Journal of Cleaner Production 172 (2018) 1121e1127

Contents lists available at ScienceDirect

Journal of Cleaner Production

journal homepage: www.elsevier.com/locate/jclepro

Optimization of surface area loading rate for an anaerobic moving bed

biofilm reactor treating brewery wastewater
Alessandro di Biase*, Tanner Ryan Devlin, Maciej Kowalski, Jan A. Oleszkiewicz
Department of Civil Engineering, University of Manitoba, Winnipeg, R3T 5V6, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Three acclimated anaerobic moving bed biofilm reactors filled to 40% by volume with cubic AC920 media
Received 6 June 2017 (protected surface area of 680 m2 m
3) were studied to treat brewery wastewater at hydraulic retention
Received in revised form times of 6e24 h and mesophilic condition (35.0 ± 0.1  C). The reactors were seeded with media that had
20 October 2017
developed mature anaerobic biofilm prior to the start of this study. Soluble chemical oxygen demand of
Accepted 22 October 2017
the brewery wastewater was equal to 4.0 ± 1 kg m
3, total chemical oxygen demand 5.2 ± 2.1 kg-COD
Available online 27 October 2017
m
3, and the ratio of 5-day biological oxygen demand to chemical oxygen demand was equal to
0.77 ± 0.02. Dissolved oxygen concentration in the influent wastewater was below detection limit
Keywords:
Hydraulic retention time
(0.1 mg-O2 L
1). This study focused on the assessment of optimal design surface area loading rates for an
Anaerobic moving bed-biofilm reactor anaerobic moving bed biofilm process by varying the hydraulic retention times from 6 to 24 h. Volu-
Brewery wastewater treatment metric removal rates of 3.0 ± 0.2, 4.2 ± 0.3, 7.2 ± 0.6, 12.4 ± 0.8, 13.4 ± 1.0, and 12.0 ± 1.1 g-sCOD d
1 were
Methane yield achieved at 24, 18, 12, 10, 8, and 6 h hydraulic retention times. Hydraulic retention time of 18 h exhibited
Surface area loading rate the highest chemical oxygen demand removal, at 88 ± 2.5%, and the highest carbon conversion into
methane at 0.34 ± 0.06 m3-CH4 kg-sCOD
1. At this hydraulic retention time the specific surface area
activity was 13 ± 1 g-sCOD m
2d
1, considering that the organic loading rate was 4.0 ± 0.3 kg-sCOD
m
3d
1. Biogas composition was 71 ± 2% methane and 26 ± 4% carbon dioxide. Media clogging and mass
transfer limitations due to overgrowing biomass in the interior media channels was observed at surface
area loading rates above 54 g-sCOD m
2 d
1. However, kinetic tests revealed that mass transfer limitation
started at surface area loading rate of 44 g-sCOD m
2d
1. Anaerobic moving bed biofilm reactor has been
shown to be a reliable treatment option for high-strength wastewaters and could be used in brewery
wastewater management as a resource to comply with the discharge limit regulation.
© 2017 Elsevier Ltd. All rights reserved.

1. Introduction
Anaerobic digestion processes treating high strength wastewa-
ters require extended solid residence time for maintaining meth-
anogenic microbial communities. However, industries must
simultaneously aim for low hydraulic retention time (HRT) in order
to allow for high volumetric throughput of wastewater. In the past
few decades, different configurations of high-rate anaerobic in-
dustrial wastewater treatment processes have been developed. The
most common full-scale treatment in food and beverage industry
(e.g., winery, dairy, brewery, potato and slaughter house) is the up-
flow anaerobic sludge blanket reactor (UASB) or the expanded
* Corresponding author.
E-mail addresses: alessandro.dibiase.fi@gmail.com, dibiasea@myumanitoba.ca
(A. di Biase).
granular sludge bed (EGSB) reactor (Castillo et al., 2017). The up-
flow anaerobic sludge blanket reactors were developed for the
effective treatment of domestic wastewaters. An alternative to
conventional UASB, providing equally robust performance, could be
an anaerobic fixed film reactor such as fluidized bed, anaerobic
baffled, anaerobic hybrid bioreactor and anaerobic moving bed
biofilm reactor (AMBBR; di Biase et al., 2017).
Within biofilm based processes, the moving bed-biofilm reactor
has proven to be very reliable because of its high volumetric
organic loading rate (OLR) and low sludge production (Ødegaard,
2016). The principle behind this technology is biomass growing
on a protected surface area within carriers, which are physically
retained in the reactor by retention sieves. Studies carried out
within the last two decades have demonstrated the effectiveness
and feasibility of moving bed biofilm reactor technology in both
aerobic and anoxic environments; aerobic removal of biological
oxygen demand (BOD), chemical oxygen demand (COD), and

