Thing Teen818 AnMBBR Reactor 2024

Abstract: An anaerobic continuously stirred tank reactor (CSTR) integrated with an external tubular nanofiltration membrane module was
operated for 466 days. Studies were conducted to compare the efficiencies in flux and operational permeability recovery of two different
chemical cleaning processes. Chemically enhanced backwashes (CEBs) were applied for approximately 176 days of the study and their
influence on chemical cleaning performance and membrane fouling rate following chemical cleanings were also investigated. The anaerobic
nanofiltration membrane bioreactor used in the study was fed a high-strength, potato starch–based synthetic wastewater (average chemical
oxygen demand of 83.4 g=L and average total suspended solids of 5.8 g=L). Important findings were that a static, one-phase cleaning process
using caustic cleaning chemicals outperformed a dynamic, three-phase cleaning process that also incorporated acidic cleaning chemicals; the
acidic phase of the three-phase cleaning had a detrimental effect on the overall cleaning procedure’s performance; CEBs enhanced the per-
formance of cleaning processes; and, finally, caustic cleaning solutions provided benefits to both membrane cleaning effectiveness and rate of
fouling in the short term but may lead to an increased rate of membrane fouling in the long term. DOI: 10.1061/(ASCE)EE.1943-
7870.0001524. © 2019 American Society of Civil Engineers.
Author keywords: Anaerobic membrane bioreactor; Chemical cleaning; Chemically enhanced backwash; Membrane fouling;
Nanofiltration membrane.
Introduction anaerobic systems such as high concentration of biomass retention

(as the effectiveness of biomass retention becomes independent of
With an ever-growing population, sustainable water and waste- sludge settleability due to membranes’ near-complete solid–liquid
water treatment technologies will become essential in solving water separation capabilities), compact design, and, most importantly, the
scarcity problems prone to develop in upcoming years. Anaerobic ability to produce a superior quality effluent for possible recycle
membrane bioreactors (AnMBRs) offer a promising means of or high-quality discharge to meet stringent wastewater disposal
sustainable and value-added wastewater treatment, coalescing the regulations (Skouteris et al. 2012).
benefits of anaerobic digestion for the treatment of municipal and What continues to remain the most significant factor limiting the
industrial wastewaters with the almost 100% solid–liquid separa- cost effectiveness of membranes in wastewater treatment applica-
tion potential of membrane integrated wastewater treatment tech- tions (both municipal and industrial) is membrane fouling (Ramos
nologies. Anaerobic digestion for wastewater treatment provides et al. 2014). Membrane fouling may be defined as the combination
certain benefits over more commonly used conventional aerobic of processes leading to flux deterioration and transmembrane pres-
technologies, such as high-rate suspended-growth systems, very sure (TMP) increase due to surface or internal pore blockage of the
high removal efficiencies of biochemical oxygen demand (BOD) membrane (Lousada-Ferreira et al. 2014). Surface or internal pore
and chemical oxygen demand (COD), resource recovery (value- blockage is generated through the contact and deposition of sus-
added products such as methane in its biogas and the possible pended particles, colloids, and solutes present in the mixed liquor
recovery of select biochemicals such as volatile fatty acids in its on the membrane surface or in its pores, henceforth referred to as
mixed-liquor and effluent), lower sludge yield, and low carbon foulants (Iorhemen et al. 2016). Membrane fouling increases TMP
footprint (Kinnunen et al. 2015; Meng et al. 2009; Ozgun et al. and decreases flux, while the control of foulant deposition increases
2013; Pilli et al. 2016; Skouteris et al. 2012). Through the integra- manpower and energy requirements, demands costly chemical
tion of membrane separation technologies within anaerobic di- cleaning (that further requires waste handling), and may also
gestion, further benefits are provided over non-membrane-based result in increased deterioration of the membrane material, affecting
its lifespan (Ramos et al. 2014). Foulant mitigation for a membrane
1
Graduate Student, Dept. of Civil Engineering, Univ. of New module is a two-pronged approach: first, the membrane-fouling
Brunswick, 17 Dineen Dr., Fredericton, NB, Canada E3B 5A3 (corre- rate should be minimized, and second, technologies to restore the
sponding author). Email: h557d@unb.ca flux of a fouled membrane must be implemented (Liao et al. 2006).
2
Professor, Dept. of Civil Engineering, Univ. of New Brunswick, 17 Physical cleaning processes (e.g., gas scouring, backwashing,
Dineen Dr., Fredericton, NB, Canada E3B 5A3. Email: singhk@unb.ca relaxation) have only proven to be effective at removing some pore-
3
Process Engineer, ADI Systems, Inc., 370 Wilsey Rd., Fredericton,
clogging and cake-layer membrane foulants (Liao et al. 2006).
NB, Canada E3B 6E9. Email: gustavo.zanatta@evoqua.com
Note. This manuscript was submitted on July 18, 2018; approved on Chemical cleanings, using low- to moderate-strength corrosive or
October 26, 2018; published online on March 6, 2019. Discussion period caustic solutions, are highly effective at removing both the gel layer
open until August 6, 2019; separate discussions must be submitted and cake layer developed through solids deposition on the mem-
for individual papers. This paper is part of the Journal of Environmental brane surface and pore-blocking foulants, making them typically
Engineering, © ASCE, ISSN 0733-9372. more efficient at recovering membrane flux. These chemical
© ASCE 04019018-1 J. Environ. Eng.
J. Environ. Eng., 2019, 145(5): 04019018

cleanings can be performed either as a preventive maintenance op- are, to the authors’ best knowledge, no studies investigating the
eration to maintain permeate flux [through incorporating chemical impact of daily CEBs on the effectiveness of chemical cleaning
cleanings into regular physical cleaning processes, for example processes.
chemically enhanced backwashes (CEBs)], or as a recovery oper-
ation to recover membrane flux from a severely fouled membrane
module (Ramos et al. 2014). Chemically enhanced backwashes Materials and Methods
combine conventional physical backwashing with chemical clean-
ing in one process, which could be of high importance for the
Reactor Configuration
design and operation of membrane bioreactor (MBR) systems
for wastewater treatment. Recent studies using CEBs have typically A 1,000 L stainless steel anaerobic CSTR was equipped with an
been performed at long intervals (weekly or monthly cleaning proc- external nanofiltration tubular membrane module and operated in
esses) (Zhou et al. 2014). There have been few comparative studies cross-flow mode. A simplified process flow diagram (PFD) of the
on different cleaning methods and conditions, and scant informa- configuration may be viewed in Fig. 1. The anaerobic CSTR was
tion exists on CEBs, in general, in the literature (Ramos et al. 2014; equipped with a heat wrap, which maintained the reactor temper-
Zhou et al. 2014). Snowdon et al. (2018) observed that one 60 s ature at an average value of 36.2°C (0.5°C) throughout the dura-
CEB every 24 h resulted in increased permeate flux, operational tion of the study. The operational temperature of approximately
permeability, and net flux, demonstrating CEBs’ effectiveness as 36°C was selected since it is the upper limit of mesophilic anaerobic
a preventive maintenance operation. digestion (25°C–35°C) and maximized biogas generation, which
This research project compared the efficiency of two different was desired (Metcalf & Eddy, Inc. 1991).
chemical cleaning processes applied to one membrane module Reactor-mixed liquor was continuously recirculated from the
integrated within a continuously operated pilot-scale AnMBR sys- CSTR to the membrane module and back into the CSTR by a pro-
tem. The two chemical cleaning processes were applied to the gressive cavity pump (Nemo, New Brunswick, Canada). This pump
membrane module only once it was severely fouled, acting as a maintained a mixed-liquor flow rate of 6,000 L=h, which resulted
recovery operation. As a secondary objective, the project investi- in the targeted membrane surface cross-flow velocity of 2.5 m=s
gated the impact of daily CEBs on the effectiveness of each and a complete recirculation of reactor contents every 10 min.
chemical cleaning process. The novel pilot-scale AnMBR system A cross-flow velocity of 2.5 m=s was selected since it is deemed
consisted of a 1,000 L anaerobic continuously stirred tank reactor to be the minimum operational cross-flow velocity that must be
(CSTR) integrated with an external nanofiltration tubular mem- maintained for the membrane module based on manufacturer rec-
brane module. The reactor and membrane configuration provided ommendations. It was desired to operate at the lower spectrum of
benefits over conventional cleaning processes in that the membrane the operational cross-flow velocity to minimize energy demands.
module did not need to be removed for chemical cleanings (since it The membrane module was cylindrical in shape, composed of poly-
could be isolated from the CSTR through a recirculation loop). This vinylidene fluoride (PVDF), with an outer diameter of 50 mm and
resulted in minimal downtime and maintenance work required for a total length of 1.436 m. The module contained 13 tubular mem-
chemical cleaning processes. With few direct comparative studies brane channels, each 8 mm in diameter, which resulted in a total
on different chemical cleanings applied to the same membrane membrane area of 0.42 m2 . The membrane surfaces had a mean
module, and little information in the literature on CEBs, there pore size of 30 nm.
Fig. 1. Simplified PFD of experimental setup.
J. Environ. Eng., 2019, 145(5): 04019018

Table 1. Chemical cleaning conditions
Cleaning Recirculation Soak time Number of backwashes and
process Cycle time (min) (min) duration (number and min) Chemical used
1 1 30 N/A 2 and 2 min 1% sodium hydroxide
2 N/A 20 N/A
3 30 N/A 2 and 2 min
4 30 N/A 2 and 2 min 1% citric acid
5 N/A 20 N/A
7 30 N/A 2 and 2 min 1% sodium hydroxide and sodium hypochlorite
8 N/A 20 N/A
2 1 30 N/A N/A 1% sodium hydroxide and sodium hypochlorite
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 20 N/A
5 N/A 20 N/A
8 N/A 20 N/A
chemical cleaning process. Finally, the chemicals used outline the module were also measured. These values allowed the MLF
exact concentration of chemical cleaning solutions applied to the and MLOP to be calculated. The mixed-liquor recoveries were
membrane module during each cleaning condition. calculated by dividing the post–chemical cleaning MLF or post–
Clean water flux was measured by isolating the CSTR recircu- chemical cleaning MLOP values by the initial MLF or initial
lation line from the CSTR itself, allowing a 50-L tank to be filled MLOP values measured on Day 2 of operation, and then multiply-
with clean water (maintained at around 30°C) and filtered through ing by 100. Equations used to calculate clean water and mixed-
the membrane module. Each clean water test recirculated the clean liquor recoveries are displayed in Eqs. (2) and (3), respectively:
water at five flow rates in increasing order (790, 1,580, 2,360,
3,150, and 3,940 L=h), and the resultant membrane fluxes and J; OPPC
TMPs were measured. Five different increasing flow rates were se- CW J;OP ð%Þ ¼ × 100 ð2Þ
J; OPo
lected to provide incremental increases in flux and TMP, reaching
the maximum flow rate of the recirculation loop at approximately
3,940 L=h. Operational permeability is defined as membrane flux
divided by instantaneous operational TMP, in other words pressure- J; OPPC
MLJ;OP ð%Þ ¼ × 100 ð3Þ
averaged flux, and may be defined by Eq. (1) (Hamden de Andrade J; OPo
et al. 2013):
Ji In Eq. (2), CW J;OP is the clean water flux or operational per-
OPi ¼ ð1Þ
ΔPi meability recovery (%), J; OPPC is the post–membrane cleaning
clean water membrane flux or clean water operational permeability
In Eq. (1), OPi is the instantaneous operational permeability (L=m2 =h, L=m2 =h=bar), and J, OPo is the initial clean water mem-
(L=m2 =h=bar), J i is the instantaneous membrane flux (L=m2 =h), brane flux or clean water operational permeability (L=m2 =h,
and ΔPi is the instantaneous transmembrane pressure (bar). Opera- L=m2 =h=bar). In Eq. (3), MLJ;OP is the mixed-liquor flux or opera-
tional permeability was considered so that TMP could be factored tional permeability recovery (%), J; OPPC is the post–membrane
into the flux recovery analysis. Clean water recoveries were calcu- cleaning mixed-liquor membrane flux or mixed-liquor operational
lated by dividing the post–chemical cleaning clean water flux or permeability (L=m2 =h, L=m2 =h=bar), and J; OPo is the initial
post–chemical cleaning clean water OP by the initial clean water mixed-liquor membrane flux or mixed-liquor operational per-
flux or initial OP values measured on Day 1 of operation, and then meability (L=m2 =h, L=m2 =h=bar).
multiplying by 100. Once the membrane module was placed back Membrane resistance coefficients were calculated using the
inline for mixed-liquor recirculation, and after 24 h of recirculation resistance-in-series model [Eq. (4)]. This equation was adopted and
at a flow rate of 6,000 L=h, the flux and TMP of the membrane developed from other studies (Ruigomez et al. 2017):
J. Environ. Eng., 2019, 145(5): 04019018

