2 g COD/L day) and H2 the H2 concentration in bulk
liquid (H2 mg/L COD). Fig. 14. Comparison of experimental data and model predic-
At constant pH2=4  10
5 atm (corresponding H2 tion for propionate concentration.
concentration=6  10
5 mg/L COD), Rt (s
1)=193
OLR and from Eqs. (10) and (11), the following As shown in Fig. 13, propionate concentration
equation can be obtained: increases rapidly at some point of OLR, which results
C2 OLR DZ2 61 in reactor failure, and the OLR at this point increases as
C1 : 12 DZ decreases. If DZ of MB is 2 mm and DZ of TNMR is
61
OLR DZ2
1.4 mm, Fig. 14 shows that this model reasonably
From Eqs. (9) and (12), the following equation can be
matches the experimental data.
obtained:

200C2 f1:33 log0:083C2 8g

Pr     : 13
C2 OLR DZ 2 61
0:43
1:25 log 0:083
5 0:0005 C 2
C 2 f 1:33 log0:083C2 8g
61
OLR DZ2

When C2 0:0001 mg/L, using Excel spreadsheet Fig. 15 also shows a signicant effect of the distance
software, Fig. 13 shows the propionate concentration, of two syntrophic microorganisms on the extent of
which is a function of OLR and DZ: propionate and H2 degradation at steady state. From
4384 M. Kim et al. / Water Research 36 (2002) 43694385
Fig. 15. Effect of the distance of syntrophic microorganisms on the extent of propionate and H2 degradation.
Eqs. (5), (7) and (12), the extent of propionate and H2
degradation vs. H2 partial pressure based on the H2
concentration on the surface of the H2 producer can be
plotted. At steady state, the extent of propionate and H2
degradation should be equal, which is shown by three
intersections in Fig. 15. As DZ increases, the extent of
degradation decreases, which indicates the important
role of close microbial consortia proximity.
4. Conclusions
The overall conclusions from the results presented in
this study can be summarized as:
* During the start-up period, stable pH, lower VFA
concentrations, higher gas production, and stable VS
removal of non-mixed reactors at both temperatures
clearly showed efcient performance between acid
producers and consumers. This implies the impor-
tance of close microbial consortia proximity to
enhance microbial scavenging of dissolved H2 con-
centrations to lower levels, improving the thermo-
dynamic advantage of the process at both
temperatures. Therefore, thermophilic disadvantages
can be solved by the efcient use of close microbial
consortia proximity.
* Supplementing inorganic nutrients signicantly re-
duced all VFA concentrations dramatically, except
for propionate. On the other hand, in the case of the
non-mixed reactor conguration, propionate was
also considerably reduced. Therefore, it turned out
that nutrients bioavailability played an important
role in anaerobic digestion at both temperatures.
* Propionate was efciently reduced by the non-mixed
reactor conguration which enhanced close microbial
consortia proximity.
* The comparative results of the reactor perfor-
mance at steady state clearly showed the superior
performance of TNMR with respect to lower
VFA, higher gas production and VS removal
conrming that microbial consortia proximity can
solve the poor efuent quality problem of thermo-
philic.
* During the long-term operation with OLR increases,
the continuously fed reactors failed at the highest
OLR due to the steady pH in the reactor. The
thermophilic non-mixed reactor also failed at the
highest OLR showing much better performance than
the continuously fed reactors until reactor failure. On
the other hand, the mesophilic non-mixed reactor,
which was never supplemented by inorganic nutri-
ents, failed earlier than the continuously fed reactors
even though it showed much better performance than
the continuously fed reactors prior to reactor failure.
This result conrms the important role of both close
microbial consortia proximity and nutrients supple-
mentation.
* Two-phase digestion at both temperatures did not
show any signicant benet. However, thermophilic
two-phase showed slightly better performance than
mesophilic during both the start-up and the long-
term periods.
* The results for the same conditions during OLR
increase also clearly showed that the non-mixed
reactor conguration at both temperatures removed
VFAs more efciently and produced more gas than
other reactor congurations conrming efcient
conversion of solubilized compounds to CH4 with
propionate below detection concentration which is
quite signicant.
* The proposed model for the effect of the distance
between two syntrophic bacteria reasonably matched
the experimental data in this study.
 M. Kim et al. / Water Research 36 (2002) 43694385
References
[1] Buhr HO, Andrews JF. Review paper, the thermo-
philic anaerobic digestion process. Water Res 1977;11:
12943.
[2] Krugel S, Nemeth L, Peddle C. Extended thermophilic
anaerobic digestion for producing Class A biosolids at the
Greater Vancouver Regional Distreicts Annacis Island
wastewater treatment plant. Water Sci Technol 1998;38(8
9):40916.
[3] Rimkus RR, Ryan JM, Cook EJ. Full-scale thermophilic
digestion at the West-Southwest Sewage Treatment
Works, Chicago, Illinois. J Water Pollut Control Fed
1982;54(11):144757.
[4] Kugelman IJ, Guida VG. Comparative evaluation of
mesophilic and thermophilic anaerobic digestion. EPA/
600/S2-89/001, August 1989.
[5] Harris WL, Dague RR. Comparative performance of
anaerobic lters at mesophilic and thermophilic tempera-
tures. Water Environ Res 1993;65(6):76471.
[6] Wiegant WM, Hennink M, Lettinga G. Separation of the
propionate degradation to improve the efciency of
thermophilic anaerobic treatment of acidied wastewater.
Water Res 1986;20(4):51724.
[7] Zinder SH, Anguish T, Cardwell SC. Effects of tem-
perature on methanogenesis in a thermophilic (581C)
4385
anaerobic digester. Appl Environ Microbiol 1984;47(4):
80813.
[8] Speece RE. Anaerobic biotechnology for industrial waste-
waters. Nashville, TN: Archae Press, 1996.
[9] Zeikus JG. Thermophilic bacteria: ecology, physio-
logy, and technology. Enzyme Microb Technol 1979;1:
24351.
[10] McCarty PL, Smith DP. Anaerobic wastewater treatment.
Environ Sci Technol 1986;20:12006.
[11] Schmidt JE, Ahring BK. Effects of hydrogen and formate
on the degradation of propionate and butyrate in
thermophilic granules from an UASB reactor. Appl
Microbiol 1993;59:254651.
[12] van Lier JB, Boersma F, Debets M, Lettinga G. High rate
thermophilic anaerobic wastewater treatment in compart-
mentalized upow reactor. Water Sci Technol
1994;30(12):25161.
[13] APHA. Standard methods for the analysis of water and
wastewater, 18th ed. AWWA, WPCF, 1992.
[14] US EPA. Control of pathogens in municipal wastewater
sludge for land application under 40 CFR Part 257. EPA/
625/10-81/006, Washington, DC, 1989.
[15] US EPA. Environmental regulations and technology:
control of pathogens and vector attraction in sewage
sludge under 40 CFR Part 503. EPA/625/R-92/013,
Washington, DC, 1992.