https://doi.org/10.1016/j.jclepro.2017.10.256
0959-6526/© 2017 Elsevier Ltd. All rights reserved.
1122 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127

area of 680 m2 m
3 resulted in a total surface area of 1.1 m2

List of abbreviations available for biofilm development. The reactors were originally
inoculated with mixed culture anaerobic digestate from the North
AMBBR Anaerobic moving bed biofilm reactor End Water Pollution Control Centre in Winnipeg, Canada, to an
BOD Biological oxygen demand initial suspended solids concentration of 15 kg-TSS m
3. The vola-
COD Total chemical oxygen demand tile (VSS) to total suspended solids (TSS) ratio of 0.6 VSS TSS
1
EGSB Expanded granular sludge bed corresponded to 40% inert suspended solids. Synthetic and brewery
HRT Hydraulic retention time wastewater were previously run through the reactors according to
OLR Organic loading rate di Biase et al. (2016), and the maximum volumetric load previously
SALR Surface area loading rate experienced was approximately 23 ± 1 kg-COD m
3d
1. The pre-
sCOD Soluble chemical oxygen demand viously colonized media were re-distributed between the three
SSA Specific surface area reactors in order to achieve 40% filling by volume (500 ± 50 units
SSAA Specific surface area activity each reactor).
TSS Total suspended solids Brewery wastewater was stored in a 200 L tank at 4  C, and
UASB Upflow anaerobic sludge blanket reactor dissolved oxygen concentration in the effluent was below detection
VSS Volatile suspended solids limits (0.1 mg-O2 L
1) making sure anaerobic condition were ach-
ieved. A heat exchanger raised the temperature of wastewater to
35 ± 0.1  C before entering the reactors. Mechanical mixing was
employed and operational temperature was maintained at
ammonia nitrogen (Raudkivi et al., 2017; Zekker et al., 2016; 35 ± 0.1  C via an environmental chamber (SR-Chamber, CON-
McQuarrie and Boltz, 2011). Other studies, demonstrate that in VIRON). Gas-proof tubing (Nalgene 180 PVC 0.635 cm ID, 1.5 OD)
moving bed biofilm reactor systems pollutants removal is depen- and inline digital mass flow meters (FMA 4000, OMEGA, Laval, PQ)
dant on temperature and toxic nitrite concentrations, and oxidized were applied to determine the total biogas production rate. The
nitrogen forms (Rikmann et al., 2017; Zekker et al., 2017). working volume of 4 L was maintained by gravity flow effluent
Anaerobic moving bed biofilm reactors for cultivation of meth- through an inverted syphon, in order to limit oxygen transfer. Hy-
anogenic communities have been applied to treat winery and milk draulic retention times were changed from and initial value of 24 h
permeate wastewaters (Sheli et al., 2014; Sheli and Moletta, 2007, to a final value of 6 h by increasing flowrate. Hydraulic retention
2010; Wang et al., 2009). Despite having been tested on these times of 24, 18, 12, 10, 8, and 6 h were each maintained for at least
wastewaters, very little information is available on proper design three weeks. Influent alkalinity concentrations were kept at
parameters in order to avoid mass transfer limitation. 1.5 ± 0.1 kg of calcium carbonate (CaCO3) per cubic meter and
Hydraulic retention time is the key design parameter to be increased to 1.7 ± 0.2 kg-CaCO3 m
3 to enhance pH buffering ca-
investigated in order to evaluate feasibility of AMBBR technology in pacity at the lowest HRT applied in order to maintain influent pH
treating industrial wastes, such as brewery wastewater. Hydraulic values of 6.7 ± 0.1.
retention time defines footprint and therefore capital cost of
wastewater pre-treatment prior to discharge. The goal of this study
2.2. Wastewater characteristics
was to assess the optimal design surface area loading rate (SALR) of
an AMBBR process treating brewery wastewater at HRT from 6 to
Raw fermenter wastewater from the Fort Garry Brewery in
24 h at constant media filling fraction.
Winnipeg, Canada, was collected for this study. The Fort Garry
Brewery is a medium sized brewery equipped with 15 continuously
2. Material and methods operating fermenters and 8 maturation tanks. Their average daily
beer production is approximately 50 m3, resulting in an estimated
2.1. Reactor setup 150 m3 d
1 of wastewater production. At 3 L wastewater generated
per one L of beer produced, the Fort Garry Brewery has lower water
Brewery wastewater was treated in three identical 4 L sealed consumption than typical range of 4e7 L of water per liter of beer
acrylic reactors e Fig. 1. Each reactor contained 500 ± 50 AC 920 produced reported by Olajirea (2012). Table 1 reports raw waste-
media (Headworks BIO, Victoria BC, Canada), which corresponded water characteristics for the Fort Garry Brewery during highest
to 40% media fill by volume. The media's specific protected surface production in May and June. The biological to total chemical oxygen
demand ratio (BOD5/COD) of the For Garry raw wastewater was
0.77; higher than typical brewery wastewater ratio of 0.6e0.7
(Brito et al., 2007).
In this study, COD concentrations were kept as close as possible
to average North American brewery wastewater characteristics
(Brewery Association, 2012) by diluting waste sampled from the
bottom of fermenters (140 ± 10 kg-COD m
3). Fermenter underflow

Table 1
Brewery wastewater characteristics.

Parameter Value Unit

sCOD 6±2 kg m
3

COD 7±3 kg m
3
BOD5 5.3 ± 0.2 kg m
3
BOD5/COD 0.77 e
TSS 0.4 ± 0.1 kg m
3
VSS 0.3 ± 0.1 kg m
3
Fig. 1. Reactor configuration.
 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127 1123

was used as substrate because it is the main source of organics in  

brewery wastewater (Chastain et al., 2011). This resulted in influent
concentrations of soluble COD (sCOD) of 4.0 ± 1.0 kg-sCOD m
3 and
total COD (COD) of 5.2 ± 2.1 kg-COD m
3. The biological oxygen
demand was measured to be 3.9 ± 1.6 kg-BOD5 m
3 with a BOD5/
COD ratio of 0.75. Even though biodegradation process produces
bicarbonate alkalinity neutralising the acidity of mixed liquor
(Tenno et al., 2016), additional alkalinity was dosed to prevent
excess accumulation of volatile fatty acids in the reactors. Initially,
1.5 ± 0.1 kg-CaCO3 m
3 was added before being increased to
1.7 ± 0.2 kg-CaCO3 m
3 as sodium bicarbonate at the 6 h HRT to
promote pH-buffering capacity.
2.3. Analytical methods
Samples were taken from the reactor influent and effluent three
times per week. The pH was recorded with Alpha pH200 meter
(Eutech Instruments, USA) immediately upon sampling and alka-
linity was measured according to Standard Methods (APHA, 2012).
Total and soluble COD were analyzed spectrophotometrically
(filtered through 0.45 mm GF/C filters, Whatman, UK) as per Stan-
dard Methods, as well as TSS and VSS. Furthermore, BOD was
determined using Oxitop bottles (model IS 12, Xylem, CA).
Surface removal rates expressed as specific surface area activity
(SSAA) was determined assuming the protected surface area was
entirely available for microbial growth:
  Q  ðS0
SÞ
SSAA g CODm
2 d
1 ¼  
VM
SSA  100 V
m3 CH4 VCH4
Methane yield ¼ (2)
kg sCODrem Q  ðS0
SÞ
where VCH4 is the volume of methane produced (m3 d
1), Q is the
continuous flow applied (m3 d
1), S0 is the substrate concentration
in the influent and S in the effluent as kg-sCOD m
3.
Performance of suspended biomass was quantified with kinetic
tests at the end of each period in which an interval of HRT was
maintained and before reducing HRT to desirable values. Kinetic
tests were conducted by withdrawing mixed liquor from each
reactor and by transferring it to 0.5 L flasks in which as much as
5.0 ± 0.3 kg-sCOD m
3 was supplied as 98% glucose (Sigma-Aldrich
Canada Co.). The kinetic tests were 4 h long and every hour (i.e., at
0, 1, 2, 3, 4, 24 h) 5 mL sample was collected, filtered through a
0.45 mm filter for sCOD measurement. Averaged removal for each
HRT was plotted to develop a correlation curve from which a
removal rate coefficient was extrapolated. The contribution of
suspended biomass to total removal of sCOD as kg-sCOD m
3d
1 as
well as the percentage of the total sCOD removed was thus deter-
mined. Maximum removal rates observed during kinetic tests were
assumed as representative of rates occurring in the reactors at that
particular HRT. Glucose was chosen as substrate since it is a base
sugar and it can be reliably metabolized by the acetogenic micro-
organisms into acetate and therefore be available for methanogenic
microbial community in a short time period. This approach was
held since it will overestimate the suspended biomass contribution
in case any removal was observed and therefore gives conservative
(1) and consistent results.

where, Q is the continuous flow applied (m3 d
1), S0 is the substrate
concentration in the influent and S in the effluent as kg-sCOD m
3
or kg-COD m
3, SSA is the specific surface area of the AC920 equal
to 680 m2 m
3, V is the reactor volume (i.e., 0.004 m3) while VM was
the total volume of media represented as percentage of working
volume (40%).
The total gas volume was recorded every second in mL/min
using FMA 4000 OMEGA mass flow meter. The composition of gas
was determined by gas chromatography (490 Micro GC, Agilent
Technologies, Santa Clara, CA) once a week in triplicates on samples
collected in septum gas bottles. Since the flow meter registered gas
produced as nitrogen, it was necessary to convert gas produced into
actual biogas after gas chromatography analysis was conducted.
Therefore, the total gas recorded was first converted to m3 d
1 by
applying gas correction factor of K ¼ 0.721 and temperature
correction factor T ¼ 1.1 at 35  C provided by OMEGA.
Specific methane generation per kg of sCOD removed was
determined according to the following equation.
3. Results and discussion
3.1. Influent and effluent characteristics
The influent and effluent average results, reported in Table 2,
refer to the best performance obtained at 18 h HRT.
The first week of data for each interval of HRT was not consid-
ered in the calculations because it was considered an acclimation
period. Alkalinity concentration in the influent was initially sup-
plied as 1.5 ± 0.1 kg-CaCO3 m
3, but was increased to 1.7 ± 0.2 kg-
CaCO3 m
3 when the HRT was 6 h because a drop in pH to 6.5 was
observed. The pH was close to neutral values for the entire study,
and therefore non-acclimated methanogenic microorganisms
should not have been inhibited by acidic conditions during HRT
transition periods (Taconi et al., 2008). Effluent pH averaged
7.1 ± 0.1 while influent pH was 6.7 ± 0.1, demonstrating that the
buffering capacity was capable of preventing acidification. Influent
TSS and VSS were highly variable, with TSS ranging from 0.1 to

Table 2
Influent and effluent results at 18 h HRT.

Parameter Minimum Maximum Average Standard Deviation Unit

Influent pH 6.5 6.9 6.7 0.1 e

TSS 0.1 0.5 0.2 0.1 kg-TSS m
3
VSS 0.1 0.5 0.2 0.1 kg-VSS m
3
sCOD 2.5 3.4 3.0 0.2 kg-sCOD m
3
COD 2.8 6.7 4.2 1.2 kg-COD m
3
BOD5 2.1 5.0 3.5 1.4 kg-BOD5 m
3
Effluent pH 6.9 7.2 7.1 0.1 e
Alkalinity 1.4 1.6 1.5 0.1 kg-CaCO3 m
3
TSS 0.1 0.3 0.2 0.1 kg-TSS m
3
VSS 0.1 0.3 0.2 0.1 kg-VSS m
3
sCOD 0.1 0.3 0.2 0.1 kg-sCOD m
3
COD 0.3 0.8 0.5 0.1 kg-COD m
3
BOD5 0.2 0.6 0.4 0.2 kg-BOD5 m
3
1124 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127
Fig. 2. A) Soluble COD removal shown with total organic loading rate; different in-
tervals of hydraulic retention time are presented as blocks at the top. B) Surface area
loading rate correlated to soluble COD removal at different hydraulic retention time.
0.5 kg-TSS m
3 over the entire study. This was most likely due to
fluctuating presence of trub in the waste samples from the brewery.
The AMBBR was able to reliably remove significant quantities of
soluble and total COD in spite of inconsistent solids loading.
Wastewater can be treated with microscreens prior to moving bed-
biofilm reactor processes such that solids do not interfere with
removal rates (Leyva-Díaz et al., 2017). This suggests that the
AMBBR is a robust system even though its biomass comprises slow
growing microorganisms.
Percent removal of soluble and total COD decreased as OLR
increased e Fig. 2A. At an HRT of 24 h, which corresponded to an
OLR of 3.8 kg-COD m
3d
1, the AMBBR removed 93% of sCOD and
84% of COD. An 18 h HRT resulted in comparable COD removal even
though the OLR was 5.5 kg-COD m
3d
1. Performance was
observed to decrease at each following interval of HRT, with 86%
sCOD removal during the 12 h HRT.
At the shortest HRT applied, which was 6 h, sCOD removal was
the lowest at 61% even though the OLR (18.7 kg-COD m
3d
1) was
actually lower than at an 8 h HRT. The discrepancy in OLR resulted
from a less concentrated feed with lower concentrations of COD
than the previous HRT period. It was expected that the system
would not show differences in removal due to this phenomenon,
but a decrement in performance was observed. For this reason, the
OLR was maintained at values similar to the previous 8 h HRT and
the potential to recover performance to 80% sCOD removal was
evaluated.
The overall trend in performance demonstrated that 80% sCOD
removal was reliably attained until HRT decreased below 10 h. This
corresponded with SALR of 39 ± 6 g-sCOD m
2d
1. By plotting
surface area loading rates applied and sCOD removal achieved at
those loading rate, it was possible to extrapolate a correlation curve
e Fig. 2B.
A correlation coefficient of 0.6901 was found between sCOD
removal and SALR for the dataset obtained when values at 6 h HRT
where excluded. These results were neglected because of was
observed mass transfer limitations, resulting from excessive biofilm
growth overgrown the interior media channels. By using the
correlated equation in Fig. 2B, it was possible to estimate surface
area loading rates ensuring a desired % sCOD removal. Surface area
loading rates should not exceed 56 g-COD m
2d
1 to achieve 80%
removal of sCOD. At higher SALR the system's sCOD removal ca-
pabilities decrease due to substrate mass transfer limitation and
biomass accumulation in the interior media channels. However, the
AMBBR reliably removed more than 90% of sCOD with a SALR less
than 24 g-COD m
2d
1. Total COD removal rates, on the other hand,
were more variable and ranged from 59% to 93% during the study.
This was attributed to the fluctuating solids concentration in the
influent.
Studies on AMBBR processes treating industrial wastewater are
summarized in Table 3. Research on winery wastewaters (Sheli and
Moletta, 2010) demonstrated that to achieve 80% sCOD removal a
SALR of 87 g-sCOD m
2d
1 could be applied using R9 media with a
SSA of 320 m2 m
3 and 65% media filling. On the other hand, Wang
et al. (2009) investigated the capability of the same R9 media at
similar volumetric fill treating a different substrate. The authors
showed that with a SALR of 58 g-sCOD m
2 d
1, removals as high as
85% sCOD were achieved in treating milk permeate wastewater.
Treating a more complex substrate may have had a negative impact
on the R9 media. In theory, the organic carbon present in the wine
distillery wastewater has a higher biodegradability and is easier to
breakdown than the dairy wastewater. The anaerobic biodegrad-
ability rates of fat-rich wastewaters are slower because the action
of lipase and beta oxidation required for utilization of fats is a
slower process than carbohydrate utilization (Vidal et al., 2000). In
addition, the higher content of particulate matter in the dairy
wastewater may have also limited mass transfer for the R9 media
(Demirel et al., 2005).
Brewery wastewater composition is more comparable to winery
than dairy wastewater since both brewery and winery processes
have similar production methods (i.e., fermentation), utilize yeast,
and have low fat quantities. The yeast is the highest source of COD
in these wastewaters. By comparing the study on the AC920
manufactured by Headworks Bio with the R9, different SALR values
were obtained for similar percentage of sCOD removal. Particularly
at 85% sCOD removal, the AC920 media were able to achieve less
than half the SALR of the R30. Similarly, a difference of 30 g-sCOD
m
2d
1 in SALR was observed for 80% sCOD removal. It was ex-
pected that the SALR would be greater for the AC920 media
compared to the others presented because of its physical proper-
ties. Hence, the higher SSA of the media used in this study (i.e.,
680 m2 m
3) was expected to reach a greater loading rate than the
one obtained for 320 m2 m
3. However, the AC920 removal prop-
erty was limited because the cubical shape of the media and small
spacing within the protected specific surface area (i.e., 4 mm).
Hence, mass transfer limitation was found to be the cause of un-
expectedly decreased performance at 6 h HRT due to both the
shape of the media and the mixing properties linked with the
cubical feature in a small bench scale reactor. Increasing the mixing,
and therefore the shear, had little impact on the mass transfer
limitation observed.
3.2. Biomass activity
Biomass activity was quantified using specific surface area
 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127 1125

Table 3
Review of AMBBR processes treating industrial wastewater.

Wastewater treated Carrier sCOD rem (%) SALR (g-sCOD m
2 d
1) Reference

Type SSA (m2 m
3)

Wine distillery R30a 320 85 87 Sheli and Moletta (2010)

R9a 530 80 85 Sheli et al. (2014)
Dairy R9a 530 85 58 Wang et al. (2009)
Brewery AC920 680 70 87 This study
80 56
85 40
90 24
a
Bioflow manufactured by Raushert Co.

activitye Equation (1). For the most part, SSAA increased with OLR
as the HRT decreased e Fig. 3A. At an HRT of 24 h, the SSAA was
determined to be 11 ± 1 g-COD m
2d
1, while at an 18 h HRT the
SSAA was 13 ± 1 g-sCOD m
2d
1.
When the HRT decreased to 12 h, the SSAA was calculated as
24 ± 1 g-sCOD m
2d
1. The highest SSAA, 44 ± 2 g-sCOD m
2 d
1,
was observed at a HRT of 10 h. The specific surface area activity
decreased to 32 ± 1 g-sCOD m
2d
1 when the HRT was changed to
6 h. The specific surface area activity quantified for total COD fol-
lowed similar trends compared to those calculated based on sCOD,
although with higher fluctuation. This was expected since the
observed influent solids varied significantly. Limitations in mass
transfer resulted in upset, most likely caused by excessive loading
since the biofilm had outgrown the protected surface area provided
by the media. By plotting SSAA and SALR, the limitation that
occurred at a 6 h HRT is clearly defined e Fig. 3B.
A strong correlation (R2 ¼ 0.9859) resulted from the data ob-
tained between 24 h and 8 h HRT suggesting that with AC920 media
an SALR as high as 60 g-sCOD m
2d
1 can be achieved. However,
once mass transfer limitations had developed decreasing the SALR
had not produced any positive impact. Under the presence of mass
transfer limitations even a SALR of 45 g-sCOD m
2d
1 experienced
less than optimal surface area activity. Therefore, achieving similar
SALR at a lower HRT could be addressed by implementing higher
percentage of AC920 media filling.
Kinetic tests were performed on the reactor mixed liquor before
each change in HRT to estimate the contribution of suspended
biomass to the overall performance e Fig. 4.
At a 24 h HRT, the contribution of suspended biomass was
approximately 4% of the total COD removed, which suggested that
activity was almost completely carried out by the established bio-
film. A positive correlation was observed between the suspended
biomass activity and the applied OLR as the HRT decreased from 24
to 10 h. Estimations of the suspended biomass contribution
increased from approximately 4% at 24 h to 13%, 14% and 19% at
18 h, 12 h, and 10 h HRT. The contribution of suspended biomass
was observed to decrease between 10 and 6 h HRT. Suspended
biomass contribution was estimated to be 13% and 2% at 8 h and 6 h
HRT.
With decreasing HRT, and thus increasing OLR and SALR, the
biofilm experienced increased sloughing resulting in higher con-
centrations of active suspended solids. This explains the increase in
suspended biomass activity from 24 to 10 h HRT, but not the
decrease observed in suspended activity at 8 h and 6 h HRT. At HRT
of less than 10 h the SRT of suspended biomass may have been too
low for maintaining an active microbial population. However, at

Fig. 3. A) Specific surface area activity measured on soluble COD at different hydraulic
retention time; B) Specific surface area activity correlated to surface area loading rate Fig. 4. Total COD removal compared with removal by detached suspended biomass
at different hydraulic retention time. alone, at different hydraulic retention time.
1126 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127
Fig. 5. Organic loading rate and specific surface area activity obtained at different hydraulic retention time. Percentages represent the load of COD removed relative to the oncoming
load.
lower HRT and therefore higher OLR the increased biofilm pro-
duction would also augment the suspended solids concentrations
at low suspended SRT.
Specific surface area activity (Equation (1)) increased as ex-
pected with the changes in loading rates at different HRT e Fig. 5.
However, the SSAA decreased significantly between 8 and 6 h HRT,
which supports the previous hypothesis that mass transfer limita-
tions were brought about by biofilm outgrowing the available
protected surface area. The AC920 media is cubic in shape and
divided into 9 square channels within which the biofilm can
develop. Each square channel was approximately 4 mm by 4 mm,
and thus the biofilm thickness should never reach 2 mm.
3.3. Biogas production and characterization
Biogas production and composition were monitored for the
entire study. The biogas composition remained relatively consistent
for the entire study, with 71 ± 2% methane and 26 ± 4% carbon
dioxide. Percent carbon dioxide was observed to fluctuate slightly
more than percent methane. Calculations for unit methane gener-
ation (Equation (2)) were performed using the OLR, biogas pro-
duction, and percent methane. Unit methane generation was more
stable when sCOD was considered, which was most likely due to
the high bioavailability of soluble compounds in the influent. A
maximum methane yield of 0.34 ± 0.06 CH4 kg-sCOD
1 was
observed at an OLR of 5.4 kg-sCOD m
3d
1 and HRT of 18 h;
approaching the theoretical value of 0.4 m3-CH4 kg-sCOD
1 at 35  C
(Viana et al., 2012).
4. Conclusion
Anaerobic moving bed biofilm reactors were demonstrated to
be a viable option for the treatment of brewery wastewater at 35  C.
Performance was evaluated by decreasing the HRT, and therefore
increasing the OLR. Volumetric removal rates of 3.0 ± 0.2, 4.2 ± 0.3,
7.2 ± 0.6, 12.4 ± 0.8, 13.4 ± 1.0, and 12.0 ± 1.1 g-sCOD d
1 were
achieved at 24, 18, 12, 10, 8, and 6 h hydraulic retention times.
Hydraulic retention time of 18 h showed the highest chemical ox-
ygen demand removal (88 ± 2.5%) and methane yield at
0.34 ± 0.06 m3-CH4 kg-sCOD
1. Based on SSAA alone, it can be
concluded that a SALR of 54 g-COD m
2d
1 should not be exceeded
for the AC920 media. Significant biogas production was attained,
with unit methane generations observed to be as high as
0.34 ± 0.06 m3-CH4 kg-sCOD
1 at 35  C and 18 h HRT. Mass transfer
limitations were observed above this loading due to biofilm
outgrowing the available protected surface area. However, sus-
pended biomass kinetics suggested that mass transfer may have
been limiting even at the 8 h HRT, and thus the SALR should not
exceed 44 g-COD m
2d
1 in order to promote sufficient mass
transfer.
Acknowledgement
The research was supported in kind by the Fort Garry Brewery in
Winnipeg, Canada. Special thanks to brewmasters Dan Geddes and
Matt Wolff. Headworks BIO, Canada, is acknowledged for providing
the media.
References
APHA, 2012. Standard Methods - Examination of Water and Wastewater. American
Public Health Association (American Water Works Association, Water Envi-
ronment Federation).
Brito, A.G., Peixoto, J., Oliveira, J.M., Oliveira, J.A., Costa, C., Nogueira, R.,
Rodrigues, A., 2007. Brewery and winery wastewater treatment: some focal
points of design and operation. In: Utilization of By-products and Treatment of
Waste in the Food Industry.
Castillo, A., Vall, P., Garrido-Baserba, M., Comas, J., Poch, M., 2017. Selection of in-
dustrial (food, drink and milk sector) wastewater treatment technologies: a
multi-criteria assessment. J. Clean. Prod. 143, 180e190. https://doi.org/10.1016/
j.jclepro.2016.12.132.
Chastain, C., Crandall, S., Del Grande, D., Flores, T., Gilliland, M., Horwitz, L., Opela, C.,
Swersey, C., Utz, M., 2011. Water and Wastewater: Treatment/Volume Reduction
Manual 1e47.
Demirel, B., Yenigun, O., Onay, T.T., 2005. Anaerobic treatment of dairy wastewaters:
a review. Process Biochem. 40, 2583e2595. https://doi.org/10.1016/
j.procbio.2004.12.015.
di Biase, A., Devlin, T., Oleszkiewicz, J., 2016. Start-up of an anaerobic moving
bedebiofilm reactor and transition to brewery wastewater treatment.
J. Environ. Eng. 142 (12).
di Biase, A., Devlin, T.R., Kowalski, M.S., Oleszkiewicz, J.A., 2017. Performance and
design considerations for an anaerobic moving bed biofilm reactor treating
brewery wastewater: impact of surface area loading rate and temperature.
J. Environ. Manag. 1e7. https://doi.org/10.1016/j.jenvman.2017.05.093.
Leyva-Díaz, J.C., Martín-Pascual, J., Poyatos, J.M., 2017. Moving bed biofilm reactor to
treat wastewater. Int. J. Environ. Sci. Technol. 14, 881e910. https://doi.org/
10.1007/s13762-016-1169-y.
McQuarrie, J.P., Boltz, J.P., 2011. Moving bed biofilm reactor technology: process
applications, design, and performance. Water Environ. Res. 83, 560e575.
https://doi.org/10.2175/106143010X12851009156286.
Ødegaard, H., 2016. A road-map for energy-neutral wastewater treatment plants of
 A. di Biase et al. / Journal of Cleaner Production 172 (2018) 1121e1127
the future based on compact technologies (including MBBR). Front. Environ. Sci.
Eng. 10 https://doi.org/10.1007/s11783-016-0835-0.
Olajire, A.A., 2012. The brewing industry and environmental challenges. J. Clean.
Prod. 1e21. https://doi.org/10.1016/j.jclepro.2012.03.003.
Raudkivi, M., Zekker, I., Rikmann, E., Vabam€ ae, P., Kroon, K., Tenno, T., 2017. Nitrite
inhibition and limitation - the effect of nitrite spiking on anammox biofilm,
suspended and granular biomass. Water Sci. Technol. 75, 313e321. https://
doi.org/10.2166/wst.2016.456.
Rikmann, E., Zekker, I., Tenno, T., Saluste, A., Tenno, T., 2017. Inoculum-free start-up
of biofilm- and sludge-based deammonification systems in pilot scale. Int. J.
Environ. Sci. Technol. https://doi.org/10.1007/s13762-017-1374-3.
Sheli, C., Guo, J., Yuan, C., Jing, C., Lina, G., 2014. Anaerobic treatment of winery
wastewater in moving bed biofilm reactors. Desalin. Water Treat. 52,
1841e1849. https://doi.org/10.1007/s00449-009-0387-9.
Sheli, C., Moletta, R., 2010. Treatment of wine distillery wastewater using an
anaerobic moving bed biofilm reactor with low density of polyethylene sup-
port. In: Int. Conf. Challenges Environ. Sci. Comput. Eng. CESCE 2010, vol. 2,
pp. 232e235. https://doi.org/10.1109/CESCE.2010.229.
Sheli, C., Moletta, R., 2007. Anaerobic treatment of vinasses by a sequentially mixed
moving bed biofilm reactor. Water Sci. Technol. 56, 1. https://doi.org/10.2166/
wst.2007.465.
Taconi, K.A., Zappi, M.E., Todd French, W., Brown, L.R., 2008. Methanogenesis under
acidic pH conditions in a semi-continuous reactor system. Bioresour. Technol.
99, 8075e8081. https://doi.org/10.1016/j.biortech.2008.03.068.
1127
Tenno, T., Rikmann, E., Zekker, I., Tenno, T., Daija, L., Mashirin, A., 2016. Modelling of
equilibrium distribution of carbonaceous ions and molecules in a heteroge-
neous system of CaCO3ewater e gas. Proc. Est. Acad. Sci. 65 (1), 68e77. https://
doi.org/10.3176/proc.2016.1.07.
Viana, M.B., Freitas, A.V., Leit~
ao, R.C., Pinto, G.A.S., Santaella, S.T., 2012. Anaerobic
digestion of crude glycerol: a review. Environ. Technol. Rev. 1, 81e92. https://
doi.org/10.1080/09593330.2012.692723.
ndez, R., Lema, J.M., 2000. Influence of the content in fats
Vidal, G., Carvalho, A., Me
and proteins on the anaerobic biodegradability of dairy wastewaters. Bioresour.
Technol. 74, 231e239. https://doi.org/10.1016/S0960-8524(00)00015-8.
Wang, S., Rao, N.C., Qiu, R., Moletta, R., 2009. Performance and kinetic evaluation of
anaerobic moving bed biofilm reactor for treating milk permeate from dairy
industry. Bioresour. Technol. 100, 5641e5647. https://doi.org/10.1016/
j.biortech.2009.06.028.
Zekker, I., Rikmann, E., Kroon, K., Mandel, A., Mihkelson, J., Tenno, T., Tenno, T., 2017.
Ameliorating nitrite inhibition in a low-temperature nitritationeanammox
MBBR using bacterial intermediate nitric oxide. Int. J. Environ. Sci. Technol. 14,
2343e2356. https://doi.org/10.1007/s13762-017-1321-3.
Zekker, I., Rikmann, E., Mandel, A., Kroon, K., Seiman, A., Mihkelson, J., Tenno, T.,
Tenno, T., 2016. Step-wise temperature decreasing cultivates a biofilm with high
nitrogen removal rates at 9 C in short-term anammox biofilm tests. Environ.
Technol. 37, 1933e1946. https://doi.org/10.1080/09593330.2015.1135995
(United Kingdom).