Table 2. Membrane resistance coefficients calculated following each would allow the CEBs to be quite efficient in both removing
chemical membrane cleaning process (unit: 10−12 × m−1 ) built-up membrane foulants and preventing their rate of accu-
Cleaning process KM KR K IR mulation while operating. Zhang et al. (2007) found that, when
operating a tubular polyethersulfone side stream AnMBR for the
1 0.62 4.66 0.73
2 9.33 2.00 treatment of swine manure, long-term operation without chemical
3 17.3 2.60 cleaning processes resulted in residual membrane fouling that was
4 18.7 1.45 difficult to remove with the chemicals used in their study (nitric
5 6.27 1.32 acid and ethylenediaminetetraacetic acid). By using a constant
6 6.34 3.72 chemical cleaning process, the build-up of residual membrane
foulants might have been prevented. In this study, the prevention
of the build-up of difficult-to-remove membrane foulants was ob-
served, and a constant source of chemical cleaning was supplied
first placed inline. Reversible membrane fouling occurred rapidly through CEBs, resulting in more effective fouling mitigation. It was
when the membrane module was first placed inline, resulting in a found that K R increased by 85% over 106 days between Cleaning
quick drop in membrane flux (a loss of 35% in MLF over 5 days), Process 2 and Cleaning Process 3, whereas K R increased by only
which merited the first chemical cleaning. The membrane resis- 8% over 121 days between Cleaning Process 3 and Cleaning
tance coefficients were calculated (Table 2) to further support this Process 4, indicating a decrease in the rate of K R accumulation.
observation. Following Cleaning Process 1, K R was found to be With the cleaning conditions of Cleaning Processes 2 and 3 being
much greater than K IR (values of 4.66 × 10−12 m−1 and 0.73 × identical, and operating conditions being fairly consistent in the
10−12 m−1 , respectively), indicating that the majority of foulant periods following these cleaning processes, the primary reason for
removed was reversible membrane fouling. Since this was the first the decline in reversible foulant accumulation may be attributed to
chemical cleaning applied to the membrane module, K IR was also CEBs being in place 60% of the days following Cleaning Process 3.
at its lowest value throughout the study. As the majority of mem- K R was 6.27 × 10−12 m−1 following Cleaning Process 5. A de-
brane fouling was contributed from reversible foulants, which are crease in this value from Cleaning Process 4 to 5 indicates that the
removed effectively through chemical cleaning processes, and there rate of reversible foulant build-up was greatly reduced during
was little contribution from irreversible membrane foulants relative this period of operation, which may again be attributed to regular
to other cleaning methods, Cleaning Process 1 appeared much more CEBs being in place.
effective. With this consideration in mind, it is difficult to distin- The irreversible fouling resistance coefficient typically increases
guish whether Cleaning Process 1 was superior in terms of clean over the duration of membrane filtration and would not be antici-
water flux recovery over all other cleaning processes due to its pated to decrease unless operational or cleaning conditions could
cleaning conditions or whether its enhanced performance was sim- be enhanced significantly. Considering that K IR decreased from
ply due to occurring so soon after the membrane was first placed Cleaning Process 3 to 4, and again from Cleaning Process 4 to 5,
inline. Wang et al. (2017) did find that the maximum amount while all of these cleaning processes operated under the same
of organics in their chemical cleanings was desorbed through the conditions, indicates that the presence of CEBs might also en-
use of an alkaline and acidic cleaning process when treating MBR hance regular cleaning protocols in such a way that irreversible
effluent. Cleaning Process 1 results agree with this observation, as foulants (that could not be removed without CEBs in place) could
the cleaning conditions provided both alkaline and acidic chemicals then be removed. Zhang et al. (2007) concluded that weekly mem-
and it had the largest clean water flux recovery, indicating more brane cleanings provided no significant benefits in membrane per-
foulant removal. formance compared to monthly membrane cleanings, attributing
Cleaning Processes 4 and 5 outperformed Cleaning Processes 2 the lack of benefits to the fact that irreversible fouling accumulated
and 3 in terms of clean water flux recovery. This result was not more rapidly than the frequency of either cleaning. Perhaps the
initially anticipated because, due to the accumulation of irreversible daily CEBs in this study occurred frequently enough that irre-
membrane fouling over the course of membrane operation, succes- versible foulants were unable to accumulate effectively between
sively applying the same cleaning conditions typically results in membrane cleanings. From Cleaning Process 5 to 6 the K R value
a decrease in flux and OP recoveries with each successive clean remained quite steady between cleans, whereas K IR increased by a
(Sun et al. 2016). Since all four of these cleaning processes were factor of 2.8 with CEBs regularly in place for 26 days of operation.
identical in conditions, it was expected that irreversible membrane This indicates that Cleaning Process 5 may have been superior to
foulants would accumulate, making each successive membrane Cleaning Process 6 since K IR increased once again between cleans.
cleaning process less efficient than the one preceding it. A differ- Similar to the results for clean water flux recovery, Cleaning
ence that was present leading into Cleaning Processes 4 and 5 was Processes 1, 5, and 4 offered the largest clean water OP recoveries
that there were CEBs present 60% (73 days) and 100% (77 days) of with values of 49%, 31%, and 30%, respectively. Since the recov-
the time, respectively, leading into each of these chemical cleaning eries in clean water operational permeability are near identical to
processes. Cleaning Process 5 had more CEBs prior to it and out- the recoveries in clean water flux, this indicates that the chemical
performed Cleaning Process 4. Similarly, Cleaning Process 4 had cleanings had similar recoveries in TMP. One reason why Clean 1
more CEBs prior to it compared to Cleaning Processes 2 and 3 (as outperformed all of the others by a large margin may again be
there were no CEBs in place leading into Cleans 2 and 3) and out- attributed to its early application and lack of irreversible foulant
performed them. From this observation, it would appear that regu- accumulation.
lar CEBs synergize with chemical cleaning processes and enhance In terms of MLF and MLOP recoveries, all cleaning processes
their performance in terms of clean water flux recovery. Regular exhibited good performance, with the lowest MLF recovery being
contact between a low-strength caustic chemical cleaning solution Cleaning Process 3 (with a value of 67%) and the lowest MLOP
known for its efficiency in removing organic membrane foulants recovery being Cleaning Process 2 (62%). Compared to the largest
and a wastewater/mixed liquor composed almost entirely of organic clean water flux and OP recoveries of 53% and 49%, respectively,
matter as a source of membrane foulants (with average volatilities this indicates that the chemical cleanings were much more effective
for the wastewater and mixed liquor of 99% and 94%, respectively) at recovering inline mixed-liquor flux rather than clean water flux.
J. Environ. Eng., 2019, 145(5): 04019018

Cleaning Processes 1, 4, and 5 offered the largest MLF recoveries Rate of Operational Permeability Decline
with values of 100%, 84%, and 84%, respectively. Similarly, Clean-
As a final means of comparison between the membrane cleanings
ing Processes 1, 4, and 5 offered the largest MLOP recoveries with and the influence of frequent CEBs on their performance, the rate of
values of 100%, 85%, and 83%, respectively. OP decline and standardized duration of operation following each
The first chemical cleaning process appears to be superior to chemical cleaning was calculated (excluding Cleaning Process 6
chemical Cleaning Processes 2–5 until both its early occurrence because membrane operation was ceased following this cleaning
and the results from Cleaning Process 6 are factored into the analy- process). The standardized duration of operation was found by first
sis. Cleaning Process 6 offered a clean water flux recovery and a calculating the rate of OP decline following each membrane clean-
clean water OP recovery of only 14%. These recoveries were ing process and then projecting it to a standardized OP low set point
achieved after 26 days of operation following Cleaning Process 5 of 10 L=m2 =h=bar. This set point was chosen since it was the low-
(compared to 121 and 77 days of operation leading into Cleaning est OP value prior to any chemical cleaning. By standardizing the
Processes 4 and 5, respectively), with CEBs in place daily. This other durations of operation to the set point, direct comparisons
short time interval between cleaning processes provided less time could be made between their standardized values. The values for
for irreversible membrane foulant accumulation, especially with the rate of operational permeability decline, duration of operation,
CEBs regularly in place. However, following Cleaning Process 6, standardized duration of operation, and days of CEB following
the contribution of irreversible membrane fouling to the overall each chemical cleaning may be found in Table 4. The average op-
membrane resistance increased from 1.32 × 10−12 m−1 following erating conditions associated with each of these periods are also
Cleaning Process 5 to 3.72 × 10−12 m−1 (a 2.8 times increase). provided (Table 5). Snowdon et al. (2018) provided more detailed
Cleaning Process 6 was not only inefficient in clean water flux reactor performance data from Days 113 to 277 of operation. The
and OP recovery, it also led to the largest increase in K IR between operating conditions displayed are for select parameters known to
cleans while also having the shortest period of operation prior to it. affect the fouling propensity of a membrane (Zhang et al. 2006).
The K R value following Cleaning Process 6 was similar to the K R Operational permeability decline was highest following the
value following Cleaning Process 5 (values of 6.34 × 10−12 m−1 first membrane cleaning process, with a rate of decline of
and 6.27 × 10−12 m−1 , respectively), indicating that both processes 0.18 L=m2 =h=bar=day. Following Cleaning Process 2 was the
were nearly equivalent in removing reversible foulants; the large period of operation that had the slowest decline in OP, with a rate
discrepancy lies in the irreversible foulants. Since preventing irre- of decline of 0.02 L=m2 =h=bar=day. Comparison of the rate of
versible foulant accumulation could lead to a longer membrane decline in operational permeability following Cleaning Processes 1
lifespan, Cleaning Processes 2–5 appear to be superior to Cleaning and 2 revealed that operational permeability decreased at an 89%
Processes 1 and 6. When one also factors in that Cleaning Process 1 slower rate following Cleaning Process 2. The operating conditions
occurred only 5 days after the new membrane module was placed in terms of MLSS, organic loading rate (OLR), and sludge retention
inline, and that this could be the primary reason for its initial en- time (SRT) were similar between these two periods of operation,
hanced performance, it appears that the cleaning conditions applied with SRT varying by the largest amount and being 25% larger dur-
in Cleaning Processes 2–5 are the superior option. ing the period following Cleaning Process 2. The impact of SRT on
the membrane fouling rate has been debated, with too large of a
Performance for Each Part of Clean 6
Clean water flux, clean water OP, and K IR values were calculated Table 4. Rate of operational permeability decline, duration of operation,
(Table 3) following each part of Cleaning Process 6 to compare standardized duration of operation, and days of CEB being present fol-
the effectiveness of each part of the clean in removing membrane lowing each chemical cleaning process
foulants. It was found that the citric acid portion of the cleaning, Duration of Standardized Time CEBs
Part 2, had a detrimental effect to the overall cleaning performance. Cleaning rOP operation duration of present
Clean water flux recovery and clean water OP recovery decreased, process (L=m2 =h=bar=day) (day) operation (day) (day)
and the irreversible fouling resistance coefficient increased follow- 1 −0.18 81 81 0
ing Part 2 of Cleaning Process 6. Part 3 of the clean was able to 2 −0.02 106 264 0
recover performance in each of these parameters; however, its ef- 3 −0.03 121 305 73
fectiveness may have been hindered from Part 2 taking place prior 4 −0.04 77 276 77
to it. The citric acid portion of the clean may have lagged in the 5 −0.12 26 75 26
three performance parameters since citric acid is primarily used Note: CEB = chemically enhanced backwashes.
as a means of inorganic foulant removal and the wastewater and
mixed liquor were both highly organic (Liao et al. 2006). From
these results it was concluded that Cleaning Processes 1 and 6 might
Table 5. Average MLSS, OLR, backwash flow, and SRT for period of
be enhanced through the removal of the citric acid portion of the operation following each chemical cleaning process
cleaning under the conditions used in this study.
OLR Backwash flow
MLSS (g=L) (kg=m3 =d) (L=h) SRT (day)
Cleaning
process Average SD Average SD Average SD Average SD
Table 3. Irreversible fouling resistance coefficients, clean water flux re-
coveries, and clean water OP recoveries calculated following each part of 1 21.8 1.9 2.8 0.8 56.4 6.0 76 234
chemical membrane Cleaning Process 6 2 19.6 1.5 2.7 0.7 47.0 11.8 95 154
3 20.9 2.4 2.2 0.7 35.7 13.7 232 307
Variable Part 1 Part 2 Part 3
4 21.1 1.6 2.0 0.6 51.3 11.4 357 266
K IR (10−12 m−1 ) 6.84 9.09 3.72 5 23.6 1.3 2.1 0.4 65.7 6.4 ∞ —
CW J (%) 9 6 14
Note: MLSS = mixed liquor suspended solids; OLR = organic loading rate;
CW OP (%) 9 6 14
SD = standard deviation; and SRT = sludge retention time.
J. Environ. Eng., 2019, 145(5): 04019018

J. Environ. Eng., 2019, 145(5): 04019018

When mixed liquor flowing through the recirculation line sludge changed slightly over the course of operation (shifting from
entered the membrane module, a pressure gradient generated from a granular consistency to a more viscous consistency); however,
a peristaltic pump (Masterflex, Ontario, Canada) instantaneously 87 days after placing this membrane module inline the mixed liquor
and continuously filtered a portion of the mixed liquor through the had a TSS concentration of 21.3 g=L, a VSS concentration of
pores of the membrane generating filtrate, which will henceforth 20.4 g=L, and a SMA of 0.17g-CODCH4 =g-VSS=day, making it
be referred to as permeate. The permeate flow direction was con- quite similar to the initial seed sludge in terms of sludge biomass
trolled by two solenoid valves. For forward filtration, the permeate activity.
flowed through the permeate recycle line back into the CSTR. The
solenoid valves would flip every 15 min and a secondary peristaltic
pump (Masterflex, Ontario, Canada) would force a portion of the Analytical Methods
permeate into a secondary holding tank, where it would exit as ef- The COD, TSS, VSS, BOD, and total alkalinity were determined
fluent or be recycled for backwashing processes. The backwashes following Standard Methods (APHA 2005). Total phosphorous
would occur every 30 min, again controlled by the solenoid valves (TP), total kjeldahl nitrogen (TKN), phosphate-phosphorous, and
and secondary pump. For CEBs, a mixed solution of 2,600 ppm ammonia nitrogen were measured according to the HACH Water
sodium hydroxide and 2,600 ppm sodium hypochlorite was in- Analysis Handbook (Hach Company 2002). The methane con-
jected into the permeate backwash just prior to entering the mem- tent of the biogas was measured using a Varian CP-3800 gas
brane module (at a ratio of 80% permeate and 20% chemical chromatograph. The chromatograph was fitted with an Alltech
solution), creating an approximate concentration of 520 ppm so- CTR 1 concentric packed column and a thermal conductivity de-
dium hydroxide and 520 ppm sodium hypochlorite entering the tector maintained at a constant temperature of 180°C. Helium was
membrane module. The chemical concentrations used for CEBs used as a carrier gas operating at a flow rate of 30 mL=min. The
were recommended by the membrane manufacturer based on past temperature of the injection port was set to 120°C. Biogas injec-
performance data. tions were done in triplicate, and SMA tests were conducted in
Waste anaerobic sludge was pumped out of the reactor on a duplicate using batch studies operated at 36°C. The tests were con-
2-h cycle. Every 2 h a peristaltic pump (Masterflex, Ontario, ducted in 500 mL serum bottles with a working volume of 450 mL.
Canada) would pump mixed liquor from the base of the CSTR out The bottles were filled with reactor sludge, acetic acid as a sub-
of the reactor as waste sludge. The specific flow rate of the waste strate, pH buffer, and mineral bases. Young and Cowan (2004)
sludge was adjusted to maintain reactor-mixed liquor suspended outlined the procedures and mixed solutions used. All flow and
solids (MLSS) concentration at the desired level of approximately pressure readings were noted once daily at similar times using in-
19–21 g=L. line flow meters and pressure sensors.
Biogas exited the CSTR through a gas trap at the top of the
reactor and passed into a plastic gas collection bag, where the flow
of biogas was regulated. The regulated outflow of biogas moved Experimental Design
through a mass-flow meter (AALBORG, Orangeburg, New York), The experimental design was developed through applying the
where its total mass was measured and logged, before exiting the pretest–posttest control group design methodology. This method-
system. The mass-flow meter was calibrated with a headspace ology demonstrated that an effect occurred only following the
gas having a composition of 70% methane gas and 30% carbon change in a certain variable from control conditions (Leedy and
dioxide gas. Ormrod 2005). Applying the pretest–posttest control group design
methodology an initial clean water flux, clean water operational
Wastewater Characterization permeability (OP), mixed-liquor flux (MLF), and mixed-liquor op-
erational permeability (MLOP) were established for the membrane
The anaerobic CSTR was fed with a high-strength (average COD of module prior to its placement inline (for clean water values) and on
84.8 8.9 g=L) potato starch–based synthetic wastewater simulat- Day 2 of operation (for mixed-liquor values). Following a chemical
ing typical food-processing industry wastewater. Snowdon et al. cleaning process, each of these parameters was measured and cal-
(2018) provided a complete characterization of the synthetic waste- culated to identify how efficient each chemical cleaning process
water. The CSTR was equipped with a level gauge, and whenever was in recovering flux and OP.
the mixed-liquor level fell below a preset threshold of 0.94 m, A total of six chemical cleaning processes were applied to the
an influent pump (Masterflex, Ontario, Canada) came online and membrane module, and their respective cleaning conditions may be
pumped wastewater into the system until the mixed-liquor level viewed in Table 1. The two different cleaning conditions were
returned to the threshold level. The daily influent flow rate was chosen because they offered the largest recoveries in clean water
equivalent to the sum of the effluent and waste flows, minus the flux and clean water OP during other membrane studies conducted
backwash flow. by the authors. Cleaning Processes 1 and 6 were dynamic cleanings
(with backwashing processes in place during the cleaning) and
consisted of three parts using three different chemical solutions.
Seed Sludge Characteristics
Cleaning Processes 2–5 were static cleans (with no backwashing
The reactor was initially seeded with granular sludge acquired processes in place) and consisted of one part with one chemical
from an anaerobic digester treating a fruit-juice-processing plant solution. Through the application of each set of cleaning conditions
wastewater in Lassonde, Quebec, Canada. The seed sludge had multiple times, repetitions were accomplished for comparison.
an initial total solid (TS) concentration of 23.2 g=L, total volatile In Table 1, the recirculation time is the amount of time the mem-
solid (TVS) concentration of 19.8 g=L, total suspended solid brane module was operated in forward filtration mode during the
(TSS) concentration of 21.4 g=L, volatile suspended solid (VSS) cleaning process. The soak time is the amount of time forward and
concentration of 19.5 g=L, and a specific methanogenic activity backward filtration was ceased, allowing the membrane module to
(SMA) of 0.18 g-CODCH4 =g-VSS=day. The membrane module sit and soak in the chemical cleaning solution. The number of back-
used throughout the duration of this study was first placed inline washes and duration are the quantity and duration of backward fil-
171 days after initial seeding of the reactor. The consistency of the tration periods, or backwashes, accomplished during a cycle of the
J. Environ. Eng., 2019, 145(5): 04019018

Table 1. Chemical cleaning conditions
Cleaning Recirculation Soak time Number of backwashes and
process Cycle time (min) (min) duration (number and min) Chemical used
2 N/A 20 N/A
5 N/A 20 N/A
8 N/A 20 N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 60 N/A
3 30 N/A N/A
2 N/A 20 N/A
5 N/A 20 N/A
8 N/A 20 N/A
chemical cleaning process. Finally, the chemicals used outline the module were also measured. These values allowed the MLF
exact concentration of chemical cleaning solutions applied to the and MLOP to be calculated. The mixed-liquor recoveries were
membrane module during each cleaning condition. calculated by dividing the post–chemical cleaning MLF or post–
Clean water flux was measured by isolating the CSTR recircu- chemical cleaning MLOP values by the initial MLF or initial
lation line from the CSTR itself, allowing a 50-L tank to be filled MLOP values measured on Day 2 of operation, and then multiply-
with clean water (maintained at around 30°C) and filtered through ing by 100. Equations used to calculate clean water and mixed-
the membrane module. Each clean water test recirculated the clean liquor recoveries are displayed in Eqs. (2) and (3), respectively:
water at five flow rates in increasing order (790, 1,580, 2,360,
3,150, and 3,940 L=h), and the resultant membrane fluxes and J; OPPC
TMPs were measured. Five different increasing flow rates were se- CW J;OP ð%Þ ¼ × 100 ð2Þ
J; OPo
lected to provide incremental increases in flux and TMP, reaching
the maximum flow rate of the recirculation loop at approximately
3,940 L=h. Operational permeability is defined as membrane flux
divided by instantaneous operational TMP, in other words pressure- J; OPPC
MLJ;OP ð%Þ ¼ × 100 ð3Þ
averaged flux, and may be defined by Eq. (1) (Hamden de Andrade J; OPo
et al. 2013):
Ji In Eq. (2), CW J;OP is the clean water flux or operational per-
OPi ¼ ð1Þ
ΔPi meability recovery (%), J; OPPC is the post–membrane cleaning
clean water membrane flux or clean water operational permeability
In Eq. (1), OPi is the instantaneous operational permeability (L=m2 =h, L=m2 =h=bar), and J, OPo is the initial clean water mem-
(L=m2 =h=bar), J i is the instantaneous membrane flux (L=m2 =h), brane flux or clean water operational permeability (L=m2 =h,
and ΔPi is the instantaneous transmembrane pressure (bar). Opera- L=m2 =h=bar). In Eq. (3), MLJ;OP is the mixed-liquor flux or opera-
tional permeability was considered so that TMP could be factored tional permeability recovery (%), J; OPPC is the post–membrane
into the flux recovery analysis. Clean water recoveries were calcu- cleaning mixed-liquor membrane flux or mixed-liquor operational
lated by dividing the post–chemical cleaning clean water flux or permeability (L=m2 =h, L=m2 =h=bar), and J; OPo is the initial
post–chemical cleaning clean water OP by the initial clean water mixed-liquor membrane flux or mixed-liquor operational per-
flux or initial OP values measured on Day 1 of operation, and then meability (L=m2 =h, L=m2 =h=bar).
multiplying by 100. Once the membrane module was placed back Membrane resistance coefficients were calculated using the
inline for mixed-liquor recirculation, and after 24 h of recirculation resistance-in-series model [Eq. (4)]. This equation was adopted and
at a flow rate of 6,000 L=h, the flux and TMP of the membrane developed from other studies (Ruigomez et al. 2017):
J. Environ. Eng., 2019, 145(5): 04019018

Table 2. Membrane resistance coefficients calculated following each would allow the CEBs to be quite efficient in both removing
chemical membrane cleaning process (unit: 10−12 × m−1 ) built-up membrane foulants and preventing their rate of accu-
Cleaning process KM KR K IR mulation while operating. Zhang et al. (2007) found that, when
operating a tubular polyethersulfone side stream AnMBR for the
1 0.62 4.66 0.73
2 9.33 2.00 treatment of swine manure, long-term operation without chemical
3 17.3 2.60 cleaning processes resulted in residual membrane fouling that was
4 18.7 1.45 difficult to remove with the chemicals used in their study (nitric
5 6.27 1.32 acid and ethylenediaminetetraacetic acid). By using a constant
6 6.34 3.72 chemical cleaning process, the build-up of residual membrane
foulants might have been prevented. In this study, the prevention
of the build-up of difficult-to-remove membrane foulants was ob-
served, and a constant source of chemical cleaning was supplied
first placed inline. Reversible membrane fouling occurred rapidly through CEBs, resulting in more effective fouling mitigation. It was
when the membrane module was first placed inline, resulting in a found that K R increased by 85% over 106 days between Cleaning
quick drop in membrane flux (a loss of 35% in MLF over 5 days), Process 2 and Cleaning Process 3, whereas K R increased by only
which merited the first chemical cleaning. The membrane resis- 8% over 121 days between Cleaning Process 3 and Cleaning
tance coefficients were calculated (Table 2) to further support this Process 4, indicating a decrease in the rate of K R accumulation.
observation. Following Cleaning Process 1, K R was found to be With the cleaning conditions of Cleaning Processes 2 and 3 being
much greater than K IR (values of 4.66 × 10−12 m−1 and 0.73 × identical, and operating conditions being fairly consistent in the
10−12 m−1 , respectively), indicating that the majority of foulant periods following these cleaning processes, the primary reason for
removed was reversible membrane fouling. Since this was the first the decline in reversible foulant accumulation may be attributed to
chemical cleaning applied to the membrane module, K IR was also CEBs being in place 60% of the days following Cleaning Process 3.
at its lowest value throughout the study. As the majority of mem- K R was 6.27 × 10−12 m−1 following Cleaning Process 5. A de-
brane fouling was contributed from reversible foulants, which are crease in this value from Cleaning Process 4 to 5 indicates that the
removed effectively through chemical cleaning processes, and there rate of reversible foulant build-up was greatly reduced during
was little contribution from irreversible membrane foulants relative this period of operation, which may again be attributed to regular
to other cleaning methods, Cleaning Process 1 appeared much more CEBs being in place.
effective. With this consideration in mind, it is difficult to distin- The irreversible fouling resistance coefficient typically increases
guish whether Cleaning Process 1 was superior in terms of clean over the duration of membrane filtration and would not be antici-
water flux recovery over all other cleaning processes due to its pated to decrease unless operational or cleaning conditions could
cleaning conditions or whether its enhanced performance was sim- be enhanced significantly. Considering that K IR decreased from
ply due to occurring so soon after the membrane was first placed Cleaning Process 3 to 4, and again from Cleaning Process 4 to 5,
inline. Wang et al. (2017) did find that the maximum amount while all of these cleaning processes operated under the same
of organics in their chemical cleanings was desorbed through the conditions, indicates that the presence of CEBs might also en-
use of an alkaline and acidic cleaning process when treating MBR hance regular cleaning protocols in such a way that irreversible
effluent. Cleaning Process 1 results agree with this observation, as foulants (that could not be removed without CEBs in place) could
the cleaning conditions provided both alkaline and acidic chemicals then be removed. Zhang et al. (2007) concluded that weekly mem-
and it had the largest clean water flux recovery, indicating more brane cleanings provided no significant benefits in membrane per-
foulant removal. formance compared to monthly membrane cleanings, attributing
Cleaning Processes 4 and 5 outperformed Cleaning Processes 2 the lack of benefits to the fact that irreversible fouling accumulated
and 3 in terms of clean water flux recovery. This result was not more rapidly than the frequency of either cleaning. Perhaps the
initially anticipated because, due to the accumulation of irreversible daily CEBs in this study occurred frequently enough that irre-
membrane fouling over the course of membrane operation, succes- versible foulants were unable to accumulate effectively between
sively applying the same cleaning conditions typically results in membrane cleanings. From Cleaning Process 5 to 6 the K R value
a decrease in flux and OP recoveries with each successive clean remained quite steady between cleans, whereas K IR increased by a
(Sun et al. 2016). Since all four of these cleaning processes were factor of 2.8 with CEBs regularly in place for 26 days of operation.
identical in conditions, it was expected that irreversible membrane This indicates that Cleaning Process 5 may have been superior to
foulants would accumulate, making each successive membrane Cleaning Process 6 since K IR increased once again between cleans.
cleaning process less efficient than the one preceding it. A differ- Similar to the results for clean water flux recovery, Cleaning
ence that was present leading into Cleaning Processes 4 and 5 was Processes 1, 5, and 4 offered the largest clean water OP recoveries
that there were CEBs present 60% (73 days) and 100% (77 days) of with values of 49%, 31%, and 30%, respectively. Since the recov-
the time, respectively, leading into each of these chemical cleaning eries in clean water operational permeability are near identical to
processes. Cleaning Process 5 had more CEBs prior to it and out- the recoveries in clean water flux, this indicates that the chemical
performed Cleaning Process 4. Similarly, Cleaning Process 4 had cleanings had similar recoveries in TMP. One reason why Clean 1
more CEBs prior to it compared to Cleaning Processes 2 and 3 (as outperformed all of the others by a large margin may again be
there were no CEBs in place leading into Cleans 2 and 3) and out- attributed to its early application and lack of irreversible foulant
performed them. From this observation, it would appear that regu- accumulation.
lar CEBs synergize with chemical cleaning processes and enhance In terms of MLF and MLOP recoveries, all cleaning processes
their performance in terms of clean water flux recovery. Regular exhibited good performance, with the lowest MLF recovery being
contact between a low-strength caustic chemical cleaning solution Cleaning Process 3 (with a value of 67%) and the lowest MLOP
known for its efficiency in removing organic membrane foulants recovery being Cleaning Process 2 (62%). Compared to the largest
and a wastewater/mixed liquor composed almost entirely of organic clean water flux and OP recoveries of 53% and 49%, respectively,
matter as a source of membrane foulants (with average volatilities this indicates that the chemical cleanings were much more effective
for the wastewater and mixed liquor of 99% and 94%, respectively) at recovering inline mixed-liquor flux rather than clean water flux.
J. Environ. Eng., 2019, 145(5): 04019018

sludge wastage rate increasing the foulant contribution from solu- Process 5, with CEBs in place 100% of the time, the rate of decline
ble microbial products (SMPs) known to block membrane pores in operational permeability increased to 0.12 L=m2 =h=bar=day, a
and too low of a sludge wastage rate promoting the development threefold increase over the rate following Cleaning Process 4. The
and growth of extracellular polymeric substances (EPSs), again duration of operation and the standardized duration of operation
contributing to membrane fouling (Iorhemen et al. 2016; Ibeid both reached their lowest values of 26 and 75 days, respectively,
et al. 2016; Jarusutthirak and Amy 2006; Le-Clech et al. 2006). during this period of operation as well. Zhou et al. (2014) found
Studies have found that an optimum SRT range for both the min- that the average pore size of their PVDF membranes increased
imization of SMP and EPS is around 25–30 days (Barker and after 1 week of operation with 1% sodium hypochlorite cleaning.
Stuckey 1998). Since the SRT for both of these periods of operation This resulted in a decrease in both their fouling resistance coeffi-
was well above this value and within 25% of one another, SRT most cients and operating periods between cleans. Wang et al. (2018)
likely did not substantially contribute to their observed differences noticed both their clean water flux and rate of hydraulic resistance
in the rate of decline in operational permeability. Additionally, the accumulation increased with an increased soak time in a sodium
average backwash flow rate was greater during the period of oper- hypochlorite cleaning solution, attributing the increase to an
ation following Cleaning Process 1, which would have contributed enlargement in the pore size of the membrane module. Constant
towards a slower rate of operational permeability decline. Despite contact between caustic cleaning solutions and the membrane sur-
the operating conditions being on average quite similar, no CEBs face may provide initial benefits; however, over time it may lead to
being in place, and post–Chemical Cleaning Process 1 conditions an increased rate of fouling due to pore size enlargement.
having more favorable physical cleaning processes in place, the Recent studies have also noted an increasing trend between the
operating period following Cleaning Process 2 showed a much quantity of dissolved organic matter and dose of sodium hypochlo-
lower rate of OP decline. The lower rate of OP decline may have rite the biomass is exposed to (Cai et al. 2016; Cai and Liu 2016).
been primarily due to the nature of the chemical cleaning itself, Cai and Liu (2016) noted that increasing the concentration of dis-
again supporting earlier conclusions that Cleaning Processes 2–5 solved organic matter in the mixed liquor resulted in an increased
were superior to Cleaning Processes 1 and 6. membrane-fouling rate, although they found that following an
Chemically enhanced backwashes were first applied to the exposure of 55 mg-NaOCl=g-MLVSS the soluble COD concentra-
membrane module 40% of the way through the period of operation tion increased by only 180 mg=L, and under the operating condi-
following Cleaning Process 3. The average operating conditions tions provided in this study the dose would have been in the range
following Cleaning Processes 2 and 3 were relatively constant. of 6–7 mg-NaOCl=g-MLVSS, meaning the impact in this case
The largest discrepancies between these periods of operation were would have been very minimal. Nonetheless, it is important to con-
the average backwash flow rate, which decreased by 11.3 L=h, and sider that increased exposure to sodium hypochlorite may have
the average SRT, which increased by 137 days from the period of increased the generation of SMPs, which would have contributed
operation following Cleaning Process 2 to the period of operation to the increased rate of decline in operational permeability as well.
following Cleaning Process 3. Increasing the SRT would be ex- Initially, the CEBs may be beneficial in retarding the rate of irre-
pected to promote SMP foulant development, which would lead versible foulant accumulation on the membrane module and in its
to increased pore blockage. This in turn would result in less effi- pores, but over time the constant contact between the chemical sol-
cient backwashes and, paired with a decreased backwash flow rate, ution and the membrane surface could be damaging. The pore size
should have resulted in an increased rate of operational permeabil- of the membrane module used for this study was unable to be in-
ity decline. The rate of operational permeability decline did investigated prior to and following exposure to CEBs because this
crease; however, the increase was very minimal, with a rate of would have required destroying the module, and it was decided that
decline of 0.03 L=m2 =h=bar=day. The presence of CEBs for the membrane would be stored for reuse following the conclusion
60% of the operational period following Cleaning Process 3 likely of the study. If the average pore size of the membrane did increase,
assisted in preventing the build-up of pore-blocking irreversible and there was no impact on its TSS removal efficiency: it remained
reversible foulants, resulting in this minimal increase in the rate of above 99.9% throughout the duration of the entire study.
operational permeability decline. Additionally, following Cleaning
Process 3 was the longest period of operation and the longest stand-
ardized period of operation. Zhou et al. (2014) found that the use of Conclusions
sodium hydroxide backwashes retarded the rates of total foulant
and irreversible foulant accumulation, attributing their increased Two different chemical cleaning conditions were applied to a tubu-
performance to the combined hydraulic and chemical mechanisms lar, nanofiltration membrane module a total of six times, and each
enhancing the detachment of biopolymers. It appears that CEBs are chemical cleaning process’s individual performance was evaluated
beneficial, initially, in terms of reducing the rate of operational per- and compared to that of the others. The membrane module was
meability decline and extending the duration of operation between operated as part of an AnMBR system for 466 days of continuous
membrane cleanings. Ramos et al. (2014) were able to delay the operation. Over its course of operation, CEBs were eventually in-
requirement for recovery cleaning by 3 years by incorporating corporated as a means of enhanced physical cleaning, and their
CEBs. This study fell short of a 3-year extension, likely due to dif- influence on membrane fouling mitigation and the chemical clean-
ferences in membrane modules, and mixed-liquor/feed wastewater ings applied were also investigated. Overall, the following conclu-
characteristics, yet an extension over conditions without CEBs in sion may be drawn from this study:
place was possible. • The static, one-solution chemical cleaning processes offered
The period of operation following Cleaning Process 4, with higher recoveries in clean water flux, clean water operational
CEBs in place 100% of the time, saw the rate of decline in opera- permeability, mixed-liquor flux, and mixed-liquor operational
tional permeability increase by only 0.01 L=m2 =h=bar=day over permeability, on average, than the three-solution chemical
the period of operation following Cleaning Process 3. The reason cleaning processes if the performance of Cleaning Process 1
for this minimal increase in the rate of operational permeability de- can be attributed to its having taken place so early in the mem-
cline, and the second highest standardized duration of operation, is brane’s operational period, making the one-solution chemical
most likely, once again, the presence of CEBs. Following Cleaning cleaning process the highest performing alternative.
J. Environ. Eng., 2019, 145(5): 04019018

• No cleaning processes were able to recover clean water flux Leedy, P., and J. Ormrod. 2005. Practical research: Planning and design.
and operational permeability very efficiently (with the highest New York: Pearson.
clean water flux recovery being 53%). However, all cleaning Liao, B. Q., J. T. Kramer, and D. M. Bagley. 2006. “Anaerobic membrane
processes were able to recover mixed-liquor flux and opera- bioreactors: Applications and research directions.” Crit. Rev. Environ.
tional permeability effectively (with a lowest mixed-liquor flux Sci. Technol. 36 (6): 489–530. https://doi.org/10.1080/1064338060
0678146.
recovery of 67%).
Lousada-Ferreira, M., J. B. van Lier, and J. H. J. M. van der Graaf. 2014.
• The citric acid portion of Cleaning Process 6 had a detrimental
“Impact of suspended solids concentration on sludge filterability in full-
effect on its overall foulant removal effectiveness. scale membrane bioreactors.” J. Membr. Sci. 476: 69–75. https://doi.org
• Daily CEBs enhanced the foulant removal of successive chemi- /10.1016/j.memsci.2014.11.012.
cal cleaning processes applying the same cleaning conditions. Meng, F., S. R. Chae, A. Drews, M. Kraume, H. S. Shin, and F. Yang. 2009.
• Daily CEBs decreased the rate of foulant accumulation. “Recent advances in membrane bioreactors (MBRs): Membrane fouling
• Considering these results and those of Snowdon et al. (2018), and membrane material.” Water Res. 43 (6): 1489–1512. https://doi.org
daily low-strength CEBs were observed to be an effective strat- /10.1016/j.watres.2008.12.044.
egy for both membrane fouling preventive maintenance and as Metcalf & Eddy, Inc.. 1991. Chap. 10 in Wastewater engineering treat-
assistance in chemical cleaning recovery operations. ment, disposal, and reuse, 987–988. New York: McGraw-Hill.
Ozgun, H., R. Kaan Dereli, M. Evren Ersahin, C. Kinaci, H. Spanjers, and
J. B. van Lier. 2013. “A review of anaerobic membrane bioreactors for
Acknowledgments municipal wastewater treatment: Integration options, limitations, and
expectations.” Sep. Purif. Technol. 118: 89–104. https://doi.org/10
The authors would like to acknowledge and thank the following .1016/j.seppur.2013.06.036.
funding agencies and industrial collaboration: the Natural Science Pilli, S., T. T. More, S. Yan, R. D. Tyagi, and R. Y. Surampalli. 2016.
and Engineering Research Council (NSERC), the New Brunswick “Fenton pre-treatment of secondary sludge to enhance anaerobic diges-
Innovation Foundation (NBIF), and ADI Systems Inc. They would tion: Energy balance and greenhouse gas emissions.” Chem. Eng. J.
also like to thank Jonathan Dargavel and Dr. Dennis Connor for 283: 285–292. https://doi.org/10.1016/j.cej.2015.07.056.
their help and contributions with data collection and testing Ramos, C., F. Zecchino, D. Ezquerra, and V. Diez. 2014. “Chemical clean-
throughout the duration of the study. ing of membranes for an anaerobic membrane bioreactor treating food
industry wastewater.” J. Membr. Sci. 458: 179–188. https://doi.org/10
.1016/j.memsci.2014.01.067.
References Ruigomez, I., E. Gonzalez, S. Guerra, L. E. Rodriguez-Gomez, and L. Vera.
2017. “Evaluation of a novel physical cleaning strategy based on HF
APHA (American Public Health Association). 2005. Standard methods for membrane rotation during the backwashing/relaxation phases for
the examination of water and wastewater. Washington, DC: APHA. anaerobic submerged MBR.” J. Membr. Sci. 526: 181–190. https://doi
Barker, D. J., and D. C. Stuckey. 1998. “A review of soluble microbial .org/10.1016/j.memsci.2016.12.042.
products (SMP) in wastewater treatment systems.” Water Res. 33 (14): Skouteris, G., D. Hermosilla, P. Lopez, C. Negro, and A. Blanco. 2012.
3062–3082. https://doi.org/10.1016/S0043-1354(99)00022-6. “Anaerobic membrane bioreactors for wastewater treatment: A review.”
Cai, W., J. Liu, X. Zhang, W. J. Ng, and Y. Liu. 2016. “Generation of Chem. Eng. J. 198–199: 138–148. https://doi.org/10.1016/j.cej.2012.05
dissolved organic matter and byproducts from activated sludge during .070.
contact with sodium hypochlorite and its implications to on-line chemi- Snowdon, J., K. S. Singh, and G. Zanatta. 2018. “Optimization of an
cal cleaning in MBR.” Water Res. 104: 44–52. https://doi.org/10.1016/j external nanofiltration anaerobic membrane bioreactor treating a high-
.watres.2016.07.065. strength starch-based wastewater.” J. Environ. Eng. 144 (6): 04018032.
Cai, W., and Y. Liu. 2016. “Enhanced membrane biofouling potential https://doi.org/10.1061/(ASCE)EE.1943-7870.0001376.
by on-line chemical cleaning in membrane bioreactor.” J. Membr. Sci. Sun, Y., Y. Fang, P. Liang, and X. Huang. 2016. “Effects of online chemical
511: 84–91. https://doi.org/10.1016/j.memsci.2016.03.039. cleaning on removing biofouling and resilient microbes in a pilot mem-
Hach Company. 2002. Water analysis handbook. Loveland, CO: Hach brane bioreactor.” Int. Biodeterior. Biodegrad. 112: 119–127. https://
Company. doi.org/10.1016/j.ibiod.2016.05.010.
Hamden de Andrade, L., F. D. dos Santos Mendes, J. Cawettiere Espindola, Wang, J., K. Li, D. Yu, J. Zhang, and Y. Wei. 2017. “Fouling characteristics
and M. C. Santos Amaral. 2013. “Internal versus external submerged and cleaning strategies of NF membranes for the advanced treatment of
membrane bioreactor configurations for dairy wastewater treatment.”
antibiotic production wastewater.” Environ. Sci. Pollut. Res. 24 (10):
Desalin. Water Treat. 52 (16–18): 2920–2932. https://doi.org/10
8967–8977. https://doi.org/10.1007/s11356-015-5408-4.
.1080/19443994.2013.799048.
Wang, Q., H. Zeng, Z. Wu, and J. Cao. 2018. “Impact of sodium hypochlo-
Ibeid, S., M. Elektorowicz, and J. A. Oleszkiewicz. 2016. “Impact of elec-
rite cleaning on the surface properties and performance of PVDF mem-
trocoagulation of soluble microbial products on membrane fouling at
branes.” Appl. Surf. Sci. 428: 289–295. https://doi.org/10.1016/j.apsusc
different volatile suspended solids concentrations.” Environ. Technol.
.2017.09.056.
38 (4): 385–393. https://doi.org/10.1080/09593330.2016.1195879.
Iorhemen, O. T., R. A. Hamza, and J. H. Tay. 2016. “Membrane bioreac- Young, J., and R. Cowan. 2004. Respirometry for environmental science
tor (MBR) technology for wastewater treatment and reclamation: and engineering. Springdale, AR: SJ Enterprises.
Membrane fouling.” Membranes 6 (2): 33. https://doi.org/10.3390 Zhang, J., H. C. Chua, J. Zhou, and A. G. Fane. 2006. “Factors affecting
/membranes6020033. the membrane performance in submerged membrane bioreactors.”
Jarusutthirak, C., and G. Amy. 2006. “Role of soluble microbial products J. Membr. Sci. 284 (1–2): 54–66. https://doi.org/10.1016/j.memsci
(SMP) in membrane fouling and flux decline.” Environ. Sci. Technol. .2006.06.022.
40 (3): 969–974. https://doi.org/10.1021/es050987a. Zhang, J., S. I. Padmasiri, M. Fitch, B. Norddahl, L. Raskin, and E.
Kinnunen, V., A. Yla-Outinen, and J. Rintala. 2015. “Mesophilic anaerobic Morgenroth. 2007. “Influence of cleaning frequency and membrane his-
digestion of pulp and paper industry biosludge-long-term reactor per- tory on fouling in an anaerobic membrane bioreactor.” Desalination
formance and effects of thermal pretreatment.” Water Res. 87: 105–111. 207 (1–3): 153–166. https://doi.org/10.1016/j.desal.2006.07.009.
https://doi.org/10.1016/j.watres.2015.08.053. Zhou, Z., F. Meng, H. Lu, Y. Li, X. Jia, and X. He. 2014. “Simultaneous
Le-Clech, P., V. Chen, and T. A. G. Fane. 2006. “Fouling in membrane alkali supplementation and fouling mitigation in membrane bioreactors
bioreactors used in wastewater treatment.” J. Membr. Sci. 284 (1–2): by on-line NaOH backwashing.” J. Membr. Sci. 457: 120–127. https://
17–53. https://doi.org/10.1016/j.memsci.2006.08.019. doi.org/10.1016/j.memsci.2014.01.008.
J. Environ. Eng., 2019, 145(5): 04019018

Wat. Sci. Tech. Vol. 28. No.2. pp. 165-176. 1993. 0273-1223/93 $6'00 0'00
Printed in Great Britain. All rights reserved. Copyright © 1993 IAWQ
ANAEROBIC-AEROBIC TREATMENT OF
POTATO-STARCH WASTEWATER
U. Abeling and C. F. Seyfried
Institut for Siedlungswasserwirtscha/t und Abfalltechnik (Water Supply and Treatment of

Solid Wastes). University of Hannover. Welfengarten I. D-3000 Hannover I. Germany
ABSTRACT
Production of potato starch produces wastewaters witb high concentrations of COD, nitrogen and
phosphorus. The best known and most economical solution for tbe nearly full elimination of tbese substances
is tbe two-stage anaerobic-aerobic treatment. The anaerobic pretreatment must only go so far as to maintain
enougb COD for biological nitrogen and phosphorus elimination in tbe aerobic stage. To optimize tbis
process, several tests have been carried out on nitrification/denitrification by means of tbe intermediate
product nitrite. The carbon consumption amounts to only 60% in comparison witb denitrification via nitrate.
The essential parameter for regulating tbe process is tbe concentration of free ammonia in tbe reactor.
Concentrations of I to 5mg NH31l inhibit tbe nitratation but not tbe nitritation. A separation of botb partial
steps is possible. The content of ammonia was controlled by means of continuous NH4 and pH measuring.
The process is suitable for many wastewaters witb low CODrrKN ratios and higb nitrogen concentrations. If
tbere are undesired nitrite peaks in an aerobic treatment plant caused by bigh pH values and temperatures,
specific nitrogen elimination by means of nitrite is a reliable treatment system.
KEYWORDS
Potato-starch wastewater; anaerobic-aerobic treatment; high-strength ammonium wastewater;

nitrification/denitrification via nitrite; nitritation; nitratation; industrial wastewater; nitrogen removal.
INTRODUCTION
In Europe, about 4.7 Mio. t starch and starch decay products are produced each year. The main raw materials
used are com (60% of the production) and potatoes and wheat (nearly 20% each). The largest product group
is the true starches with nearly 30%, followed by glucose (23%), isoglucose (13%) and modified starch
(12%). Most starch is employed in the food industry (for example sweets, drinks, fruit manufacturing - total
57%), the paper and paperboard industry (25%), as well as the chemical and pharmaceutical industry (10%).
Whereas a short time ago most of the attention was directed to production processes and commercializing,
nowadays wastewater avoidance and wastewater treatment are becoming more and more important. The cost
share of advanced wastewater treatment in the total production costs amounts to nearly 15%.
The starch wastewaters have high nitrogen and phosphorus concentrations. In Germany both nutrients must
be largely eliminated, and this will soon apply to the whole of the European Community as far as the
harmonization of environmental protection is concerned.
For purification of industrial wastewaters, especially for starch factories, more and more two-stage
anaerobic-aerobic processes have been used during the last few years. This is useful - for energetic reasons
- when highly concentrated wastewaters are to be treated, as a substantial part of the organic load can be
1 65
166 U. ABELING and C. F. SEYFRIED
removed anaerobically, hereby producing methane gas. Only the pollution finally remaining has to be
eliminated aerobically by using certain aeration energy.
Minimizin� the Eneuy Consumption
The combined anaerobic-aerobic technology is the most economical solution for as long as only the removal
of carbon compounds is concerned. With regard to the elimination of nitrogen and phosphorus, however, this
process is problematic, as no nitrogen and phosphorus removal worth mentioning takes place during the
elimination of the greater part of the carbon compounds in the anaerobic pretreatment. The BODIN ratio
decreased from 9:1 to nearly 1:1, the BOD:P ratio from 50:1 to 5:1. It is no problem to nitrify the
ammonium in the following aerobic treatment stage, but denitrification and biological phosphorus removal
are often incomplete, possibly due to the low content of organic carbon compounds. Phosphorus may be
eliminated by precipitation with lime or metal ions, whereas the chemicaVphysical methods for ammonium
elimination cause many problems with the resulting products. A total nitrification, however, as well as a
denitrification as far as possible and an effluent concentration of 1 or 2mg PIl, is prescribed in Germany for
most industrial wastewaters (alcohol and food production (pectin, potato, starch and sugar processing
industry, slaughter-houses etc.), pulp and paper industry, animal waste).
In order to achieve a total denitrification sometimes external carbon sources (methanol, acetic acid) are used
(very expensive). In practice, anaerobic pretreatment is often evaded by means of a bypass, but this causes a
lower gas production and also sometimes problems concerning the sludge composition (bulking) in the
aerobic stage.
Our investigations aimed to realize a process which enables total denitrification in spite of far-reaching
anaerobic pre-elimination. In contrast to the process which has been used so far, nitrification/denitrification
was carried out with assistance of the intermediate product nitrate. The carbon consumption amounts to only
60% in comparison with denitrification via nitrate. The essential parameter for regulating the process is the
concentration of free ammonia in the reactor (controlled by means of continuous NH4 and pH measuring).
Concentrations of 1 to 5mg NHyt inhibit the nitratation but not the nitritation.
The experiments were carried out in half-technical pilot plants for a period of three years, using the
wastewater of a potato starch factory. At the same time detailed examinations were made in laboratories
with regard to the kinetics and to the inhibition of nitrification/denitrification by NH3 and HN02. The new
process is available for all wastewaters with high ammonium concentrations and pH > 7.5, which often
cause problems concerning the inhibition of nitrogen removal dependent on high nitrite or ammonium
concentrations.
TREATMENT OF POTATO-STARCH WASTEWATER
Wastewater Production
During potato starch production, there are three partial wastewater streams with different concentrations of
COD, TKN and P. Figure 1 shows a flow diagram of potato starch production. The concentrations and
specific wastewater quantities are listed in Table 1.
o 1
Potato-s1arCb wastewater
00 2P ..---""T'I'-r----,.---.,..---, 98
171
10 I;?, 7" 9o
III 80
I.S +---��,..-_+---+_---;
20
NH.
I I 70
II 1/-lS·C
_UJ
30·C J I _S·C 60
60
/I HNo,-N
III
O.S+---+-�r\t---+_---;
70
I Vi NHJ
20
90
1/ /" 10
�V OP�,--��--�6-��---��
8
o
6 7 8 9 10 11 12 13
pH pH
+ -
17 NH -N . 10PH 46 N0 -N
4 2
(mg/I) HN0 -- (mg/l)
2
= • ---
14 YKw + 10pll 14 K · 10pll

.
"t, .. ioniz.tion constant of the _i. equilibrhlD equation K

•
• •(-2,300/(273 + ·C»
•
lew Ionization constant of water ANTHONISEN ET AL. (1976)
•(6,344/273+·C)
Figure 4 . Dissociation balances NlLt+/NH3 and N02IHN02 (Nybuis, 1985).
/K
"t, w ..
The inhibition of Nitrosomonas and Nitrobacter by ammonia has been proved in much fundamental research
(Anthonisen, 1974; Bergeron, 1978; Neufeld et al., 1980; Nyhuis, 1985; Dombrowski, 1991; Balmelle et al.,
1992). Recently, there have appeared several speculations that - especially concerning Nitrobacter - the
inhibition is not caused by NH3' but by the intermediate product hydroxylamine (NH20H) (Yang and
Alleman, 1992). Here also the non-dissociated form is supposed to be responsible. Further tests referring to
this remain to be reported.
If nitrification should be reduced to the fIrSt partial step (nitritation), the activity of the Nitrobacter must be
largely reduced and the Nitrosomonas must correspondingly gain growth advantages. In several tests with
continuously operated laboratories Nyhuis (1985) could prove that Nitrosomonas (NH4 � NOV and
Nitrobacter (N02 � N03 ) show different reactions on ammonia and nitric acid. The sensibility of
Nitrobacter vis-a.-vis ammonia is much bigger than the substrate inhibition of Nitrosomonas. Nyhuis found
for Nitrobacter an inhibition even from 1 mg/l NH3 onwards; over 10 mg/l there was no more growth.
Nitrosomonas was inhibited only at concentrations of 40 mg/l onwards. Theoretically, at an ammonia
concentration of 10 mg NHyt in the aerobic tank the nitratation should be completely inhibited, while the
nitritation happens at maximal rate. The tests described hereafter were carried out to fix the inhibition limits
in continuous operation, and above all to find out whether both processes can be completely decoupled.
EJij!!<rlmental units
Two plants had been operated in parallel each with one anoxic and one aerobic reactor. In the first plant a
complete nitrification/denitrification with nitrate had been carried out. The arrangement of the second plant
is shown in Figure 5. In the aerobic reactor the pH value was controlled by means of a NaOH dosage and
regulated to pH 8.2-8.4. Furthermore the content of ammonium was controlled by means of regulation of the
oxygen supply. The nitrification tanks in unit 2 were filled with plastic media (Fa. NSW AG) with a specific
surface of 150m2 /m3 . As residual concentrations of ammonium and nitrite occur due to this kind of system,
Potato-starch wastewater 173
TABLE 5. Chemical Characterization of Potato Starch Wastewater
COD BDD
S
organic TKN NH -N
4
N�-N Tot P SS BDD ITKN
S
CDD/TKN
acids
[mslll [-l [-l
ra.. wastewater 17,000 10,000 870 150 190 11.5 19.5
anaerobic effl. 1,600 700 500 820 730 10 180 1,000 0.85 1.95
CQntrol 2f the DitritaUIlD
Both plants were started with nitrifying sludge from the aerobic stage of the large-scale. The inhibition of
the nitritation was planned to be achieved in the second plant by increasing the pH value. According to the
basic research work by Anthonisen (1974), each of the non-ionized forms of the ammonium and of the
nitrite have - as ammonia (NH3) and as nitrous acid (HNOz) - an inhibition effect on the Nitrosomonas and
the Nitrobacter. The Nitrobacter react more sensitively so that the concentrations in the range O. I to 1.0 mg
NH� are sufficient for starting an inhibition. The dissociation balance for ammonium-ammonia is
dependent on the temperature and on the pH value (Fig. 4.). At the pH value of 8.3 and a temperature of
20DC, 7.3% of the ammonium are present as ammonia. A concentration of 20 mg/l NH4-NIl in the reactor
thus produces an ammonia concentration of 1.8 mg NH31l.
The adjustment of the pH value was effected by means of a computer controlled dosage of soda solution
(NaOH). The fluctuation range amounted to ± O. I pH. As the content of ammonia at a constant pH value is
dependent on the altitude of the ammonia concentration in the reactor, and as this fluctuated, a pH control
alone is not sufficient. At a total nitrification the ammonia concentration is equal to zero, anyhow, due to
lack of ammonium in the reactor. This is the reason why also the content of ammonium had been adjusted
by computer-controlled aeration in order to achieve a constant concentration of ammonia. The measurement
was carried out with the help of a continuous NH4 analyzer (Bran & Liibbe). The temperature remained
constant during the measurement phases, which is important because the inhibiting effect of decreasing
temperatures is greater for Nitrobacter than for Nitrosomonas (Randall and Buth, 1984). Hence, nitrite
accumulation is possible due to lack of temperature control.
Nitrification
Inboth plants a complete nitrification respectively, and nitritation at space loadings of 0.3 kg NH4-N/m 3.d
could be reached. Thus, nitrification by means of nitrite with equal COD value in the influent has no
advantages as far as the required nitrification volume is concerned, because the necessary sludge age is the
same in both cases. Anyway, there is need of only 75% of the oxygen for complete nitrification.
As ammonia in higher concentrations also has an inhibiting influence on Nitrosomonas, several tests were
carried out to fmd out the influence of ammonia concentration on nitritation rate. Figure 6 shows the results.
Contrary to other authors, who analyzed much higher inhibition limits, a reduction was stated of the
nitritation rate from 7 mg/l NH3 onwards (Hyhuis, 1985: 40-200 mg NH� Anthonisen, 1974: 10-150 mg
NH�). Neufeld (1980) also found the start of inhibition to be from 10 mg NH� onwards.
The curve of nitritation rate follows an enlarged Haldane function (Eq.1), which - in addition - considers
the HNOrconcentration as non-competitive inhibition.
Potato-slarch wastewater 175
Nitrogen removal
Both pilot plants were operated under substrate limited conditions in order to guarantee a direct
comparability of denitrification results over nitrite and nitrate. Furthermore, the maximum possible
denitrification was limited to 92% by the recycle water quantity. To reach low nitrate effluent concentrations
at high influent concentrations, a simultaneous/intermitting denitrification process should be chosen.
Otherwise, a very high recycle rate is required.
In Table 6, the average results of nitrogen elimination over nitrite and nitrate are shown. It shows that by
means of nitrite a considerably higher elimination rate can be achieved. It must be take into account that
both processes were substrate-limited. With sufficient substrate supply, with plant 2 there can be achieved
effluent rates from about 60 mg N03-Nn (maximum possible value depending on recycle water quantity);
meanwhile in plant I at equal substrate supply the effluent concentration would be come up to 330 mg
N03-Nn.
TABLE 6. Results - Nitro!:en Remoyal - Ayera" values <April-June 1991)
influent effluent
plant plant 2
nitritation nitratation
COO [mg/l] 1,600 80.0 90.0 70.0

TKN [mg/l] 820 3.0 23.6 3.2
NH4-N [mg/l] 750 1.0 20.0 1.2
N0 2-N [mg/l] 0.5 21.6 0.6
r
NO N [mg/l] 10 348.0 63.4 102.3
ef ect of
N-elimination [%] 57.1 86.8 87.1
effect of
denitrification [%] 59.6 95.9
COOt NOx-N [-] 3.63 2.28
- 63 % of plant 1
As already mentioned, according to the results of the batch test, a complete decoupling of nitritation is not
possible (Figure 4). Anyway, the COD consumption in the process over nitrite compared to over nitrate
came to only 63% in the semi-technical tests, which means only a little margin to the 60% which is possible
by way of calculation (refer to Table 3). Therefore, a large separation may be supposed.
It was important to find out whether the improved elimination results of plant 2 were possibly due to strip
effects in the activation; hence a balance-sheet of the basins was established. Independent of the pH value,
wrong amounts of ± 7% resulted from this (Abeling and Seyfried, 1992). Thus, considerable nitrogen
elimination by means of stripping of ammonia can be excluded.
CONCLUSIONS
The example of potato starch wastewater was showed that wastewaters with high nitrogen concentrations
and an unfavourable CODIN ratio < 3.5 can be completely cleaned and made operational with the
nitrification/denitrification process. 37% of the carbon can be saved, compared to the conventional nitrogen
elimination over nitrite. With the help of anaerobic pretreatment, this carbon can be transferred to methane,
which contributes to optimization of the anaerobic-aerobic process from an economical point of view. The
operating costs are reduced by about 20%, and in total there are lower investment costs (tank volumes,
sludge treatment, aeration).
In order to achieve low nitrate effluent concentrations at high nitrogen concentrations, a simultaneous
denitrificatiuon process is recommended. The process must be two-staged because oxidation of the
176 U. ABELINO and C. F. SEYFRIED
remaining nitrite and ammonia concentrations is required. The process is controlled by continuous pH value
and ammonium measurements.
As the pH value must be increased to 8-8.5, the process is suitable above all for wastewaters which are
located in this pH value range. On the one hand, often undesired nitrite upgradings appear which cause
considerable operation break-downs, but with a specific decoupling of nitritation from nitratation, this can
be avoided.
RFERENCES
Abeling, U. and Seyfried, C. F. (1992) Anaerobic-areobic treatment of high-strength ammonium wastewater - nitrogen removal
via nitrite. Wilt. Sci. Techn., 26, No. 5-6, 1007-1015
Anthonisen, A. C. (1974) The effects of the free ammonia and free nitrous acid on the nitrification process. Ph.D. Thesis, Cornell
University, Engineering, agriCUltural
Anthonisen, A. C., Loehr, R C., Prakasam, T. B. S. and Srinatu, E. O. (1976) Inhibition of nitrification by ammonia and nitrous
acid. J. Wat. Poilu!. Control. Fed., 48, 835-852
Austermann-Haun, U. (1990) Stiltkefabriken. Vortrag beim Lebrgang .. Abwasserbebandlung in der Erui1hrungs- und
Getrllnkeindustrie", Technische Akademie Esslingen, Ostfildern, 11JI2.10.1990
Beccari, M., Passino, R, Ramadori, R and Tandoi, V. (1983) Kinetics of dissimilatory nitrate and nitrite reduction in suspended
growth culture. J. Wat. Poilu!. Control Fed., 55, 58-64
Balmelle, B., Nguyen, K. M., Capdeville, B., Cornier, J. C. and Degiun, A. (1992) Study of factors controlling nitrite build-up in
biological processes for water nitrification. Wat. Sci. Techn., 26, No. 5-6, 1017-1025
Bergeron, P. (1978) Untersuchungen zur Kinetik der Nitriflkation. Karlsruher Berichte zur Ingenieurbiologie, Heft 12
Chen, S. K., Juaw, C. K. and Cheng, S. S. (1991) Nitrification and denitrification of high-strength ammonium and nitrite
wastewater with biofilm reactors. Wat. Sci. Tech., 23, No. 1-3, 1417-1425
Dombrowski, T. (1991) Kinetik der Nitrif1kation und Reaktionstecbnik der Stickstoffeliminierung aus hochbeIasteten Abwllssern.
Forschungsberichte VD!, Reihe 15, Nr. 87, VOl-Verlag Diisseldorf
Nanninga, H. J. and Gottschal, J. C. (1986) Anaerobic purification of wastewater from a potato-starch producing factory. Water
Res., 20, 97-103
Neufeld, R. D., Hill, A. J. and Adekoya, D. O. (1980) Phenol and free ammonia inhibition to Nitrosomonas avtivity. Wat. Res., 14,
1695-1703
Nyhnis, O. (1985) Beitrag zu den MOglichkeiten der Abwasserbehandlung bei Abwllssem mit erMhten Stickstoffkonzentrationen.
Verliffentlichungen des Institutes /fir Siedlungswasserwirtschaft und Abfalltechnik, University Hannover, FRG, 61,
Randall, C. W. and Buth, D. (1984) Nitrite build-up in activated sludge resulting from temperature effects. J. Wat. Pollut. Control
Fed., 56, 1039-1044
Seyfried, C. F. and Austermann-Haun, U. (1990) Large scale anaerobic/aerobic treatment plants for wastewaters from a molasses
distillery, pectin-factory and waste of starch factories. Wat. Sci. Techn., 22, No. In, 353-360
Tunmermans, P. and Van Haute, A. (1983) Denitrification with methanol. Hyphomicrobrium sp. Wat. Res., 17, 1249-1255
Van Versefeld, H. W., Meijer, E. M. and Stouthamer, A. H. (1977) Energy conservation during nitrate respiration in Paracoccus
denitrificans. Arch Microbiol., 112, 17-23
Yang, L. and Alleman, J. E. (1992) Investigation of batchwise nitrite build-up by enriched nitrification culture. Wat. Sci. Techn.
26, No. 5-6, 997-1005
Reproduced with permission of copyright
owner. Further reproduction prohibited
without permission.
and CEBs on Membrane Fouling
Joshua Snowdon 1; Kripa S. Singh 2; and Gustavo Zanatta 3
Abstract: An anaerobic continuously stirred tank reactor (CSTR) integrated with an external tubular nanofiltration membrane module was
operated for 466 days. Studies were conducted to compare the efficiencies in flux and operational permeability recovery of two different
chemical cleaning processes. Chemically enhanced backwashes (CEBs) were applied for approximately 176 days of the study and their
influence on chemical cleaning performance and membrane fouling rate following chemical cleanings were also investigated. The anaerobic
nanofiltration membrane bioreactor used in the study was fed a high-strength, potato starch–based synthetic wastewater (average chemical
oxygen demand of 83.4 g=L and average total suspended solids of 5.8 g=L). Important findings were that a static, one-phase cleaning process
using caustic cleaning chemicals outperformed a dynamic, three-phase cleaning process that also incorporated acidic cleaning chemicals; the
acidic phase of the three-phase cleaning had a detrimental effect on the overall cleaning procedure’s performance; CEBs enhanced the per-
formance of cleaning processes; and, finally, caustic cleaning solutions provided benefits to both membrane cleaning effectiveness and rate of
fouling in the short term but may lead to an increased rate of membrane fouling in the long term. DOI: 10.1061/(ASCE)EE.1943-
7870.0001524. © 2019 American Society of Civil Engineers.
Author keywords: Anaerobic membrane bioreactor; Chemical cleaning; Chemically enhanced backwash; Membrane fouling;
Nanofiltration membrane.
Introduction anaerobic systems such as high concentration of biomass retention

(as the effectiveness of biomass retention becomes independent of
With an ever-growing population, sustainable water and waste- sludge settleability due to membranes’ near-complete solid–liquid
water treatment technologies will become essential in solving water separation capabilities), compact design, and, most importantly, the
scarcity problems prone to develop in upcoming years. Anaerobic ability to produce a superior quality effluent for possible recycle
membrane bioreactors (AnMBRs) offer a promising means of or high-quality discharge to meet stringent wastewater disposal
sustainable and value-added wastewater treatment, coalescing the regulations (Skouteris et al. 2012).
benefits of anaerobic digestion for the treatment of municipal and What continues to remain the most significant factor limiting the
industrial wastewaters with the almost 100% solid–liquid separa- cost effectiveness of membranes in wastewater treatment applica-
tion potential of membrane integrated wastewater treatment tech- tions (both municipal and industrial) is membrane fouling (Ramos
nologies. Anaerobic digestion for wastewater treatment provides et al. 2014). Membrane fouling may be defined as the combination
certain benefits over more commonly used conventional aerobic of processes leading to flux deterioration and transmembrane pres-
technologies, such as high-rate suspended-growth systems, very sure (TMP) increase due to surface or internal pore blockage of the
high removal efficiencies of biochemical oxygen demand (BOD) membrane (Lousada-Ferreira et al. 2014). Surface or internal pore
and chemical oxygen demand (COD), resource recovery (value- blockage is generated through the contact and deposition of sus-
added products such as methane in its biogas and the possible pended particles, colloids, and solutes present in the mixed liquor
recovery of select biochemicals such as volatile fatty acids in its on the membrane surface or in its pores, henceforth referred to as
mixed-liquor and effluent), lower sludge yield, and low carbon foulants (Iorhemen et al. 2016). Membrane fouling increases TMP
footprint (Kinnunen et al. 2015; Meng et al. 2009; Ozgun et al. and decreases flux, while the control of foulant deposition increases
2013; Pilli et al. 2016; Skouteris et al. 2012). Through the integra- manpower and energy requirements, demands costly chemical
tion of membrane separation technologies within anaerobic di- cleaning (that further requires waste handling), and may also
gestion, further benefits are provided over non-membrane-based result in increased deterioration of the membrane material, affecting
its lifespan (Ramos et al. 2014). Foulant mitigation for a membrane
1
Graduate Student, Dept. of Civil Engineering, Univ. of New module is a two-pronged approach: first, the membrane-fouling
Brunswick, 17 Dineen Dr., Fredericton, NB, Canada E3B 5A3 (corre- rate should be minimized, and second, technologies to restore the
sponding author). Email: h557d@unb.ca flux of a fouled membrane must be implemented (Liao et al. 2006).
2
Professor, Dept. of Civil Engineering, Univ. of New Brunswick, 17 Physical cleaning processes (e.g., gas scouring, backwashing,
Dineen Dr., Fredericton, NB, Canada E3B 5A3. Email: singhk@unb.ca relaxation) have only proven to be effective at removing some pore-
3
Process Engineer, ADI Systems, Inc., 370 Wilsey Rd., Fredericton,
clogging and cake-layer membrane foulants (Liao et al. 2006).
NB, Canada E3B 6E9. Email: gustavo.zanatta@evoqua.com
Note. This manuscript was submitted on July 18, 2018; approved on Chemical cleanings, using low- to moderate-strength corrosive or
October 26, 2018; published online on March 6, 2019. Discussion period caustic solutions, are highly effective at removing both the gel layer
open until August 6, 2019; separate discussions must be submitted and cake layer developed through solids deposition on the mem-
for individual papers. This paper is part of the Journal of Environmental brane surface and pore-blocking foulants, making them typically
Engineering, © ASCE, ISSN 0733-9372. more efficient at recovering membrane flux. These chemical
J. Environ. Eng., 2019, 145(5): 04019018

J. Environ. Eng., 2019, 145(5): 04019018

• No cleaning processes were able to recover clean water flux Leedy, P., and J. Ormrod. 2005. Practical research: Planning and design.
and operational permeability very efficiently (with the highest New York: Pearson.
clean water flux recovery being 53%). However, all cleaning Liao, B. Q., J. T. Kramer, and D. M. Bagley. 2006. “Anaerobic membrane
processes were able to recover mixed-liquor flux and opera- bioreactors: Applications and research directions.” Crit. Rev. Environ.
tional permeability effectively (with a lowest mixed-liquor flux Sci. Technol. 36 (6): 489–530. https://doi.org/10.1080/1064338060
0678146.
recovery of 67%).
Lousada-Ferreira, M., J. B. van Lier, and J. H. J. M. van der Graaf. 2014.
• The citric acid portion of Cleaning Process 6 had a detrimental
“Impact of suspended solids concentration on sludge filterability in full-
effect on its overall foulant removal effectiveness. scale membrane bioreactors.” J. Membr. Sci. 476: 69–75. https://doi.org
• Daily CEBs enhanced the foulant removal of successive chemi- /10.1016/j.memsci.2014.11.012.
cal cleaning processes applying the same cleaning conditions. Meng, F., S. R. Chae, A. Drews, M. Kraume, H. S. Shin, and F. Yang. 2009.
• Daily CEBs decreased the rate of foulant accumulation. “Recent advances in membrane bioreactors (MBRs): Membrane fouling
• Considering these results and those of Snowdon et al. (2018), and membrane material.” Water Res. 43 (6): 1489–1512. https://doi.org
daily low-strength CEBs were observed to be an effective strat- /10.1016/j.watres.2008.12.044.
egy for both membrane fouling preventive maintenance and as Metcalf & Eddy, Inc.. 1991. Chap. 10 in Wastewater engineering treat-
assistance in chemical cleaning recovery operations. ment, disposal, and reuse, 987–988. New York: McGraw-Hill.
Ozgun, H., R. Kaan Dereli, M. Evren Ersahin, C. Kinaci, H. Spanjers, and
J. B. van Lier. 2013. “A review of anaerobic membrane bioreactors for
Acknowledgments municipal wastewater treatment: Integration options, limitations, and
expectations.” Sep. Purif. Technol. 118: 89–104. https://doi.org/10
The authors would like to acknowledge and thank the following .1016/j.seppur.2013.06.036.
funding agencies and industrial collaboration: the Natural Science Pilli, S., T. T. More, S. Yan, R. D. Tyagi, and R. Y. Surampalli. 2016.
and Engineering Research Council (NSERC), the New Brunswick “Fenton pre-treatment of secondary sludge to enhance anaerobic diges-
Innovation Foundation (NBIF), and ADI Systems Inc. They would tion: Energy balance and greenhouse gas emissions.” Chem. Eng. J.
also like to thank Jonathan Dargavel and Dr. Dennis Connor for 283: 285–292. https://doi.org/10.1016/j.cej.2015.07.056.
their help and contributions with data collection and testing Ramos, C., F. Zecchino, D. Ezquerra, and V. Diez. 2014. “Chemical clean-
throughout the duration of the study. ing of membranes for an anaerobic membrane bioreactor treating food
industry wastewater.” J. Membr. Sci. 458: 179–188. https://doi.org/10
.1016/j.memsci.2014.01.067.
References Ruigomez, I., E. Gonzalez, S. Guerra, L. E. Rodriguez-Gomez, and L. Vera.
2017. “Evaluation of a novel physical cleaning strategy based on HF
APHA (American Public Health Association). 2005. Standard methods for membrane rotation during the backwashing/relaxation phases for
the examination of water and wastewater. Washington, DC: APHA. anaerobic submerged MBR.” J. Membr. Sci. 526: 181–190. https://doi
Barker, D. J., and D. C. Stuckey. 1998. “A review of soluble microbial .org/10.1016/j.memsci.2016.12.042.
products (SMP) in wastewater treatment systems.” Water Res. 33 (14): Skouteris, G., D. Hermosilla, P. Lopez, C. Negro, and A. Blanco. 2012.
3062–3082. https://doi.org/10.1016/S0043-1354(99)00022-6. “Anaerobic membrane bioreactors for wastewater treatment: A review.”
Cai, W., J. Liu, X. Zhang, W. J. Ng, and Y. Liu. 2016. “Generation of Chem. Eng. J. 198–199: 138–148. https://doi.org/10.1016/j.cej.2012.05
dissolved organic matter and byproducts from activated sludge during .070.
contact with sodium hypochlorite and its implications to on-line chemi- Snowdon, J., K. S. Singh, and G. Zanatta. 2018. “Optimization of an
cal cleaning in MBR.” Water Res. 104: 44–52. https://doi.org/10.1016/j external nanofiltration anaerobic membrane bioreactor treating a high-
.watres.2016.07.065. strength starch-based wastewater.” J. Environ. Eng. 144 (6): 04018032.
Cai, W., and Y. Liu. 2016. “Enhanced membrane biofouling potential https://doi.org/10.1061/(ASCE)EE.1943-7870.0001376.
by on-line chemical cleaning in membrane bioreactor.” J. Membr. Sci. Sun, Y., Y. Fang, P. Liang, and X. Huang. 2016. “Effects of online chemical
511: 84–91. https://doi.org/10.1016/j.memsci.2016.03.039. cleaning on removing biofouling and resilient microbes in a pilot mem-
Hach Company. 2002. Water analysis handbook. Loveland, CO: Hach brane bioreactor.” Int. Biodeterior. Biodegrad. 112: 119–127. https://
Company. doi.org/10.1016/j.ibiod.2016.05.010.
Hamden de Andrade, L., F. D. dos Santos Mendes, J. Cawettiere Espindola, Wang, J., K. Li, D. Yu, J. Zhang, and Y. Wei. 2017. “Fouling characteristics
and M. C. Santos Amaral. 2013. “Internal versus external submerged and cleaning strategies of NF membranes for the advanced treatment of
membrane bioreactor configurations for dairy wastewater treatment.”
antibiotic production wastewater.” Environ. Sci. Pollut. Res. 24 (10):
Desalin. Water Treat. 52 (16–18): 2920–2932. https://doi.org/10
8967–8977. https://doi.org/10.1007/s11356-015-5408-4.
.1080/19443994.2013.799048.
Wang, Q., H. Zeng, Z. Wu, and J. Cao. 2018. “Impact of sodium hypochlo-
Ibeid, S., M. Elektorowicz, and J. A. Oleszkiewicz. 2016. “Impact of elec-
rite cleaning on the surface properties and performance of PVDF mem-
trocoagulation of soluble microbial products on membrane fouling at
branes.” Appl. Surf. Sci. 428: 289–295. https://doi.org/10.1016/j.apsusc
different volatile suspended solids concentrations.” Environ. Technol.
.2017.09.056.
38 (4): 385–393. https://doi.org/10.1080/09593330.2016.1195879.
Iorhemen, O. T., R. A. Hamza, and J. H. Tay. 2016. “Membrane bioreac- Young, J., and R. Cowan. 2004. Respirometry for environmental science
tor (MBR) technology for wastewater treatment and reclamation: and engineering. Springdale, AR: SJ Enterprises.
Membrane fouling.” Membranes 6 (2): 33. https://doi.org/10.3390 Zhang, J., H. C. Chua, J. Zhou, and A. G. Fane. 2006. “Factors affecting
/membranes6020033. the membrane performance in submerged membrane bioreactors.”
Jarusutthirak, C., and G. Amy. 2006. “Role of soluble microbial products J. Membr. Sci. 284 (1–2): 54–66. https://doi.org/10.1016/j.memsci
(SMP) in membrane fouling and flux decline.” Environ. Sci. Technol. .2006.06.022.
40 (3): 969–974. https://doi.org/10.1021/es050987a. Zhang, J., S. I. Padmasiri, M. Fitch, B. Norddahl, L. Raskin, and E.
Kinnunen, V., A. Yla-Outinen, and J. Rintala. 2015. “Mesophilic anaerobic Morgenroth. 2007. “Influence of cleaning frequency and membrane his-
digestion of pulp and paper industry biosludge-long-term reactor per- tory on fouling in an anaerobic membrane bioreactor.” Desalination
formance and effects of thermal pretreatment.” Water Res. 87: 105–111. 207 (1–3): 153–166. https://doi.org/10.1016/j.desal.2006.07.009.
https://doi.org/10.1016/j.watres.2015.08.053. Zhou, Z., F. Meng, H. Lu, Y. Li, X. Jia, and X. He. 2014. “Simultaneous
Le-Clech, P., V. Chen, and T. A. G. Fane. 2006. “Fouling in membrane alkali supplementation and fouling mitigation in membrane bioreactors
bioreactors used in wastewater treatment.” J. Membr. Sci. 284 (1–2): by on-line NaOH backwashing.” J. Membr. Sci. 457: 120–127. https://
17–53. https://doi.org/10.1016/j.memsci.2006.08.019. doi.org/10.1016/j.memsci.2014.01.008.
J. Environ. Eng., 2019, 145(5): 04019018

Wat. Sci. Tech. Vol. 28. No.2. pp. 165-176. 1993. 0273-1223/93 $6'00 0'00
Printed in Great Britain. All rights reserved. Copyright © 1993 IAWQ
ANAEROBIC-AEROBIC TREATMENT OF
POTATO-STARCH WASTEWATER
U. Abeling and C. F. Seyfried
Institut for Siedlungswasserwirtscha/t und Abfalltechnik (Water Supply and Treatment of

Solid Wastes). University of Hannover. Welfengarten I. D-3000 Hannover I. Germany
ABSTRACT
Production of potato starch produces wastewaters witb high concentrations of COD, nitrogen and
phosphorus. The best known and most economical solution for tbe nearly full elimination of tbese substances
is tbe two-stage anaerobic-aerobic treatment. The anaerobic pretreatment must only go so far as to maintain
enougb COD for biological nitrogen and phosphorus elimination in tbe aerobic stage. To optimize tbis
process, several tests have been carried out on nitrification/denitrification by means of tbe intermediate
product nitrite. The carbon consumption amounts to only 60% in comparison witb denitrification via nitrate.
The essential parameter for regulating tbe process is tbe concentration of free ammonia in tbe reactor.
Concentrations of I to 5mg NH31l inhibit tbe nitratation but not tbe nitritation. A separation of botb partial
steps is possible. The content of ammonia was controlled by means of continuous NH4 and pH measuring.
The process is suitable for many wastewaters witb low CODrrKN ratios and higb nitrogen concentrations. If
tbere are undesired nitrite peaks in an aerobic treatment plant caused by bigh pH values and temperatures,
specific nitrogen elimination by means of nitrite is a reliable treatment system.
KEYWORDS
Potato-starch wastewater; anaerobic-aerobic treatment; high-strength ammonium wastewater;

nitrification/denitrification via nitrite; nitritation; nitratation; industrial wastewater; nitrogen removal.
INTRODUCTION
In Europe, about 4.7 Mio. t starch and starch decay products are produced each year. The main raw materials
used are com (60% of the production) and potatoes and wheat (nearly 20% each). The largest product group
is the true starches with nearly 30%, followed by glucose (23%), isoglucose (13%) and modified starch
(12%). Most starch is employed in the food industry (for example sweets, drinks, fruit manufacturing - total
57%), the paper and paperboard industry (25%), as well as the chemical and pharmaceutical industry (10%).
Whereas a short time ago most of the attention was directed to production processes and commercializing,
nowadays wastewater avoidance and wastewater treatment are becoming more and more important. The cost
share of advanced wastewater treatment in the total production costs amounts to nearly 15%.
The starch wastewaters have high nitrogen and phosphorus concentrations. In Germany both nutrients must
be largely eliminated, and this will soon apply to the whole of the European Community as far as the
harmonization of environmental protection is concerned.
For purification of industrial wastewaters, especially for starch factories, more and more two-stage
anaerobic-aerobic processes have been used during the last few years. This is useful - for energetic reasons
- when highly concentrated wastewaters are to be treated, as a substantial part of the organic load can be
1 65
166 U. ABELING and C. F. SEYFRIED
removed anaerobically, hereby producing methane gas. Only the pollution finally remaining has to be
eliminated aerobically by using certain aeration energy.
Minimizin� the Eneuy Consumption
The combined anaerobic-aerobic technology is the most economical solution for as long as only the removal
of carbon compounds is concerned. With regard to the elimination of nitrogen and phosphorus, however, this
process is problematic, as no nitrogen and phosphorus removal worth mentioning takes place during the
elimination of the greater part of the carbon compounds in the anaerobic pretreatment. The BODIN ratio
decreased from 9:1 to nearly 1:1, the BOD:P ratio from 50:1 to 5:1. It is no problem to nitrify the
ammonium in the following aerobic treatment stage, but denitrification and biological phosphorus removal
are often incomplete, possibly due to the low content of organic carbon compounds. Phosphorus may be
eliminated by precipitation with lime or metal ions, whereas the chemicaVphysical methods for ammonium
elimination cause many problems with the resulting products. A total nitrification, however, as well as a
denitrification as far as possible and an effluent concentration of 1 or 2mg PIl, is prescribed in Germany for
most industrial wastewaters (alcohol and food production (pectin, potato, starch and sugar processing
industry, slaughter-houses etc.), pulp and paper industry, animal waste).
In order to achieve a total denitrification sometimes external carbon sources (methanol, acetic acid) are used
(very expensive). In practice, anaerobic pretreatment is often evaded by means of a bypass, but this causes a
lower gas production and also sometimes problems concerning the sludge composition (bulking) in the
aerobic stage.
Our investigations aimed to realize a process which enables total denitrification in spite of far-reaching
anaerobic pre-elimination. In contrast to the process which has been used so far, nitrification/denitrification
was carried out with assistance of the intermediate product nitrate. The carbon consumption amounts to only
60% in comparison with denitrification via nitrate. The essential parameter for regulating the process is the
concentration of free ammonia in the reactor (controlled by means of continuous NH4 and pH measuring).
Concentrations of 1 to 5mg NHyt inhibit the nitratation but not the nitritation.
The experiments were carried out in half-technical pilot plants for a period of three years, using the
wastewater of a potato starch factory. At the same time detailed examinations were made in laboratories
with regard to the kinetics and to the inhibition of nitrification/denitrification by NH3 and HN02. The new
process is available for all wastewaters with high ammonium concentrations and pH > 7.5, which often
cause problems concerning the inhibition of nitrogen removal dependent on high nitrite or ammonium
concentrations.
TREATMENT OF POTATO-STARCH WASTEWATER
Wastewater Production
During potato starch production, there are three partial wastewater streams with different concentrations of
COD, TKN and P. Figure 1 shows a flow diagram of potato starch production. The concentrations and
specific wastewater quantities are listed in Table 1.
Potato-starch wastewater 167
t:----I:;;.;;;,;::;;;;�)_---__1 transport and

'---.--./ washing water
starch/potato - protein -coagulation )..-----l potato juice

juice seperation
S��t-----1 process water
Figure I. Flow diagram of potatO staJeh production
TABLE 1. Potato-Starch Production - Wastewater Characteristics and Specific volumes
washing water process water potato juice

(after protein-coagulation)
pH-Value 6 - 7 6 - 7 5 - 6
COD [mg/l] 2,000 - 4,000 4,000 - 8,000 50,000 - 60,000 (25,000 - 30,000)
TKN [mg/l] 150 - 30 150 - 600 3,600 - 4,400 (1,800 - 2,200)
tot. P [mg/l] 20 - 40 20 - 40 800 - 1,000 (400 - 500)
m3/t potatoes 0. 3 - 1.0 0.6 - 1.3 0.8 - 1.0
The washing water resulting from potato washing is recycled. The solid matter (earth, plant parts etc.) is
eliminated in a sedimentation tank, and any lost part of the water is substituted by fresh water continuously.
The longer the campaign lasts, COD and TKN concentration increases, because the potato quality decreases
for the reason of storage. The process water (starch washing water) contains the contamination component
fme starch, which cannot be gained. The main component of the high strength potato juice is protein. It goes
into a protein extraction, where about 50% of the protein is coagulated, separated, dried and packed for sale.
A still highly contaminated wastewater with 25-30g CODn remains.
Modern manufacturers use circular movement of the water during washing and starch extraction to reduce
the specific wastewater quantities to keep the values preferably in the low regions of the range in Table 1.
The application of an effective technique for starch extraction also reduces the concentrations. Anyway, in
the field of potato juice the technical limits are reached by reducing the concentration by 50% by means of
protein extraction. The wastewater percentage is also layed down by the water content of the potatoes and
changes only slightly. Table 2 shows the specific wastewater quantities and concentrations of a potato starch
factory which was opened in 1987 in Germany (Liichow). The treatment capacity amounts to 300 000
t1campaign (end August until mid March).
176 U. ABELINO and C. F. SEYFRIED
remaining nitrite and ammonia concentrations is required. The process is controlled by continuous pH value
and ammonium measurements.
As the pH value must be increased to 8-8.5, the process is suitable above all for wastewaters which are
located in this pH value range. On the one hand, often undesired nitrite upgradings appear which cause
considerable operation break-downs, but with a specific decoupling of nitritation from nitratation, this can
be avoided.
RFERENCES
Abeling, U. and Seyfried, C. F. (1992) Anaerobic-areobic treatment of high-strength ammonium wastewater - nitrogen removal
via nitrite. Wilt. Sci. Techn., 26, No. 5-6, 1007-1015
Anthonisen, A. C. (1974) The effects of the free ammonia and free nitrous acid on the nitrification process. Ph.D. Thesis, Cornell
University, Engineering, agriCUltural
Anthonisen, A. C., Loehr, R C., Prakasam, T. B. S. and Srinatu, E. O. (1976) Inhibition of nitrification by ammonia and nitrous
acid. J. Wat. Poilu!. Control. Fed., 48, 835-852
Austermann-Haun, U. (1990) Stiltkefabriken. Vortrag beim Lebrgang .. Abwasserbebandlung in der Erui1hrungs- und
Getrllnkeindustrie", Technische Akademie Esslingen, Ostfildern, 11JI2.10.1990
Beccari, M., Passino, R, Ramadori, R and Tandoi, V. (1983) Kinetics of dissimilatory nitrate and nitrite reduction in suspended
growth culture. J. Wat. Poilu!. Control Fed., 55, 58-64
Balmelle, B., Nguyen, K. M., Capdeville, B., Cornier, J. C. and Degiun, A. (1992) Study of factors controlling nitrite build-up in
biological processes for water nitrification. Wat. Sci. Techn., 26, No. 5-6, 1017-1025
Bergeron, P. (1978) Untersuchungen zur Kinetik der Nitriflkation. Karlsruher Berichte zur Ingenieurbiologie, Heft 12
Chen, S. K., Juaw, C. K. and Cheng, S. S. (1991) Nitrification and denitrification of high-strength ammonium and nitrite
wastewater with biofilm reactors. Wat. Sci. Tech., 23, No. 1-3, 1417-1425
Dombrowski, T. (1991) Kinetik der Nitrif1kation und Reaktionstecbnik der Stickstoffeliminierung aus hochbeIasteten Abwllssern.
Forschungsberichte VD!, Reihe 15, Nr. 87, VOl-Verlag Diisseldorf
Nanninga, H. J. and Gottschal, J. C. (1986) Anaerobic purification of wastewater from a potato-starch producing factory. Water
Res., 20, 97-103
Neufeld, R. D., Hill, A. J. and Adekoya, D. O. (1980) Phenol and free ammonia inhibition to Nitrosomonas avtivity. Wat. Res., 14,
1695-1703
Nyhnis, O. (1985) Beitrag zu den MOglichkeiten der Abwasserbehandlung bei Abwllssem mit erMhten Stickstoffkonzentrationen.
Verliffentlichungen des Institutes /fir Siedlungswasserwirtschaft und Abfalltechnik, University Hannover, FRG, 61,
Randall, C. W. and Buth, D. (1984) Nitrite build-up in activated sludge resulting from temperature effects. J. Wat. Pollut. Control
Fed., 56, 1039-1044
Seyfried, C. F. and Austermann-Haun, U. (1990) Large scale anaerobic/aerobic treatment plants for wastewaters from a molasses
distillery, pectin-factory and waste of starch factories. Wat. Sci. Techn., 22, No. In, 353-360
Tunmermans, P. and Van Haute, A. (1983) Denitrification with methanol. Hyphomicrobrium sp. Wat. Res., 17, 1249-1255
Van Versefeld, H. W., Meijer, E. M. and Stouthamer, A. H. (1977) Energy conservation during nitrate respiration in Paracoccus
denitrificans. Arch Microbiol., 112, 17-23
Yang, L. and Alleman, J. E. (1992) Investigation of batchwise nitrite build-up by enriched nitrification culture. Wat. Sci. Techn.
26, No. 5-6, 997-1005
Reproduced with permission of copyright
owner. Further reproduction prohibited
without permission.
Bioresource Technology 98 (2007) 3298–3308
Modelling of biological processes during aerobic treatment of piggery

wastewater aiming at process optimisation
a,*
F. Béline , H. Boursier a, M.L. Daumer a, F. Guiziou a, E. Paul b
a
Cemagref, Environmental Management and Biological Treatment of Wastes Unit, 17 av. de Cucillé, CS 64427, F-35044 Rennes Cedex, France
b
INSA Toulouse, Laboratory of Environmental Process Engineering, 135 av. de Rangueil, F-31077 Toulouse Cedex 4, France
Available online 7 September 2006
Abstract
A dynamic mathematical model was developed for the simulation of the aerobic treatment of piggery wastewater. This model includes
the carbon oxidation, the nitrification and the denitrification. According to the experimental results obtained during this study, a modified
version of the activated sludge model No. 1 has been developed. The model includes (1) nitrite as intermediate of nitrification and deni-
trification, (2) the distinction between the anoxic heterotrophic yield and the aerobic heterotrophic yield, respectively equal to 0.53 and 0.6
and (3) the first-order hydrolysis of the slowly biodegradable fraction. The calibration and the validation of the model was performed
using experimental data from three experiments with two piggery wastewaters. A set of kinetic and stoichiometric parameters emerged
from these tests. Except the kinetic of hydrolysis of the slowly biodegradable organic matter varying from 6 to 25 g COD (g COD day)1,
all other parameters were similar for all experiments. The dissolved oxygen concentration was identified as the main variable influencing
the nitrite accumulation during nitrification. In the calibrated model, the oxygen half-saturation coefficient of the ammonium oxidisers
(0.3 g O2 m3) was lower than for the nitrite oxidisers (1.1 g O2 m3), leading to nitrite accumulation when the dissolved oxygen concen-
tration was low. Simulations with the proposed model could be very useful for improved design and management of biological treatment
of piggery wastewaters, particularly in case of partial nitrification to nitrite directly followed by denitrification.
2006 Elsevier Ltd. All rights reserved.
Keywords: Piggery wastewater; Modelling; Nitrification; Denitrification; Carbon oxidation; AMA
1. Introduction batch reactor without decantation phase (modified SBR)

in which the anoxic and aerobic phases alternate. The
The biological nitrogen removal by nitrification and non-biodegradable suspended solids content of the piggery
denitrification generally provides the most economical wastewaters is high, leading to a difficult decantation.
means of controlling nitrogen in wastewater effluents. Con- Thus, no decantation was applied in the reactors and the
sequently, the biological nitrogen removal by nitrification hydraulic residence time is equal to the solids residence
and denitrification is becoming increasingly common for time and ranges between 35 and 50 days. Few scientific
the treatment of piggery wastewaters in France (Béline data related to the piggery wastewater biodegradability
et al., 2004). About 300 treatment units are in operation (Boursier et al., 2005) and to the biological degradation
in France. The treatment unit is generally composed of processes are available in the literature. Consequently,
a mechanical separation of the wastewater (press-auger, the optimization of the cost of treatment, while respecting
decanter centrifuge, . . .) followed by the biological treat- nitrogen removal efficiency, is difficult. Moreover, due
ment of the liquid fraction resulting from the separation. to the complexity of the processes and the temporal
The biological treatment system is a modified sequential fluctuations of the piggery wastewater concentrations
(Béline et al., 2004), the heuristic optimization is rarely effi-
*
Corresponding author. Tel.: +33 2 23 48 21 21; fax: +33 2 23 48 21 15. cient and a model-based approach would be very useful
E-mail address: fabrice.beline@cemagref.fr (F. Béline). (Coelho et al., 2000). In fact, once calibrated to a particular
0960-8524/$ - see front matter 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.biortech.2006.07.004
F. Béline et al. / Bioresource Technology 98 (2007) 3298–3308 3299
Nomenclature
SNH NHþ 4 þ NH3 nitrogen (g N m )

3
SS readily biodegradable organic matter (g O2 m3)
SND soluble biodegradable organic nitrogen XS slowly biodegradable organic matter (g O2 m3)
(g N m3) SI soluble inert organic matter (g O2 m3)
XND particulate biodegradable organic nitrogen XI particulate inert organic matter (g O2 m3)
(g N m3) XBH active heterotrophic biomass (g O2 m3)
XNI particulate inert organic nitrogen (g N m3) XBAI active NHþ 3
4 oxidisers biomass (g O2 m )
SNOI NO 3
2 nitrogen (g N m ) XBAA
active NO2 oxidisers biomass (g O2 m ) 3

SNOA NO3 nitrogen (g N m3) XP particulate inert organic matter from biomass
SO oxygen (g O2 m3) decay (g O2 m3)
wastewater, a model allows to quickly screen many poten- chemical oxygen demand (COD) hydrolysis could signifi-
tial design and management alternatives. cantly affect the denitrification.
For the evaluation and the optimization of the treat- The objective of this work was to develop and to cali-
ment of urban wastewaters, various models have been brate a mathematical model for the simulation of the mod-
developed and edited by the IWA task group (Henze ified SBR, treating piggery wastewaters. First, an analysis
et al., 1987), such as the activated sludge models (ASM). of the processes involved in nitrogen removal was carried
These models are largely used for domestic sewage but out based on the experimental results. Then, a mathemati-
the modelling is overlooked in the treatment of livestock cal model was built and calibrated using the experimental
wastes and industrial effluents (Orhon, 1998). The available data.
activated sludge models were specifically designed and cal-
ibrated for urban wastewater treatment, and therefore, the 2. Methods
extension of such models to more concentrated influents
such as piggery wastewaters, appears particularly interest- 2.1. Description of the pilot-scale reactor
ing but requires specific experiments and research.
In contrast to the domestic sewage, a very limited A pilot-scale modified SBR was built for the experimen-
amount of data on the kinetic and stoichiometric processes tal study of the treatment of piggery wastewaters. A simpli-
and on the influent fractionation are available for piggery fied scheme of this pilot plant is shown in Fig. 1. The
wastewaters (Andreottola et al., 1997). The ASM models modified SBR consists of a succession of cycles, during
describe the processes of nitrification and denitrification which the aerobic and anoxic phases alternate. Before the
as one-step processes. However, nitrogen removal for con- biological treatment in the reactor, the piggery wastewaters
centrated wastewaters, as the piggery wastewaters, might were stored at 4 C. At the beginning of each cycle, a
result in the inhibition of nitrification by free ammonia
and nitrous acid (Anthonisen et al., 1976), and nitrification
might be limited by the dissolved oxygen (DO) concentra-
tion (Wiesmann, 1994). Therefore, the simplified assump-
tion, used in the ASM models, is no longer applicable
for the description of nitrogen removal processes in this
case. In fact, the free ammonia inhibition and/or the
nitrous acid inhibition and/or the dissolved oxygen limita-
tion frequently lead to nitrite accumulation. High temper-
atures can also lead to nitrite accumulation (Hellinga
et al., 1998). Consequently, more sophisticated model
assumptions are required for an adequate description of
the system efficiency. In this case, each process must
be described as a two-step process including nitrite as
intermediate (Andreottola et al., 1997). Moreover, some
authors studied and compared the heterotrophic yields
under aerobic and anoxic conditions (Gujer et al., 1999;
Orhon et al., 1996; Boursier et al., 2004). The observed
lower anoxic heterotrophic yield leads to a lower COD to Fig. 1. Scheme of the pilot, treating piggery wastewaters: (1) storage tank
(500 l), (2) buffer tank and weighing system, (3) reactor (100–110 l), (4)
NO 3 N ratio for the denitrification. That is more partic- foam breaker, (5) mixing system, (6) fine bubbles diffuser, (7) pH, ORP,
ularly important for the wastewaters with high N concen- DO and temperature sensors, (8) system for water addition, (9) gas flow
tration where the kinetics and the stoichiometry of the meter and (10) withdrawal and weighing system.
3300 F. Béline et al. / Bioresource Technology 98 (2007) 3298–3308
known amount of piggery wastewater was added in the the reactor reached steady-state conditions (after more
reactor through a buffer tank where the quantity added than three hydraulic residence times), the characterisation
was weighed and recorded. The reactor was equipped with of the effluent was done (COD, soluble COD and total
a foam breaker working during the aerobic phases and a nitrogen) and was followed by a study of the evolution
fine bubbles diffuser fed with a compressor (2–5 m3 h1). of NHþ
4 , NO2 and NO3 in the reactor, during one
For the mixing, a centrifugal pump (30 m3 h1) circulated cycle.
the activated sludge, during the anoxic phases from the Immediately after run 1, a second study of the evolution

bottom to the top of the reactor. Water was continuously of NHþ 4 , NO2 and NO3 , during one cycle, was performed
added in the reactor (65 l per day) to compensate for the (run 2). At the beginning of this cycle, 1 l of ammonium
evaporation caused by the high temperature and the aera- carbonate solution (13.5 kg N m3) was added in the reac-
tion. At the end of each cycle, an amount of the activated tor during the feeding phase, together with the piggery
sludge equal to the piggery wastewater added at the begin- wastewater (PW1). For this experiment, the other experi-
ning of the cycle was withdrawn from the reactor. The bio- mental conditions were similar to run 1.
logical reactor was equipped with on-line measurements of For run 3, as for run 1, the reactor was initially filled
DO (WTW Oxi197 oxygen meter), temperature (Ponselle with 50 l of activated sludge coming from the same farm
PONAPF-TTRANS), oxidation–reduction potential (Pon- and 50 l of tap water. About 600 l of raw piggery waste-
selle PONAPF-EHTRANS) and pH (Ponselle PONAPF- water was collected in an experimental farm near Rennes
PHTRANS) allowing measurements at 10 min. intervals. (France) and was mechanically sieved (1 mm) using a
The system operated without sludge recirculation nor rotary screen. The liquid fraction called piggery wastewater
decantation in the reactor, leading to a solids retention 2 (PW2) was stored at 4 C and was used for run 3. For this
time equal to the hydraulic retention time. For each exper- run, the duration of a cycle was equal to 12 h, consisting of
iment, each whole-cycle of the modified SBR was scheduled 7 h of anoxic conditions and 5 h of aerobic conditions. At
as follows: the beginning of each cycle, 4.4 l of PW2 was added in the
reactor, leading to a working volume of 104.4 l and a HRT
• a feeding phase, equal to 11.9 days. At the end of each cycle, the same
• an anoxic phase, amount of activated sludge was withdrawn. A characterisa-
• an aerobic phase, tion of the effluent (COD, soluble COD and total nitrogen)

• a withdrawal phase. and a study of the evolution of NHþ 4 , NO2 and NO3 dur-
ing a cycle, were performed when more than 3 hydraulic
residence times had passed.
2.2. Experimental The main characteristics of each piggery wastewater
used in this study are presented in Table 2.
During this study, three experiments were carried out
using the pilot-scale modified SBR (Table 1).
Table 2
For run 1, the reactor was filled with 50 l of activated Characteristics of the piggery wastewaters
sludge coming from a farm equipped with a modified
Variable PW1 PW2
SBR at the beginning of the experiment. The reactor
3
was supplemented with 50 l of tap water. At the same TS, kg m 54.0 (0.4) 31.1 (–)
VS, kg m3 42.5 (1.0) 20.1 (–)
time, about 600 l of raw piggery wastewaters were col-
COD, kg O2 m3 50.6 (0.3) 39.4 (1.0)
lected and were manually sieved (5 mm mesh). The liquid CODs, kg O2 m3 21.8 (0.1) 14.8 (0.4)
fraction called piggery wastewater 1 (PW1) was stored VFA, kg O2 m3 10.7 (0.2) 6.2 (–)
3
at 4 C and was used for this experiment. For this run, NHþ 4 N (or SNH), kg N m 3.0 (0.1) 3.4 (0.1)

the duration of a whole-cycle was equal to 24 h consisting NO2 N (or SNOI), kg N m3 0 0
NO 3 N (or SNOA), kg N m
3
0 0
of 4 h of anoxic conditions and 20 h of aerobic conditions.
TN–N, kg N m3 4.6 (0.1) 4.3 (0.05)
At the beginning of each cycle, 10 l of PW1 were added SS, kg O2 m3 5.2 (0.5) 6.0 (1.0)
in the reactor, leading to a working volume of 110 l and XS, kg O2 m3 14.1 (3.6) 23.9 (1.4)
a HRT equal to 11 days. At the end of each cycle the Inert COD, kg O2 m3 31.3 (3.4) 9.5 (1.4)
same amount of activated sludge was withdrawn. When Standard errors shown in parenthesis, n = 5.
Table 1
Summary of the conditions during the three experiments carried out with the pilot-scale modified SBR
Runs PW HRT (days) P1 (h) P2 (h) P3 (h) P4 (h) Observations
1 1 11.0 0.1 3.9 19.9 0.1 Effluent characterisation and study of a cycle
2 1 11.0 0.1 3.9 19.9 0.1 Study of a cycle (ammonium carbonate was added)
PW: piggery wastewater, P1: feeding phase, P2: anoxic phase, P3: aerobic phase, P4: withdrawal phase.
A.2. Process stoichiometry for organic matter oxidation, nitrification and denitrification
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
SI SS XI XS XBHI XBAI XBAA XP SO SNOI SNOA SNH SND XND XNI
1 ð1Y H Þ
q1 YH 1 YH iBN
1 1Y HD
q2 Y HD 1 1:71Y HD
iBN
1 1Y HD 1Y HD
q3 Y HD 1 1:14Y HD
1:14Y HD
iBN
q4 1 3:43Y
Y AI
AI 1
Y AI iBN Y1AI
q5 1 1:14Y
Y AA
AA
Y 1AA 1
Y AA iBN
q6 1 fP 1 fP iBN fP fP Æ iBN
Æ iBN
Æ iBN
Æ iBN
q9 1 1
q10 1 1
q11 1 1
A.3. Default values at 20 C for the model and parameters used for simulating run 1, 2 and 3 (-: default value was used)
Symbol Definition Default Run 1 Run 2 Run 3 Ref.

value
YH Aerobic yield heterotrophic 0.6 – – – Boursier et al. (2004)
biomass (g COD g COD1)
YHD Anoxic yield heterotrophic 0.53 – – – Boursier et al. (2004)
lH Maximum specific growth rate for 6 – – – Henze et al. (2000)
heterotrophic biomass (day1)
bH Decay coefficient for heterotrophic 0.62 – – – Henze et al. (2000)
biomass (day1)
KS SS half-saturation coefficient for 20 – – – Henze et al. (2000)
heterotrophic biomass (g COD m3)
KOH SO half-saturation coefficient for XBH 0.2 0.05 0.05 0.05 Henze et al. (2000)
(g COD m3)
KNH SNH half-saturation coefficient 0.05 – – – Henze et al. (2000)
for biomass (g N m3)
KNOI SNOI half-saturation coefficient for XBH 0.5 – – – Henze et al. (2000)
under anoxic conditions (g N m3)
KNOA SNOA half-saturation coefficient for 0.5 – – – Henze et al. (2000)
gG Correction factor for lH under anoxic 0.8 – – – Henze et al. (2000)
condition (dimensionless)
gH Correction factor for hydrolysis under 0.4 – – – Henze et al. (2000)
anoxic condition (dimensionless)
KH Maximum specific hydrolysis rate 3 25 25 6 Boursier et al. (2004)
(g COD (g COD day)1)
KA Ammonification rate 0.08 – – – Henze et al. (2000)
(m3 COD (g day)1)
(continued on next page)
Appendix (continued)
value
fp Fraction of biomass leading to inert 0.08 – – – Henze et al. (2000)
particulate products (dimensionless)
iBN Mass of nitrogen per mass of 0.086 – – – Henze et al. (2000)
COD (g N g COD1)
YAI Yield for NHþ 4 oxidisers autotrophic 0.15 – – – Hao et al. (2002b);
biomass (g COD g N1) Hao et al. (2002a)
lAI Maximum specific growth rate for NHþ 4 0.8 – – – Hao et al. (2002b);
oxidisers autotrophic biomass (day1) Hao et al. (2002a)
bAI Decay coefficient for XBAI autotrophic 0.05 – – – Hao et al. (2002b);
biomass (day1) Hao et al. (2002a)
KOAI SO half-saturation coefficient for XBAI 0.6 0.3 0.3 0.3 Hao et al. (2002b);
biomass (g COD m3) Hao et al. (2002a)
KNAI SNH half-saturation coefficient for XBAI 2.4 5 5 5 Hao et al. (2002b);
biomass (g N m3) Hao et al. (2002a)
YAA Yield for NO 2 oxidisers autotrophic 0.041 – – – Hao et al. (2002b);
lAA Maximum specific growth rate for NO 2 0.79 – – – Hao et al. (2002b);
bAA Decay coefficient for XBAA autotrophic 0.033 – – – Hao et al. (2002b);
KOAA SO half-saturation coefficient for XBAA 2.2 1.1 1.1 1.1 Hao et al. (2002b);
KNAA SNOI half-saturation coefficient for XBAA 5 2.5 2.5 2.5 Hao et al. (2002b);
References Bremner, J.M., Keeney, D.R., 1965. Steam distillation methods for
determination of ammonium, nitrate and nitrite. Anal. Chim. Acta
Abeling, U., Seyfried, C.F., 1992. Anaerobic–aerobic treatment of high- 32, 485–495.
strength ammonium waste-water nitrogen removal via nitrite. Water Choi, E., Eum, Y., 2002. Strategy for nitrogen removal from piggery
Sci. Technol. 26 (5–6), 1007–1015. waste. Water Sci. Technol. 46 (6–7), 347–354.
Andreottola, G., Bortone, G., Tilche, A., 1997. Experimental validation of Coelho, M.A.Z., Russo, C., Araujo, O.Q.F., 2000. Optimization of a
a simulation and design model for nitrogen removal in sequencing sequencing batch reactor for biological nitrogen removal. Water Res.
batch reactors. Water Sci. Technol. 35 (1), 113–120. 34, 2809–2817.
Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S., Srinath, E.G., 1976. Gujer, W., Henze, M., Mino, T., van Loosdrecht, M.C.M., 1999.
Inhibition of nitrification by ammonia and nitrous acid. J. Water Activated sludge model N3. Water Sci. Technol. 39 (1), 183–193.
Pollut. Con. F. 48 (5), 835–852. Hao, X., Heijnen, J.J., Van Loosdrecht, M.C.M., 2002a. Model-based
APHA, 1992. Standard methods for the examination of water and evaluation of temperature and inflow variations on a partial nitrifica-
wastewater. 18th ed., American Public Health Association, Washing- tion-anammox biofilm process. Water Res. 36, 4839–4889.
ton DC, USA. Hao, X., Heijnen, J.J., van Loosdrecht, M.C.M., 2002b. Sensitivity
Béline, F., Daumer, M.L., Guiziou, F., 2004. Biological aerobic treatment analysis of a biofilm model describing a one-stage completely
of pig slurry in france: nutrients removal efficiency and separation autotrophic nitrogen removal (canon) process. Biotechnol. Bioeng.
performances. Trans. ASAE 47 (3), 857–864. 77 (3), 266–277.
Bernet, N., Dangcong, P., Delgenes, J.P., Moletta, R., 2001. Nitrification Hellinga, C., Schellen, A.A.J.C., Mulder, J.W., van Loosdrecht, M.C.M.,
at low oxygen concentration in biofilm reactor. J. Environ. Eng. ASCE Heijnen, J.J., 1998. The SHARON process: an innovative method
127 (3), 266–271. for nitrogen removal from ammonium-rich wastewater. Water Sci.
Boursier, H., 2003. