Bioresource Technology 97 (2006) 1105–1118

Review Paper

The effects of operational and environmental variations

on anaerobic wastewater treatment systems: A review
Renato Carrhá Leitão a,*, Adrianus Cornelius van Haandel b,1
,
Grietje Zeeman c,2, Gatze Lettinga c,2
a
Embrapa Agroindustria Tropical (Brazilian Agricultural Research Corporation, Institute of Tropical Agroindustry),
Rua Coronel Juca, 510/902, Fortaleza, CE 60170-320, Brazil
b
Federal University of Campina Grande, Av. Aprı́gio Veloso, 882, Bodocongó, Campina Grande, PB 58.109-970, Brazil
c
Wageningen University, Sub-Department of Environmental Technology, P.O. Box 8129, 6700 EV, Wageningen, The Netherlands

Received 30 January 2004; received in revised form 12 November 2004; accepted 8 December 2004
Available online 24 February 2005

Abstract

With the aim of improving knowledge about the stability and reliability of anaerobic wastewater treatment systems, several
researchers have studied the effects of operational or environmental variations on the performance of such reactors. In general,
anaerobic reactors are affected by changes in external factors, but the severity of the effect is dependent upon the type, magnitude,
duration and frequency of the imposed changes. The typical responses include a decrease in performance, accumulation of volatile
fatty acids, drop in pH and alkalinity, change in biogas production and composition, and sludge washout. This review summarises
the causes, types and effects of operational and environmental variation on anaerobic wastewater treatment systems. However, there
still remain some unclear technical and scientific aspects that are necessary for the improvement of the stability and reliability of
anaerobic processes.
 2005 Elsevier Ltd. All rights reserved.

Keywords: Anaerobic reactors; Anaerobic wastewater treatment system; Operational variations; Environmental variations; Steady state conditions;
Shock loads; Transient conditions

1. Introduction treatment, combinations of different methods can be

used, viz. physical, chemical, and biological. The last
Human societies produce wastes that can represent a method can be divided into two classes (aerobic and
useful raw material for the production of energy, and anaerobic) which constitute the main units of most
the recovery of by-products and component water. Sev- wastewater treatment plants.
eral techniques are already available to attain the goals Anaerobic processes generate energy in the form of
of ‘‘Environmental Protection and Resource Conserva- biogas, and produce sludge in an amount that is signifi-
tion’’ (Lettinga et al., 2000). In the case of wastewater cantly lower than that resulting from aerobic systems. In
contrast, aerobic processes, which are widely used for
the treatment of wastewater, have at least two distinct
*
Corresponding author. Tel.: +55 85 32671352; fax: +55 85 disadvantages: their relatively high energy requirement
32889774. and high excess sludge production, which requires han-
E-mail addresses: renatocl@yahoo.com (R.C. Leitão), prosab@ dling, treatment and disposal. Surprisingly, there exists a
uol.com.br (A.C. van Haandel), grietje.zeeman@wur.nl (G. Zeeman),
gatze.lettinga@wur.nl (G. Lettinga).
certain reluctance to use anaerobic processes, particu-
1
Tel./fax: +55 83 3314809. larly in centralised treatment plants used for treat-
2
Tel.: +31 317 484241; fax: +31 317 484802. ment of municipal wastewater. Reasons against the

0960-8524/$ - see front matter  2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2004.12.007
1106 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

implementation of anaerobic processes that have been flow rate variations over the year (Castillo et al.,
provided by some established wastewater treatment 1997); (6) Operational procedures at the treatment
companies focus on three main points: (1) anaerobic plant can result in increased hydraulic and organic
reactors spread unpleasant odour; (2) anaerobic reactors loads, viz. when it is necessary to stop one of several
are unstable; and (3) high performance reactors such as anaerobic units for maintenance, the others have to cope
Upflow Anaerobic Sludge Blanket (UASB) reactors can- with the entire flow rate; (7) Several types of distur-
not cope with high load rate variations. In order to con- bances can manifest in case of industrial wastewa-
test the first two arguments, the example of Curitiba (a ter, even under normal operational conditions, given
city in southern Brazil) is presented here (Aisse et al., that the flow rate and waste concentration vary
2000). Several anaerobic reactors have been put into with the industrial processes routine (Puñal and Lema,
operation in densely populated areas of Curitiba. They 1999).
have performed satisfactorily for many years, i.e. they
perform without any problems of odours or instability, 2.1. Sewage flow and composition variations over
even in the relatively adverse climatic conditions of the day
Curitiba, with its air temperatures varying annually
from 5 to 25 C. However, it seems that there is still Domestic wastewater generally shows high flow vari-
not enough explicit experimental information available ations due to the number of inhabitants and dwellings
to confirm or rebut the third point, i.e. that high rate connected to the sewer system, specific characteristics
anaerobic reactors would be very sensitive to variations of the sewerage (type, material, length, maintenance,
in flow and/or concentration. infiltration, use of pump stations), as well as climate,
The overall objective of this work is to present a liter- topography and commercial/industrial contributions.
ature review on the types and impacts of several opera- The traditional approach for estimating the quantity
tional and environmental variations on the performance of wastewater in a separate sewer system assumes a daily
of anaerobic wastewater treatment (AnWT) systems. In flow per capita to give an average dry weather flow, and
the cited scientific papers, it was observed that each uses multipliers for estimating the peaks and low flows
quoted researcher was eager to study the behaviour of (Campos and von Sperling, 1996; Butler et al., 1995;
a specific reactor. Each researcher also submitted the Metcalf & Eddy, Inc. , 1991). The flow in such sewers
reactors to a strict range of variations in some opera- shows characteristic patterns on annual, seasonal, daily,
tional parameters, generating a dispersed set of informa- hourly and sub-hourly time scales.
tion on the subject. To facilitate a thorough In general, the flow of the wastewater in a separate
understanding of this review, this paper was arranged sewer system follows a pattern that can be simplified
by firstly presenting the causes and effects of the opera- using the equations presented by van Haandel and Lett-
tional and environmental variations on the AnWT sys- inga (1994). However, according to Campos and von
tems. Next, the main conclusions and the research Sperling (1996), these kinds of simplifications could lead
needs are presented. to misestimates. They presented an analysis of data re-
lated to the domestic wastewater characteristics from
several areas situated in the large Brazilian city, Belo
2. Causes of environmental variations Horizonte. They found higher flow-rate variations than
those usually used for the design of treatment plants,
In practice, the treatment system can become exposed and the extent of the variations was higher in the wealth-
to many variations, i.e. (1) In the case of sewage, the ier areas of the city. They developed models to predict
cyclical nature of human activities leads to a variable basic wastewater characteristics (water consumption,
sewage production over the day (Metcalf & Eddy, wastewater production, BOD load and concentration)
Inc., 1991); (2) Separate wastewater sewerage may have based on simple socio-economic variables. According
inappropriate connections of runoff water and rainfall to the results of their studies, the main wastewater char-
contributions, resulting in overloads in networks as well acteristics vary with the hour of the day and the day of
as in treatment plants (Dauphin et al., 1998); (3) Com- the week, in addition to depending on total family
bined sewer networks exhibit the first-flux phenomena, income.
in cases where storm water contributions result in an in- Several researchers have studied the variation of
creased suspended solids (SS) and chemical oxygen de- flows and changes in the sewage composition which oc-
mand (COD) concentration in the first minutes of the cur in sewer systems and wastewater treatment plants.
event (Deletic, 1998); (4) The sewage network often Heip et al. (1997) developed a mathematical model to
has one or more pumping stations, which convey the simulate the hydraulic behaviour of a sewer network.
sewage intermittently at a much larger flow than average The calibration was done in a city in the north of Bel-
(Dauphin et al., 1998); (5) Tourist areas dramatically in- gium utilizing data from a 9-month monitoring cam-
crease their populations during holidays, leading to high paign. Data from a monitoring station in a small
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1107

Danish town, which produced 4–5 L/s of wastewater, sented an example of a coastal tourist city in Turkey
showed diurnal variations between zero and approxi- (Bodrum), where the population increase can be as high
mately 10 L/s during dry weather (Schaarup-Jensen as 1500% during the high season.
et al., 1998). Oliva (1997) characterised the sewage of
São Carlos—Brazil (separate sewer system) on the 2.3. First-flush
basis of several parameters, throughout several hours
of the day, and different days of the week. Their results The concept of ‘‘the first-flush of storm runoff’’ is
show variations of 103% between minimum and maxi- based on the assumption that the first part of runoff is
mum COD during the week, and 117% between the most polluted. Deletic (1998) investigated whether
minimum and maximum COD throughout the day. or not this phenomenon really exists and what its char-
Higher variations were found in proteins, carbohy- acteristics are. To characterise the ‘‘first-flush’’,
drates and lipids contents, 260%, 1003%, and 650% researchers usually use curves of the cumulative fraction
respectively over the week, and 171%, 302%, and of the total pollutant mass versus the fraction of total
150% over the day. cumulative runoff volume. The phenomenon is per-
In many cases, the main source of inflow into sewer ceived if a mass cumulative curve of a pollutant is above
networks (separate system) is comprised of domestic the runoff volume curve (Bertrand-Krajewski et al.,
wastewater. Sewage however, also generally includes 1998). According to a French study, a strong first-flush
commercial and industrial wastewaters. The domestic very rarely occurs, but in more than 65% of the events
fraction of sewage is made up of contributions from var- studied, this phenomenon manifested to some extent
ious household appliances, influencing flow quantity and (Saget et al., 1996). In fact, Deletic (1998) did not find
quality, and thus the hydraulic and organic load at the a strong first-flush of suspended solids, but it manifested
end-of-pipe WWTP. The household discharges are in 30% of the events studied and particularly during
mainly derived from the WC, kitchen sink, washbasin, large storm events. Such a first-flush could not be as-
bath, shower and washing machine. Butler et al. (1995) sessed for pH or for temperature. The researcher also
and Friedler and Butler (1996) examined the quality of suggested that pollutant transformations and transport
wastewater generated by each domestic device of 28 processes might cause the first-flush.
households in South East England, and 51 homes in Independent of the question of whether first-flush
Malta. Generally, the WC was the most significant events really exist, wastewater treatment plants none-
wastewater-producing appliance, contributing around theless are presently not designed for accommodating
40% of the daily average flow of the household total dis- these types of overcharges. This implies that this extra
charge, and 40% of the total pollution load. By monitor- amount of incoming wastewater is sent through the
ing the wastewater of several houses, they were able to equalization unit of primary clarifiers, or is by-passed
analyse the daily variation of BOD (38%), orthophos- directly to water bodies (Carstensen et al., 1998).
phate (52%), ammonia (136%) and nitrate (50%), for According to the observations of Bechmann et al.
the case of South East England. Although the authors (1998), during dry weather there is a deposition of pol-
overemphasise the variations, it seems that they are not lutants in the sewer that are flushed out with the first
so significant when wave attenuations that occur into rain contribution. This idea is supported by Bertrand-
the sewer are taken into account. Krajewski et al. (1995), who found that there was an
increase of mineral and settleable solids due the catch-
2.2. Tourist areas mentÕs surface wash off during storm events. However,
the COD concentration in the influent of the treatment
An important example of the prevalence of consider- plant remained unchanged compared to that of the dry
able changes in flow-rate and organic loading rate weather periods. This means that there is a chance for
(OLR) can be found in tourist areas, where the popula- overloading (over a short period), but this conflicts
tion increases dramatically during the high season. Cas- with the view of Deletic (1998). There are actually a
tillo et al. (1997) studied the feasibility of a combined number of variables related to the first-flush, primarily
anaerobic–aerobic system (UASB + RBC) for treating based on the characteristics of the catchments and the
the wastewater of a small tourist village near Barcelona. sewer network. In steep catchments, sediments are typ-
Experiments were carried out in the summer (tourist sea- ically absent and no erosion can take place. Sediments
son) and in the winter, examining very different loads are present in a flat catchment area, but erosion is lim-
and temperatures (19 C and 12 C). Results showed ited due to low velocities. In large catchment areas, the
that removal efficiencies were similar in both seasons, first-flush potential is reduced as a result of the long
because the higher ambient temperature during summer transportation time and wave attenuation phenome-
time balanced the higher loads during that season. At non. Therefore, for catchments of steepness and areas
high temperatures, the reactor can cope with the im- of a medium size, a distinct first-flush will probably oc-
posed higher loading rates. Orhon et al. (1999) also pre- cur (Krebs et al., 1999).
1108 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

3. Effects of environmental variations on AnWT systems 1982). Another typical effect during a situation of stress
is the drastic change in biogas production rates and
Wastewater treatment plants (WWTP) are subject to compositions (Chua et al., 1997).
variations in one or more parameters that affect or de- Although some researchers, such as Inanc et al.
fine the reactor performance, viz. flow rate, influent type (1999), support the idea that a high volatile fatty acid
and concentration, sludge retention time (SRT), nutrient concentration is detrimental for the activity of methano-
availability, temperature, pH, presence of xenobiotics, gens, other authors (Cohen et al., 1982) affirm that the
as well as others. Some of these variations can be pre- effects of high VFA concentrations are better regarded
dicted and controlled, and the reactor can be designed as the result of an imbalance, rather than the cause of
to accommodate them. But this is not the case for all reactor destabilisation. In fact, during organic shocks
variations, and the reactorÕs performance can deterio- with VFA and glucose in one-phase and two-phase
rate due to extreme transient conditions. anaerobic reactors, Cohen et al. (1982) did not observe
inhibition due to a toxic action of VFA under conditions
3.1. Effect of hydraulic and organic load variations of well-controlled pH. The substantial accumulation of
propionate observed by Cohen et al. (1982) and many
In anaerobic digestion, a delicate balance exists be- other researchers suggests a saturation of the hydrogen
tween the primary processes (hydrolysis and acidogene- transfer reactions and, as a consequence, an enhanced
sis) and the conversion of the acid products by disposal of electrons via an alternative route. From this
acetogenic and methanogenic bacteria into methane information, Cohen et al. (1982) presumed that carbon
and carbon dioxide (Cohen et al., 1982). According to dioxide reduction was the rate-limiting step, rather than
many specialists working in this field, e.g. van Lier the splitting of acetate. They never found a significant
et al. (2001), strong variations in flow and concentration acetate accumulation, or lag-period for the degradation
may adversely affect the efficiency of an anaerobic reac- of VFA, in any of their experiments with organic shock
tor. The effect of fluctuations in hydraulic and organic loads. However, the authors were not clear about the
load generally depends on the applied hydraulic reten- sludge and design of the reactor they used, and inhibi-
tion time (HRT), sludge retention time (SRT), intensity tion due to VFA would depend on the conditions under
and duration of the variations, sludge properties and the which they conducted their experiments, e.g. the type of
reactor design, particularly the design of the three phase sludge used, OLR and sludge loading rate, etc.
separator. Thus far, a clear relationship between the Borja and Banks (1995) tested the effect of shock
mentioned parameters and the performance behaviour loads on the performance of a fluidised bed reactor
of UASB reactors operating under environmental varia- (FBR) by increasing the flow rate by 100 and 150%
tions has not been fully established. for 6 and 12-h periods, utilizing the same influent con-
The accumulation of volatile fatty acids (VFA) can centration. The reactor was fed with synthetic ice-cream
be a typical reactor response during overloading, and wastewater, and was operated at an HRT of 8 h and
during sudden variations in hydraulic and organic load- OLR of 15.6 kg COD/m3 day. During the shock, they
ing rates. Hydrogen partial pressure plays an important reported decreases in pH (from 7.1 to 6.6) and alkalin-
role in controlling the proportion of the various interme- ity, and increases in effluent VFA and COD. The gas
diate products of the anaerobic reactions. Under stress- production thus increased, but the methane content de-
ful conditions, there may be a shift in the metabolic creased. The change in the CH4/CO2 ratio was a direct
pathway to a less favourable route, resulting in a ratio consequence of an inhibition of methanogenesis, and
shift between VFA producers (acidogens and acetogens of the decreased solubility of CO2 at low pH values
population) and consumers (methanogens, sulphate (Fongastitkul et al., 1994; Eng et al., 1986). In the same
reducing bacteria—SRB, and nitrogen reducing bacte- investigation, Borja and Banks (1995) also tested an in-
ria—NRB). Such a highly undesirable situation could crease of the influent COD by 100 and 150% for 6 and
lead to the production of significant amounts of carbon 12 h. The effects were essentially the same as found in
dioxide and hydrogen gas in the biogas. The partial the experiments with hydraulic load variations, but less
pressure of hydrogen gas inside the reactor might in- pronounced. When they increased the concentration of
crease to values exceeding 10 4 atm, which may then the influent, they also increased the buffering capacity
cause a shift in the metabolic pathway. When slowly so that the pH remained well controlled. The most se-
growing methanogens cannot sufficiently and rapidly vere shock imposed to the system was conducted by
eliminate all H2 produced by the H2-producing bacteria increasing the influent COD by 150% for 12 h, which
(e.g. in case the sludge contains insufficient hydrogen- caused a 180% increase in effluent COD. However, the
consuming organisms), this may result in a distinct reactor recovered its normal performance within 11 h
inhibition of the degradation of propionate, butyrate after the shock ceased. The authors did not supply en-
and lactate (Fongastitkul et al., 1994; Cánovas-Dı́az ough information about the sludge for a good evalua-
and Howell, 1988; Eng et al., 1986; Cohen et al., tion about overloading, but it seems that the reactor
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1109

could cope well with the imposed shocks. This kind of liquid suffers a shift in velocity, there is an average in-
reactor actually performs well at higher OLRs (Holst crease in internal mass transfer coefficient.
et al., 1997; Borja et al., 1995), which means that the Chua et al. (1997) investigated the response of an
shocks were not so severe and they were probably assim- anaerobic fixed-film reactor (AFFR) to hydraulic shock
ilated by the reactor ‘‘buffer’’ load capacity. loads. The reactor was operated with synthetic dairy
Bhatia et al. (1985a,b) investigated the response of a wastewater—having an influent COD of 3 g/L and an
step change in concentration and flow rate in a 9.8 L HRT of 5 days. Shocks with duration of 7 days were im-
UASB reactor, using a synthetic wastewater composed posed by reducing the HRT from 5.0 days to 2.5, 1.25,
of acetic, propionic and butyric acids. The changes were 1.0 and 0.5 days, with a concomitant decreasing of the
accomplished by varying the concentrations of each acid influent concentration. They suggested that the ability
separately from 600 to 900 mg/L for 12 h. The authors of this type of reactor to cope with the imposed shocks
carried out other step-change-like tests by increasing was due to the fixed biofilm, which was not washed out
the flow rate from 1.0 to 11.8 L/h until the reactor during critical hydraulic shock loadings. Even so, the
achieved a new steady state. They concluded that the AFFR took 4 days to recover from hydraulic shocks
reactors took approximately one residence time to re- of only twice the flow rate, and the authors did not men-
cover from the imposed changes in loading rates, which tion why methanogenic bacteria were inhibited during
means that delays decreased when the flow rate in- the shocks, combined with VFA accumulation in the
creased. This behaviour could have been due to mass reactor. It is likely that the dilutions necessary to main-
transfer resistance, as the diffusion rate increases after tain the same organic load at an increased flow rate
a hydraulic shock load (Brito and Melo, 1999). Another affected the methanogenic activity.
effect observed by Bhatia et al. (1985a,b) was the exis- Eng et al. (1986) described experiments in an UASB
tence of different levels of effluent concentration (at reactor (12.7 L) fed with diluted leachate liquor
‘‘steady state’’) for the same operational condition. This (2.2 g COD/L) and operated with HRT of around
phenomenon (hysteresis) was perceived when the 6 h. They were interested in testing the capacity of an
researchers increased the flow rate from 1.0 to 11.8 anaerobic digester to cope with severe shock loads such
L/h, and then decreased it back to 1.0 L/h. The effluent as those caused by accidental spillage of sugar syrup in a
concentration for the same flow rate was lower after the sugar beet processing industry. Shock loads with sucrose
flow was increased. The authors attributed the hysteretic at different concentrations (10, 12 and 50 g/L) and differ-
effect to the structure of the cultures inside the floccu- ent durations (3 and 8 days) were applied. In all experi-
lated biomass, which can change depending upon the ments, the reactor-pH dropped to around 4.7, causing
operational condition. However, it is likely that the KS the inhibition of methanogenesis. The researchers con-
(Monod half saturation constant) increased when the cluded that the UASB system is potentially vulnerable
OLR was raised, and did not return to its previous value to shock loads, but methanogenesis resumed a few days
when the OLR was decreased. after the shock load ceased, as long as alkalinity was
Several researchers have suggested that the diffusion provided. However, the authors were not clear about
rate of substrate through a biofilm is a function of liquid the use of leachate to simulate the wastewater of sugar
velocity and substrate concentration. Under steady-state beet processing industries, and did not mention the
conditions, the diffusion rate increases with a higher sub- shock due to a distinct carbon source—the reactor was
strate concentration (FickÕs law) and decreases with a adapted to diluted leachate (VFA rich influent) and
higher flow velocity (Beyenal and Lewandowski, 2000). the shocks were imposed with sucrose. It seems that
According to these authors, the substrate concentration the UASB reactor was more robust than the authors af-
has a stronger effect on the diffusion rate than the flow firmed, as it was expected that the reactors would com-
velocity. With regards to the effect of flow velocity on pletely fail when operated under such a strong shock
the diffusion rate, there is a contradiction between Bey- load, i.e. an increase of 5 times the ‘‘steady state’’ influ-
enal and Lewandowski (2000) and Zaiat et al. (1996), ent concentration over a 3-day period, using a different
who found that the external mass transfer resistance carbon source in the influent.
can be decreased by increasing the flow velocity. Brito Oliva (1997) conducted experiments in an 18 m3
and Melo (1999) investigated the effect of a transient UASB reactor treating sewage from the city of São
shift in the external bulk liquid velocity on the internal Carlos, located in the southeast of Brazil. They exposed
mass transfer coefficient in biofilms. They found that the system to several shocks by increasing the flow rate
at a constant and laminar flow, the diffusion rates did (Q) by 50% twice a day, and 100% once a day. Accord-
not depend upon the flow rate. However, under condi- ing to the results they presented (not shown in this re-
tions of turbulent liquid flow and thus higher shear view), it appears that the reactor tended to become
stresses, the flow velocity had a pronounced effect on better adapted after each imposed shock. This observa-
the biofilm thickness and compactness, leading to differ- tion can possibly be attributed to the washout of the fine
ent mass transfer coefficients. Moreover, if the bulk sludge ingredients, or less dense particles, during the first
1110 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

shocks imposed. During the 2 · Q shock, the effluent perature was kept around 35 C. They imposed a step-
COD increased until the load was returned to its previ- change on the organic load by increasing fat concentra-
ous value, but the 1.5 · Q shock did not seriously affect tion from 2.7 to 4.8 g/L (increase of 78%), and from 2.7
reactor performance. to 6.1 g/L (increase of 126%). The increase of 78% in the
Castillo et al. (1997) investigated the effect of differ- fat concentration did not produce any remarkable effects
ent HRTs on a pilot-scale UASB reactor (750 L), fed on the performance of the reactor, and the increase of
with domestic wastewater (CODInf 600 mg/L), under 126% in fat concentration only caused a small decrease
winter (13 C) and summer (20 C) conditions. of the COD removal efficiency from 93% to 85%. The
Their results showed that the removal efficiencies of researchers also tested a step-change of organic load
total, soluble fraction and suspended fraction of by decreasing the HRT from 12 to 6 h, and found an
COD increased as the HRT increased, but there was immediate drop in the efficiency (from 93% to 69%).
a tendency for these to become constant at an HRT From their results, it can be concluded that the stabilisa-
higher than 6 h. After every change in OLR (imposed tion period (‘‘feedless’’ period) has an important influ-
by increasing the HRT), the reactors passed through ence on the stability of the reactor during transient
a transient condition before achieving the new ‘‘steady conditions.
state’’. It is worthwhile to note that anaerobic reactors Another interesting effect of hydraulic and organic
operated at lower temperatures were more sensitive to shocks is the rapid decrease in the number and length
organic variations. of free filamentous microorganisms at increased loading
Kalyuzhnyi et al. (1996) investigated the performance conditions, which in anaerobic digesters can represent
of a lab-scale UASB reactor fed with synthetic wastewa- several different species of acidifiers, acetogens, and
ter composed of sucrose (2 g/L), potassium acetate methanogens. The decrease in number of such organ-
(ranging from 2 to 5 g/L as acetic acid) and mineral isms suggests washout or disaggregation of these bacte-
medium. The reactor was operated under organic load- ria during a shock load (Alves et al., 2000). Cheng
ing rates ranging from 3.4 to 44.9 g COD/L day, at (1992) studied the morphology of attached biofilm bac-
HRTs ranging from 4.0 to 22.5 h, at a temperature teria as a function of organic loading, VFA concentra-
around 36 C, and was fed with influent concentrations tion and biogas production in a fluidised bed reactor.
ranging from 3.2 to 7.5 g COD/L. The reactor was inoc- He found that the filamentous bacteria were predomi-
ulated with suspended sludge withdrawn from another nant at lower organic loading rate (OLR), while the
reactor treating milk industry wastewater. They found number of rods and cocci increased at higher OLR and
that increasing the OLR up to 44.9 g COD/L day, at VFA concentrations.
HRT of 4 h, caused an almost complete disappearance Leitão (2004) evaluated the robustness and stability
of lighter granules in the reactor. Furthermore, they of UASB on the basis of four indicators: (i) COD re-
found that at a lower OLR, the granules consisted moval efficiency, (ii) effluent variability, (iii) pH stability,
mainly of filaments of Methanothrix cells, whereas a and (iv) recovery time. The experimental investigation
change in the population occurred at a higher OLR, was carried out using a set of seven pilot-scale UASB
showing the significant presence of Methanosarcina, as reactors fed with domestic sewage and operated under
well as other rods and cocci. During the transient condi- different operational conditions, i.e. four reactors were
tion (the period between the time when the shock started fed at a constant flow of 20 L/h (HRT = 6 h) and with
and when the reactor achieved a new ‘‘quasi-steady- different CODInf (ranging from 200 to 800 mg/L), while
state’’), the reactorÕs performance deteriorated, resulting the other three reactors were operated with approxi-
in sludge flotation, destruction of granules, and accumu- mately the same CODInf (800 mg/L), but with different
lation of VFA. The time for recovering the ‘‘quasi-stea- HRTs (ranging from 2 to 6 h). After establishment of a
dy-state’’ conditions ranged from 4 to 22.5 HRTs, and it ‘‘steady state’’, organic shock loads were imposed by
seemed that there was a trend of decreasing recovery increasing the influent concentration to approximately
time as HRT decreased, except when the reactor was five times higher than before, over a 6-h period. Next,
highly overloaded. They did not explain why this phe- hydraulic shock loads were imposed by increasing the
nomenon occurred, but perhaps the exponential increase flow rate to three times higher than before, also over a
of the biomass overcame the linear increase of the OLR. 6-h period. The results showed that UASB reactors are
Thus every single step change in loading rate was robust systems with regards to COD removal efficiency
accompanied by a relatively higher biomass concentra- and pH stability when exposed to shock loads. How-
tion, which decreased the recovery time. ever, this reactor cannot attenuate the imposed fluctua-
Nadais et al. (2000) investigated the treatment of tion in the influent COD. It seems that either a
dairy wastewater by lab-scale UASB reactors operated secondary treatment unit is needed to retain the expelled
in an intermittent mode, viz. 48 h of feeding followed sludge occurring as a result of a hydraulic shock load, or
by 48 h without feeding. The reactors were inoculated prior to the shock, a sufficient amount of sludge needs to
with flocculent sludge previously adapted and the tem- be discharged from the reactor.
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1111

3.2. Effects of variations of the reactor temperature was some methane production during the first hours of
the tests at 65–70 C. However, this production slowed
The effect of temperature shocks on reactor perfor- down and/or stopped 30 h later. No significant methane
mance depends upon factors such as the exposed tem- production was found at 35 C until the end of the test,
perature, duration of shock, sludge characteristics and 70 h later.
imposed specific sludge load. At temperatures exceeding Omil et al. (1997) investigated the competition be-
that of the maximum growth, the decay rate will gener- tween sulphate-reducing bacteria (SRB) and methane-
ally exceed the bacterial growth rate, and consequently a producing bacteria (MPB) at two different pH levels,
decrease in specific sludge activity and reactor efficiency and exposed the system to short-term temperature de-
may occur (van Lier et al., 1990). crease. They concluded that temperature shocks of 15
When an anaerobic digester is operated under steady- and 12 C for 3 days in an UASB reactor operated at
state conditions, the activities of different metabolic 30 C and pH 7.75–8.0 caused only an insignificant de-
groups are in balance, and consequently there is no accu- crease in COD removal efficiency. The short-term low
mulation of metabolic intermediate products in the reac- temperature changes had no effect on the competition.
tor. However, when the process is exposed to a sudden El-Mashad (2003) performed experiments to assess
temperature change, the digestion process conditions the effect of temperature fluctuation on the thermophilic
can become unbalanced due to the different response of anaerobic digestion of cow manure. The author used a
the various metabolic groups of microorganisms (Cha CSTR system operated with different HRTs (10 and
and Noike, 1997). 20 days), and with different temperatures (50 and
According to Borja and Banks (1995), a shock change 60 C). The fluctuations were imposed by decreasing
in temperature may be characterised by an immediate the temperature by 10 C for a period of 10 h, and
pH drop in the reactor, which then would stabilise at increasing the temperature by 10 C for a period of
a value slightly below the previous steady state pH 5 h. The results showed that temperature fluctuation sig-
value. This drop in pH is due to an increase of the mixed nificantly affects the pH and free ammonia concentra-
liquor (effluent) VFA-concentration, which tends to tion, which, in turn, negatively affect the hydrolysis,
approach a new level during operation at a reduced acidogenesis and methanogenesis steps of the anaerobic
temperature. The effluent COD increases due to the degradation of cow manure.
increase of effluent VFA concentration and suspended Variations in temperature can also affect the entrap-
solids (SS), as well as to the presence of components ment capacity of the sludge bed, as temperature changes
in the influent which remain un-converted. the viscosity, and consequently changes the hydraulic
Investigations by van Lier et al. (1990), conducted in shearing force on the particles (Mahmoud et al., 2003).
a UASB reactor at 39 C fed with synthetic wastewater,
dealt with temperature shocks of 45, 55, 61, and 65 C 3.3. Effects of pH variations
which were imposed during 5, 7, and 24 h periods. They
found that the methane production rate remained at a It is well known that methanogenic activity is more
high level at 45 C, while exposure to higher tempera- likely to proceed optimally in a narrow pH value range,
tures resulted in a serious drop in the activity of meso- between 6.3 and 7.8 (van Haandel, 1994; van Haandel
philic granular sludge due to high bacterial decay. An and Lettinga, 1994). The effect of a drastic pH-change
increase in methane production manifested immediately in the influent depends on the available alkalinity in
after raising the temperature of the system. However, a the reactor. Tests carried out by Borja and Banks
sharp drop soon followed in cases where the reactor was (1995) showed that during a 10-h period, neither an
exposed to temperatures exceeding 45 C. Propionate influent pH of 10 nor an influent pH of 3 significantly
oxidizers were found to be the most sensitive microor- affected the reactor stability. This was because the buffer
ganisms to temperature increments, and methanogens capacity of the system sufficed to maintain the pH of the
were found to be more sensitive than acidogens. These medium in the reactor in the optimal range. In experi-
conclusions are in line with those of Visser et al. ments dealing with the treatment of a synthetic wastewa-
(1993b), who investigated the effect of temperature ter containing VFA and sulphate, Visser et al. (1993a)
shocks in a mesophilic UASB reactor treating sul- concluded that methanogenesis was inhibited at a med-
phate-rich wastewater. Their results revealed that ium-pH exceeding 8, which then resulted in the develop-
increasing the temperature to 45 C did not affect degra- ment of a sludge dominated by sulphate-reducing
dation rates, but temperature elevations to 55 C and bacteria. They also concluded that sulphate-reducing
65 C resulted in a sharp decline of the treatment effi- bacteria are less sensitive to short-term (8 h) pH varia-
ciency. On the other hand, according to the results of tions than methanogenic bacteria.
Rintala and Lepisto (1997), who conducted the metha- Moletta et al. (1994) tested an on-line automatic sys-
nogenic activity test with thermophilic sludge (55 C) tem for pH control of an anaerobic fluidised-bed reac-
at temperatures of 35, 50, 55, 58, 65, and 70 C, there tor. Some of the tests applied to the system can be
1112 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

useful for elucidating what occurs to an anaerobic reac- the wastewater would be discharged without being
tor during a small change in pH. They first injected HCl treated, or would be stored in a buffer reservoir. The
to lower the reactor pH from 6.8 to 6.6, and found an introduction of adapted granules would indicate that
immediate response, viz. the gas production increased these granules would be available at the beginning of
by 40%, as well as the concentration of CO2 in the bio- the production of the new type of wastewater—which
gas. The hydrogen content remained almost unchanged. is troublesome as it is difficult to maintain a great
They also tested the reactor by adding NaOH to in- amount of adapted sludge for a long period, and it is
crease the reactor pH up to 7.4, and observed that the expensive to transport a great amount of that sludge
gas production increased, but the CO2 concentration from another reactor. Thus it seems that the solution
substantially decreased. The variations in the gaseous of the problem is through another type of treatment
phase were the consequence of a shift in CO2 solubility configuration.
with pH. Yang and Anderson (1993) tested three UASB reac-
According to Lettinga et al. (2000), based on experi- tors fed with acetate, sucrose and ice cream in order to
mental results obtained with sugar beet wastewater, pro- assess the long-term effect of distinct wastewater compo-
cess efficiency recovers almost immediately from pH sition on the UASB stability. Three reactors were inoc-
shocks once the influent pH is returned to the optimal ulated with sludge that had been previously adapted to
range. In the case of sudden drastic changes, the recov- sucrose. After a steady-state condition was reached,
ery of the process depends on the extent and duration of two of them were fed with different carbon sources,
the imposed change, as well as on the concentration of i.e. synthetic acetate wastewater and synthetic ice cream
volatile fatty acids during the event. wastewater. With the exception of the carbon source, all
other operational parameters (influent COD, flow rate,
3.4. Effects of shocks with specific compounds temperature and nutrient addition) were kept similar
in all reactors. During the 400 days of operation, the
3.4.1. Variation of the carbon source OLR was gradually increased from 3 to 29 kg COD/
A good example of drastic variations that an anaer- m3 day. The results showed that all reactors behaved
obic reactor may face concerns the sharp fluctuations similarly with rates of up to 10.5 kg COD/m3 day. How-
in the composition of the wastewater subjected to treat- ever, a further increase in OLR led to the deterioration
ment, e.g. such as those occurring in multi-product of reactors fed with sucrose, which included a decreased
food-processing industries. Schmidt and Ahring specific methanogenic activity, excessive non-methano-
(1997) investigated the treatment of these types of genic biomass, predominance of long filamentous bacte-
wastewater using an UASB reactor for a company that ria on granule surfaces and sludge washout. In general,
was processing throughout the season, viz. peas, car- variations of the carbon source present in the waste-
rots, celery roots and leeks, which obviously resulted water caused gradual changes in the physical structures,
in four very different types of wastewaters. Four lab- bacterial distribution and settling characteristics of the
scale UASB reactors were started with the individual granular seed sludge. But, under certain conditions of
wastewaters. Significant differences in the activities OLR, changes of the carbon source can lead to the dis-
and the numbers of microorganisms from different integration or floatation of granular sludge.
metabolic groups were found. After the reactors Fukuzaki et al. (1995) tested four different substrates,
reached steady state performance, each was fed with viz. starch (1.5–3.9 g COD/L), sucrose (1.25–2.5 g COD/
one of the other three wastewaters. Significant de- L), ethanol (2.0–7.0 g COD/L) and butyrate (1.5–2.9 g
creases in the overall efficiency were observed when COD/L) plus propionate (1.5–3.0 g COD/L), to assess
the feed was changed from celery wastewater to one the long-term effect of distinct wastewater composition
of the other wastewaters. This deterioration of the on UASB stability. They used lab-scale UASB reactors
reactor efficiency may be attributed to a significant in- that were operated at 37 C and inoculated with granular
crease in the organic loading rate of the reactor. Such sludge previously acclimatised to synthetic wastewater
an effect on the performance of the system was also containing starch (1.5 g/L). Their research demonstrated
found when leek wastewater, which has a high content that variations in the carbon source present in the waste-
of lipids and proteins, was fed to the reactor. The water caused changes in the physical structures, chemical
researchers proposed some strategies to overcome the contents (extracellular polymeric substances), and bacte-
problems caused by these drastic fluctuations in the rial distribution. The researchers imposed a change in the
composition (very frequent in practice), such as an influent composition of two reactors, viz. the carbon
interruption of feeding the reactor, the introduction source was changed from starch to sucrose, resulting in
of adapted granules to the system, and the use of a sludge floatation and gradual washout, drops in the
buffer tank. However, the solutions proposed by pH, and the collapse of the reactor. On the other hand,
Schmidt and Ahring (1997) are difficult to apply. An when the change was from sucrose to starch, no notice-
interruption of the feeding would mean that either able effect was observed.
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1113

3.4.2. Long-chain fatty acids tory effect of the surfactants on the active biomass
When an AnWT system is exposed to a sudden over- depends upon the adsorbed fraction, as well as the expo-
loading with long-chain fatty acids (LCFA), the risk ex- sure time (Mösche and Meyer, 2002).
ists that the sludge quality will deteriorate due to a Mensah and Forster (2003) examined an anaerobic
serious drop in methanogenic activity as a result of inhi- filter under shock loads of detergents. A mixture of three
bition (Hwu et al., 1996; Koster, 1989; Rinzema et al., different detergents was used to impose the shock, i.e. a
1989). The reactor stability can then hardly be main- concentrate of washing up detergent, a non-biological
tained, and granular sludge may deteriorate further. In hand wash and a fabric softener. The researchers im-
addition to inhibition, some researchers have observed posed a shock load by adding 2 mL/L of the mixture
severe sludge floatation at lauric acid concentrations to the base feed (starch and trace elements) over 12 h.
exceeding 100 mg/L (Koster, 1989; Rinzema et al., During the experiment there was little change in pH
1989). Floatation resulted from the poor release of gas (8.4) and alkalinity (1300 mg CaCO3/L). However, after
bubbles by the granules, due to the adsorption of LCFA 7 h of shock, the reactorÕs performance started to deteri-
at the surface of granular sludge. Moreover, the adhered orate, viz. the effluent COD and VFA concentrations in-
LCFA film may hamper the supply of substrate to the creased steadily and the methane production decreased,
bacteria present in the grains. Another harmful effect showing that the reactor would fail completely if the
is the disintegration of sludge aggregates that can occur shock persisted after 12 h.
when lipids are present. This is because at a neutral pH, Nagel et al. (1999) investigated the response of a lab-
LCFA acts as a surfactant, lowering the surface tension, scale UASB reactor under detergent shock loads. The
and consequently decreasing the aggregation of hydro- reactor was inoculated with granular sludge from a
phobic bacteria. Accordingly, this surfactant effect also plant treating brewery wastewater, and operated at an
cause disaggregation of acetogens, examples of hydro- HRT of 13.3 h. The temperature was maintained be-
phobic bacteria, that can degrade the long-chain fatty tween 30 and 35 C. The influent was the wastewater
acids (Alves et al., 2000). from the same brewery mixed with nutrients. These
researchers imposed three pulses of detergent (phospho-
3.4.3. Detergents ric acid and biodegradable non-ionic surfactant) with
Detergents belong to the category of compounds that different concentrations, viz. 0.1%, 0.4%, 0.6% v/v, rep-
are ordinarily discharged down the drain into municipal resenting the concentrations found at the industry from
sewer systems and via this route, enter sewage treatment which the sludge was obtained. They observed that there
plants. These detergents contain surfactants, which de- was a harmful effect on the reactorÕs performance, i.e.
crease the surface tension when added to a mixed system the methanogenic activity was inhibited and VFA con-
such as water and air, or water and soil. However, centration increased, but the system recovered easily as
according to investigations by Matthijs et al. (1995), a soon as the shock ceased.
distinct biodegradation of these compounds may pro-
ceed in the sewer. This results in a significant reduction 3.4.4. Oxygen
(up to 47%) of the amount of surfactants disposed into Methanogenic bacteria located in sludge granules
the environment, or of those that reach the treatment were found to be well protected, and demonstrated a
plant. However, specific industrial effluents, such as high tolerance for oxygen. However, it must be noted
those from breweries (Nagel et al., 1999), dairies (Eide that this protection can mainly be attributed to the pres-
et al., 2003) and paper and textile industries (Alvarez ence of oxygen-consuming facultative bacteria in the
et al., 2004), contain cleaning products (such as deter- immobilised consortia. They metabolise part of the
gents) at concentrations that can cause toxicity or inhib- available substrate and remove the oxygen, thus creating
itory effects on biological treatment (Nagel et al., 1999; anaerobic microenvironments. Kato (1994) concluded
Khalik et al., 1988). Important attention has to be paid from his studies that the presence of dissolved oxygen
to the linear alkylbenzene sulphonates (LAS), as these at a concentration of 3.8 mg O2/L in the influent of an
are some of the most frequently applied surfactants UASB and an Expanded Granular Sludge Bed reactor
(de Wolf and Feijtel, 1998; Prats et al., 1997) and can in- (EGSB) has no detrimental effects on the anaerobic
hibit anaerobic digestion (Mensah and Forster, 2003; treatment of low strength wastewaters.
Mösche and Meyer, 2002). Experiments with UASB reactors fed sewage showed
Despite the fact that it is very probable that WWTPs that the flocculent sludge of these reactors contained a
may periodically have to deal with a shock load of deter- certain amount of facultative bacteria that, when in
gents, there is very little research that covers this subject. ‘‘steady state’’ operation, rapidly consumed the dis-
However, if an overload of detergent occurs, it is likely solved oxygen present in the influent. This process pro-
that a significant fraction is retained in the sludge bed by ceeded within the bottommost 25 cm, thus protecting
adsorption (Jensen, 1999), while the remainder passes the methanogenic bacteria. However, during an imposed
through the reactor without being treated. The inhibi- hydraulic shock load, the amount of oxygen introduced
1114 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

to the system may exceed the capacity of the faculta- ple-pulse overloading on the FBR performance was
tive bacteria, leading to inhibition of methanogenesis more pronounced than that of a single-pulse overload-
(Leitão, 2004). ing. But considering their results, the effects of the
60-h single pulse and the 3 · 12-h multi-pulse (60 h in to-
3.5. Effects of the duration and frequency of tal) were essentially the same. This may be attributed to
‘‘disturbances’’ the fact that the time between the pulses (12 h) was not
enough to wash out all the sub-products. During the 60-
In practice, various specific disturbances can occur in h single pulse, the pH dropped from 6.9 to around
either the form of occasional pulses (Huang et al., 2000), 6.2, which did not occur during the multi-pulse. This
or step changes in the concentration of a polluting com- may perhaps have been due to the fact that the total
ponent of wastewater, in the flow rate (Nachaiyasit and organic load shock during the single pulse was higher
Stuckey, 1997a,b), or in the temperature, buffer capa- (1 pulse · 60 h/pulse · 15.6 g TOC/L · 0.036 L/h = 33.7
city, pH, etc. These variations frequently manifest g TOC) than the organic load shock during the multi-
in wastewaters originating from industries that use pulse (3 pulses · 12 h/pulse · 15.6 g TOC/L · 0.036
sequential operations or handle various raw materials, L/h = 20.2 g TOC).
e.g. tanneries (Wiegant et al., 1999), breweries (Auster- Xing et al. (1997) examined the effect of a long-term
mann-Haun et al., 1998) or food-processing industries (>400 days) periodic substrate perturbation on an
(Schmidt and Ahring, 1997; Hawkes et al., 1992). How- anaerobic CSTR. This was achieved by introducing an
ever, sharp fluctuations can also manifest periodically in influent with 16 g/L of glucose for 3 days, followed by
domestic sewage, such as those due to human activities, glucose-free influent at the same flow rate for the next
and to climate conditions (van Haandel and Lettinga, 3 days. This procedure resulted in an average concentra-
1994). As a result of these factors, a variety of fluctua- tion of 8 g/L glucose throughout the entire duration of
tions in flow rate and composition frequently occur. the reactor operation. The reactor was maintained at
These fluctuations sometimes proceed smoothly enough an HRT of 10 days and at a temperature of 35 C. Dur-
to enable the operator to take proper measures in the ing the operational period, the response of the reactor
operation of the treatment systems. At other times, the could be subdivided into four distinct phases: (1) during
fluctuations occur as a shock and the system needs to the first 49 days of the operation there was a rapid accu-
have sufficient ‘‘buffer’’ capacity to absorb these instan- mulation of the metabolic intermediates of the glucose
taneous changes, in order to avoid a drastic reduction in fermentation, consisting of VFA, hydrogen and ethanol;
the effluent quality or, in extreme cases, a complete sys- (2) in the next phase, which lasted 240 days, the reactor
tem collapse. reached a so-called ‘‘metastable steady state’’ character-
It is difficult to classify the variations with respect to ised by reduced COD removal efficiency (41%); (3) fol-
duration, as they can range from a few hours up to lowing this long operational phase, the system was
many days, or even longer. The definitions of pulse suddenly capable of degrading the VFA formed, viz.
and step-change need to be related to the operational within a period of 30 days; (4) finally, the reactors could
conditions of the treatment system, because they depend be operated at a high COD removal efficiency. From the
upon the technical features of treatment systems. For experiments described in that paper, it is not clear how
example, an imposed change of 2 days can be defined and why the reactor suddenly behaved so differently
as a step-change for a reactor operated at an HRT of after such a long period of operation. The performance
4 h, while it is a pulse in the case of a stabilisation pond, of the perturbed reactor started to improve when the
which is operated at an HRT of 10 days. hydrogen concentration decreased and pH increased
El Farhan and Shieh (1999) investigated the response from 6.1 to 7.1. However, no additional buffer was pro-
of a lab-scale fluidised bed reactor (FBR) towards single vided. The results of the experiment show that it is pos-
and multiple-pulse overloading, and towards step sible to treat a wastewater with high substrate
changes (considering a pulse time > 1 HRT will be de- fluctuation if enough buffer capacity is present.
fined as step change). The reactor was operated at an
HRT of 20 h and fed with glucose plus nutrients (total
organic carbon—TOC of 5.2 g/L). The single pulse 4. Discussion and conclusions
experiment was carried out by increasing the concentra-
tion to 200% for a period of 16 h. The concentration in Based on the information reported in this work, it
the step change was also raised to 200%, but for periods can be concluded that operational and environmental
lasting 25 and 60 h. The multi-pulse experiments con- variations exist and will always exert an effect on waste-
sisted of three sequential pulses with a duration of 6 h, water treatment systems. It is clear that the reactor re-
with 6 h between the pulses, and another at three pulses sponse varies significantly, not only depending on
of 12 h, with 12 h between the pulses. The authors af- factors related to the treatment system (reactor configu-
firmed that at similar loadings, the impact of a multi- ration, ratio of organic load and organic load potential,
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1115

available alkalinity and availability of a fault detection pH values. This means that methanogenic activity can
and control system), but also on factors related to the become inhibited at a lower pH values, while VFA are
variation itself, viz. type of the imposed shock, its extent, still produced, which may end in the acidification of
frequency and duration. the reactor.
In general terms, it can be said that anaerobic reac- Fluctuations in wastewater compositions have imme-
tors behave in a similar way when exposed to some diate effects on the performance of reactors, as the bal-
abrupt change in operational or process conditions. ance of different metabolic groups of microorganisms
The typical response is an incomplete methanogenesis, depends upon the composition of the wastewater. If
resulting in a certain accumulation of VFA (mainly pro- the change in the carbon source lasts for a long time,
pionate and butyrate), drop in pH value and alkalinity, a shift in the proportion of the various groups can
change in the biogas production and composition (in- occur, which means that a new steady state will be
creases in the CO2 and H2 gas contents), and, some- established.
times, higher sludge washout. The sudden occurrence of high concentrations of
Organic load variations can be divided into two dif- xenobiotics, heavy metals, detergents, oxygen, etc. is
ferent classes, those which are due to suspended solids very common in treatment plants. Their effects depend
variation, and those due to dissolved solids variation. on the severity (duration and concentration) of the
Each class has their own distinct effects. An extra contri- event, but these compounds inhibit the methanogens
bution in the load of SS can lead to a decrease in SRT more than the other anaerobic micro-organisms, and
and further deterioration of reactor performance. Over- consequently cause accumulation of VFA and a drop
loading due to dissolved degradable compounds can in pH. In the case of oxygen, the facultative bacteria
lead to an accumulation of VFA, a drop in pH values, present in the granules can use this element before it
and possibly an inhibition of methanogenic activity. can affect the methanogens. However, during a hydrau-
In UASB reactors, hydraulic load variations affect lic shock load, the amount of oxygen can exceed the
the dynamics of the sludge bed. They expand the bed capacity of the facultative bacteria present in the granule
due to a new equilibrium between the upflow and sludge or floc, causing inhibition of methanogenesis.
settling velocities. Depending on the variation, a higher The notion of stability or robustness of the anaerobic
SS concentration in the effluent can be expected due to reactors is still rather confusing. Robustness can be de-
the washout of lighter biomass, the decreased filtration fined as the capacity of the treatment systems to reach
capacity of the sludge bed at higher upflow velocities steady state performance under certain environmental
and the disintegration of granules or flocs under the and operational conditions. In this regard the AnWT
abrasive action of shear forces. The treatment capacity systems are as stable as any other biological system.
can also deteriorate, due to insufficient contact between But stability could also be defined in terms of variability
the sludge bed and the substrate. During an increase in of the final product of the process, i.e. the effluent. Fur-
upflow velocity, the mass transport rises and can also thermore, stability can be defined as the capacity of a
cause an organic overload, the consequences of which system to cope with more severe environmental and
were discussed previously, i.e. VFA accumulation, pH operational variations. In the case of AnWT, some seri-
drop and inhibition of methanogenesis. ous problems of reliability may manifest, which has led
Variations in temperature can dramatically affect the to a certain prejudice regarding the use of anaerobic
performance of anaerobic reactors because of the differ- reactors to treat municipal sewage.
ent responses of various metabolic groups of microor-
ganisms. A drop in the activity of methanogens occurs
at temperatures lower than 16 C, which can lead to 5. Future research needs
an accumulation of VFA and a drop in pH. Moreover,
hydrolysis significantly slows down below this tempera- Despite all of the aforementioned studies, there still
ture and an accumulation of inert suspended solids in remain some unclear aspects which could improve the
the reactor can occur, leading to a decreasing SRT stability and reliability of anaerobic processes. For in-
and deterioration of sludge quality. An increase in the stance, in the case of sewage treatment using UASB
temperature can increase the decay rate of methanogenic reactors, the applicable hydraulic retention times are
bacteria (which are more sensitive to temperature varia- still a subject of controversy. Moreover, the use of this
tions) to values exceeding the growth rate. This undesir- system for the treatment of sewage with relatively low
able situation deteriorates the reactor performance. or high COD concentrations, i.e. around 200 mg
A variation of the influent pH value can affect the COD/L (Seghezzo et al., 2002) or around 1500 mg
reactor performance, but it is dependent upon the buffer COD/L (Halalsheh, 2002), is still undergoing trials. This
capacity of the mixed liquor. Methanogenic activity has is because knowledge of the performance of anaerobic
its optimum pH value within the range of 6.5–7.5, but reactors treating municipal wastewater in extreme situa-
acidogenic bacteria are less sensitive to higher or lower tions is quite limited.
1116 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

During a hydraulic or organic shock load the sludge ment body responsible for the development of science
bed will expand (Leitão, 2004; Setiadi, 1995). However, and technology.
there is a lack of information about the relationship be-
tween the shock strength and the sludge bed expansion
phenomenon, as well as the effect of several other oper- References
ational parameters on the dynamic behaviour of the
sludge bed. Aisse, M.M., Lobato, M.B., Bona, A., Garbossa, L.H.P., Jurgensen,
Mahmoud (2002), Wang (1994) and several other D., Sobrinho, P.A., 2000. Avaliação do sistema reator UASB e
filtro biológico para o tratamento de esgoto sanitário. In: Proc.
researchers have studied two-step systems, the first step XXVII Congresso Interamericano de Engenharia Sanitária e
being used for removing and hydrolysing the suspended Ambiental, Porto Alegre, Brazil.
solids. However, for tropical conditions, methanogene- Alvarez, D., Garrido, N., Sans, R., Carreras, I., 2004. Minimization–
sis will always happen in the first step (Leitão, 2004). optimization of water use in the process of cleaning reactors and
containers in a chemical industry. Journal of Cleaner Production
Therefore, it is important to study a two-step system
12 (7), 781–787.
comprised of a upflow anaerobic reactor followed by Alves, M., Cavaleiro, A.J., Ferreira, E.C., Amaral, A.L., Mota, M., da
some type of suspended solids separation step, which Motta, M., Vivier, H., Pons, M.N., 2000. Characterisation by
is the opposite of what has been investigated. Moreover, image analysis of anaerobic sludge under shock conditions. Water
the first step (an UASB or an EGSB) should be investi- Science and Technology 41 (12), 207–214.
gated when operated at maximum sludge retention Austermann-Haun, U., Lange, R., Seyfried, C.F., Rosenwinkel, K.H.,
1998. Upgrading an anaerobic/aerobic wastewater treatment plant.
capacity. This new concept should improve the metha- Water Science and Technology 37 (9), 243–250.
nogenic potential of the reactor, as well as the SS entrap- Bechmann, H., Nielsen, M.K., Madsen, H., Poulsen, N.K., 1998.
ment capacity, as the sludge bed is maintained at its Control of sewer systems and wastewater treatment plants using
maximum volume and height. pollutant concentration profiles. Water Science and Technology 37
An imposed stress condition causes changes in the (12), 87–93.
Bertrand-Krajewski, J.L., Chebbo, G., Saget, A., 1998. Distribution of
biological population of anaerobic reactors (Alves pollutant mass vs volume in stormwater discharges and the first
et al., 2000; Kalyuzhnyi et al., 1996; Cheng, 1992). Thus flush phenomenon. Water Research 32 (8), 2341–2356.
far, little relevant information can be found about Bertrand-Krajewski, J.L., Lefebvre, M., Lefai, B., Audic, J.M., 1995.
changes in specific sludge characteristics due to imposed Flow and pollutant measurements in a combined sewer system to
changes in environmental conditions. operate a wastewater treatment plant and its storage tank during
storm events. Water Science and Technology 31 (7), 1–12.
The contradictions about the effect of a variation in Beyenal, H., Lewandowski, Z., 2000. Combined effect of substrate
the flow rate on the biofilm diffusion rate should be concentration and flow velocity on effective diffusivity in biofilms.
investigated (Beyenal and Lewandowski, 2000; Zaiat Water Research 34 (2), 528–538.
et al., 1996). It is possible that well-controlled hydraulic Bhatia, D., Vieth, W.R., Venkatasubramaniam, K., 1985a. Steady-
shock loads can improve the performance of anaerobic state and transient behaviour in microbial methanification: I.
Experimental results. Biotechnology and Bioengineering 27 (8),
reactors by decreasing the mass transfer resistance 1192–1198.
(Brito and Melo, 1999). However, an increase in OLR Bhatia, D., Vieth, W.R., Venkatasubramaniam, K., 1985b. Steady-
can lead to a increased KS (Bhatia et al., 1985a,b), state and transient behaviour in microbial methanification: II.
which can deteriorate the capacity of the reactor to treat Mathematical modelling and verification. Biotechnology and
Bioengineering 27 (8), 1199–1207.
diluted wastewater.
Borja, R., Banks, C.J., 1995. Response of an anaerobic fluidized bed
The performance of anaerobic reactors can deterio- reactor treating ice-cream wastewater to organic, hydraulic, tem-
rate when a change in the composition of the wastewater perature and pH shocks. Journal of Biotechnology 39 (3), 251–259.
occurs (Schmidt and Ahring, 1997). Detergents (Nagel Borja, R., Banks, C.J., Wang, G., 1995. Effect of organic loading rate
et al., 1999; Khalik et al., 1988) and long-chain fatty on anaerobic treatment of slaughterhouse wastewater in a fluid-
acids (Hwu et al., 1996; Koster, 1989; Rinzema et al., ized-bed reactor. Bioresource Technology 52, 157–162.
Brito, A.G., Melo, L.F., 1999. Mass transfer coefficients within
1989) can cause toxicity or inhibitory effects on anaero- anaerobic biofilms: effects of external liquid velocity. Water
bic treatment. All these aspects need more investigations. Research 33 (17), 3673–3678.
Leitão (2004) reported that anaerobic biomass has a Butler, D., Friedler, E., Gatt, K., 1995. Characterising the quantity
certain lag-phase before responding to a shock load of and quality of domestic wastewater inflows. Water Science and
VFA. This phenomenon should be investigated, as it Technology 31 (7), 13–24.
Campos, H.M., von Sperling, M., 1996. Estimation of domestic
can give important information about the capacity of wastewater characteristics in a developing country based on socio-
the reactor to cope with sudden variations of the OLR. economic variables. Water Science and Technology 34 (3–4), 71–
77.
Cánovas-Dı́az, M., Howell, J.A., 1988. Downflow fixed-film anaerobic
Acknowledgements reactors stability studies under organic and hydraulic overloading
at different working volume. Water Research 22 (5), 529–536.
Carstensen, J., Nielsen, M.K., Strandbaek, H., 1998. Prediction of
This work was accomplished with the support of hydraulic load for urban storm control of a municipal WWT plant.
CNPq (Project no. 200198/00-9), a Brazilian Govern- Water Science and Technology 37 (12), 363–370.
 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118 1117

Castillo, A., Cecchi, F., Mata-Alvarez, J., 1997. A combined anaer- Inanc, B., Matsui, S., Ide, S., 1999. Propionic acid accumulation in
obic–aerobic system to treat domestic sewage in coastal areas. anaerobic digestion of carbohydrates: an investigation on the role
Water Research 31 (12), 3057–3063. of hydrogen gas. Water Science and Technology 40 (1), 93–100.
Cha, G.C., Noike, T., 1997. Effect of rapid temperature change and Jensen, J., 1999. Fate and effects of linear alkylbenzene sulphonates
HRT on anaerobic acidogenesis. Water Science and Technology 36 (LAS) in the terrestrial environment. The Science of the Total
(6–7), 247–253. Environment 226, 93–111.
Cheng, S.G., 1992. Population dynamics of attached biofilm in Kalyuzhnyi, S.V., Sklyar, V.I., Davlyatshina, M.A., Parshina, S.N.,
anaerobic fluidized bed pilot plant. Water Science and Technology Simankova, M.V., Kostrikina, N.A., Nozhevnikova, A.N., 1996.
26, 503. Organic removal and microbiological features of UASB-reactor
Chua, H., Hu, W.F., Yu, P.H.F., Cheung, M.W.L., 1997. Response of under various organic loading rates. Bioresource Technology 55
an anaerobic fixed-film reactor to hydraulic shock loadings. (1), 47–54.
Bioresource Technology 61, 79–83. Kato, M.T., 1994. The anaerobic treatment of low strength soluble
Cohen, A., Breure, A.M., van Andel, J.G., van Deursen, A., 1982. wastewater. Ph.D. Thesis. Wageningen University.
Influence of phase separation on the anaerobic digestion of Khalik, E.F., Whitmore, T.N., Gamal, E.D.H., El Bassel, A., Lloyd,
glucose—II: stability, and kinetic response to shock loadings. D., 1988. The effects of detergents on anaerobic digestion. Applied
Water Research 16, 449–455. Microbiology and Biotechnology 29 (5), 517–522.
Dauphin, S., Joannis, C., Deguin, A., Bridoux, G., Ruban, G., Koster, I.W., 1989. Toxicity in anaerobic digestion—with emphasis on
Aumond, M., 1998. Influent flow control to increase the pollution the effect of ammonia, sulphide and long-chain fatty acids
load treated during rainy periods. Water Science and Technology on methanogenesis. Ph.D. Thesis. Wageningen Agricultural
37 (12), 131–139. University.
de Wolf, W., Feijtel, T., 1998. Terrestrial risk assessment for linear Krebs, P., Holzer, P., Huisman, J.L., Rauch, W., 1999. First flush of
alkyl benzene sulphonate (LAS) in sludge-amended soils. Chemo- dissolved compounds. Water Science and Technology 39 (9), 55–
sphere 36 (6), 1319–1343. 62.
Deletic, A., 1998. The first flush load of urban surface runoff. Water Leitão, R.C., 2004. Robustness of UASB reactors treating sewage
Research 32 (8), 2462–2470. under tropical conditions. Ph.D. Thesis. Wageningen University.
Eide, M.H., Homleid, J.P., Mattsson, B., 2003. Life cycle assessment Lettinga, G., Hulshoff Pol, L.W.H., Zeeman, G., 2000. Biological
(LCA) of cleaning-in-place processes in dairies. Lebensmittel Wastewater Treatment—Part 1—Anaerobic Wastewater Treat-
Wissenschaft und Technologie 36, 303–314. ment. Wageningen University, Wageningen, The Netherlands,
El Farhan, M., Shieh, W.K., 1999. Overloading responses of a glucose- p. 200.
fed anaerobic fluidized bed reactor. Biochemical Engineering Mahmoud, N.J.A.-H., 2002. Anaerobic pre-treatment of sewage under
Journal 3, 17–23. low temperature (15 C) conditions in an integrated UASB-digester
El-Mashad, H.E.M.H., 2003. Solar thermophilic anaerobic reactor system. Ph.D. Thesis. Wageningen University.
(STAR) for renewable energy production. Ph.D. Thesis. Wagen- Mahmoud, N., Zeeman, G., Gijzen, H., Lettinga, G., 2003. Solids
ingen University. removal in upflow anaerobic reactors, a review. Bioresource
Eng, S.C., Fernandes, X.A., Paskins, A.R., 1986. Biochemical effects Technology 90, 1–9.
of administering shock loads of sucrose to a laboratory-scale Matthijs, E., Debaere, G., Itrich, N., Masscheleyn, P., Rottiers, A.,
anaerobic (UASB) effluent treatment plant. Water Research 20 (6), Stalmans, M., Federle, T., 1995. The fate of detergent surfactants
789–794. in sewer systems. Water Science and Technology 31 (7), 321–328.
Fongastitkul, P., Mavinic, D.S., Lo, K.V., 1994. A two-phased Mensah, K.A., Forster, C.F., 2003. An examination of the effects of
anaerobic digestion process: concept, process failure and maxi- detergents on anaerobic digestion. Bioresource Technology 90,
mum system loading rate. Water Environment Research 66 (3), 133–138.
243–254. Metcalf & Eddy, Inc., 1991. Wastewater Engineering: Treatment,
Friedler, E., Butler, D., 1996. Quantifying the inherent uncertainty in Disposal, Reuse. McGraw-Hill Book Company, New York, USA,
the quantity and quality of domestic wastewater. Water Science p. 1334.
and Technology 33 (2), 65–78. Moletta, R., Escoffier, Y., Ehlinger, F., Coudert, J.P., Leyris, J.P.,
Fukuzaki, S., Nishio, N., Nagai, S., 1995. High rate performance and 1994. On-line automatic control system for monitoring an anaer-
characterization of granular and methanogenic sludges in Upflow obic fluidized-bed reactor: response to organic overload. Water
Anaerobic Sludge Blanket reactors fed with various defined Science and Technology 30 (12), 11–20.
substrates. Journal of Fermentation and Bioengineering 79 (4), Mösche, M., Meyer, U., 2002. Toxicity of linear alkylbenzene
354–359. sulfonate in anaerobic digestion: influence of exposure time. Water
Halalsheh, M.M., 2002. Anaerobic pre-treatment of strong sewage: a Research 36, 3253–3260.
proper solution for Jordan. Ph.D. Thesis. Wageningen University. Nachaiyasit, S., Stuckey, D.C., 1997a. The effect of shock loads on the
Hawkes, F.R., Rozzi, A., Black, K., Guwy, A., Hawkes, D.L., 1992. performance of an anaerobic baffled reactor (ABR). 1. Step
The stability of anaerobic digesters operating on a food-processing changes in feed concentration at constant retention time. Water
wastewater. Water Science and Technology 25 (7), 73–82. Research 31 (11), 2737–2746.
Heip, L., van Assel, J., Swartenbroekx, P., 1997. Sewer flow quality Nachaiyasit, S., Stuckey, D.C., 1997b. The effect of shock loads on the
modelling. Water Science and Technology 36 (5), 177–184. performance of an anaerobic baffled reactor (ABR). 2. Step and
Holst, T.C., Truc, A., Pujol, R., 1997. Anaerobic fluidized beds: ten transient hydraulic shocks at constant feed strength. Water
years of industrial experience. Water Science and Technology 36 Research 31 (11), 2747–2754.
(6–7), 415–422. Nadais, H., Capela, I., Arroja, L., Duarte, A., 2000. Effect of organic,
Huang, Y.H., Huang, G.H., Chou, S., Cheng, S.S., 2000. Hydrogen as hydraulic and fat shock on the performance of UASB reactors with
a quick indicator of organic shock loading in UASB. Water Science intermittent operation. In: Proc. VI Oficina e Seminário Latino-
and Technology 42 (3–4), 43–50. Americano de Digestão Anaeróbia, Recife, Pernambuco, Brazil,
Hwu, C.S., Donlon, B., Lettinga, G., 1996. Comparative toxicity of pp. 75–82.
long-chain fatty acid to anaerobic sludges from various origins. Nagel, P., Urtubia, A., Aroca, G., Chamy, R., Schiappacasse, M.,
Water Science and Technology 34 (5–6), 351–358. 1999. Methanogenic toxicity and anaerobic biodegradation of
1118 R.C. Leitão et al. / Bioresource Technology 97 (2006) 1105–1118

chemical products in use in a brewery. Water Science and reactor treating settled sewage at moderate temperatures. Water
Technology 40 (8), 169–176. Science and Technology 45 (10), 243–248.
Oliva, L.C.H.V., 1997. Tratamento de Esgotos Sanitários com Reator Setiadi, T., 1995. Predicting the bed expansion of an anaerobic
Anaeróbio de Manta de Lodo (UASB) Protótipo: Desempenho e fluidised-bed bioreactor. Water Science and Technology 31 (9),
Respostas Dinâmicas às Sobrecargas Hidráulicas. Ph.D. Thesis. 181–191.
Universidade de São Paulo, EESC, USP. van Haandel, A., 1994. Influence of the digested COD concentration
Omil, F., Bakker, C.D., Hulshoff Pol, L.W.H., Lettinga, G., 1997. on the alkalinity requirement in anaerobic digesters. Water Science
Effect of pH and low temperature shocks on the competition and Technology 30 (8), 23–24.
between sulphate reducing bacteria and methane producing bacteria van Haandel, A.C., Lettinga, G., 1994. Anaerobic Sewage Treatment:
in UASB reactors. Environmental Technology 18 (3), 255–264. a Practical Guide for Regions with a Hot Climate. John Wiley &
Orhon, D., Sözen, S., Görgün, E., Çokgör, E.U., Artan, N., 1999. Sons, Chichester, UK, p. 226.
Technological aspects of wastewater management in coastal tourist van Lier, J.B., Rintala, J., Sanz Martin, J.L., Lettinga, G., 1990. Effect
areas. Water Science and Technology 39 (8), 177–184. of short-term temperature increase on the performance of a
Prats, D., Ruiz, F., Vásquez, B., Rodriguez Pastor, M., 1997. Removal mesophilic UASB reactor. Water Science and Technology 22 (9),
of anionic and nonionic surfactants in a wastewater treatment 183–190.
plant with anaerobic digestion. A comparative study. Water van Lier, J.B., Tilche, A., Ahring, B.K., Macarie, H., Moletta, R.,
Research 31 (8), 1925–1930. Dohanyos, M., Hulshoff Pol, L.W., Lens, P., Verstraete, W., 2001.
Puñal, A., Lema, J.M., 1999. Anaerobic treatment of wastewater from New perspectives in anaerobic digestion. Water Science and
a fish-canning factory in a full-scale Upflow Anaerobic Sludge Technology 43 (1), 1–18.
Blanket (UASB) reactor. Water Science and Technology 40 (8), 57– Visser, A., Gao, Y., Lettinga, G., 1993a. Effects of pH on methano-
62. genesis and sulphate reduction in thermophilic (55 C) UASB
Rintala, J.A., Lepisto, S.S., 1997. Pilot-scale thermophilic anaerobic reactors. Bioresource Technology 44 (2), 113–121.
treatment of wastewaters from seasonal vegetable processing Visser, A., Gao, Y., Lettinga, G., 1993b. Effects of short-term
industry. Water Science and Technology 36 (2–3), 279–285. temperature increases on the mesophilic anaerobic breakdown of
Rinzema, A., Alphenaar, A., Lettinga, G., 1989. The effect of lauric sulfate containing synthetic wastewater. Water Research 27 (4),
acid shock loads on the biological and physical performance of 541–550.
granular sludge in UASB reactors digesting acetate. Journal of Wang, K., 1994. Integrated anaerobic and aerobic treatment of
Chemical Technology and Biotechnology 46 (4), 257–266. sewage. Ph.D. Thesis. Wageningen Agricultural University.
Saget, A., Chebbo, G., Bertrand Krajewski, J.L., 1996. The first flush Wiegant, W.M., Kalker, T.J.J., Sontakke, V.N., Zwaag, R.R., 1999.
in sewer systems. Water Science and Technology 33 (9), 101–108. Full scale experience with tannery water management: an inte-
Schaarup-Jensen, K., Hvitved-Jacobsen, T., Jütte, B., Nielsen, B., grated approach. Water Science and Technology 39 (5), 169–176.
Pedersen, T., 1998. A Danish sewer research and monitoring Xing, J., Criddle, C., Hickey, R., 1997. Effects of a long-term periodic
station. Water Science and Technology 37 (1), 197–204. substrate perturbation on an anaerobic community. Water
Schmidt, J.E., Ahring, B.K., 1997. Treatment of waste water from a Research 31 (9), 2195–2204.
multi product food-processing company, in Upflow Anaerobic Yang, G., Anderson, G.K., 1993. Effects of wastewater composition on
Sludge Blanket (UASB) reactors: the effect of seasonal variation. stability of UASB. Journal of Environmental Engineering 119 (5),
Pure and Applied Chemistry 69 (11), 2447–2452. 958–977.
Seghezzo, L., Guerra, R.G., González, S.M., Trupiano, A.P., Figue- Zaiat, M., Vieira, L.G.T., Foresti, E., 1996. Liquid-phase mass transfer
roa, M.E., Cuevas, C.M., Zeeman, G., Lettinga, G., 2002. Removal in fixed-bed of polyurethane foam matrices containing immobilized
efficiency and methanogenic activity profiles in a pilot-scale UASB anaerobic sludge. Biotechnology Techniques 10 (2), 121–126.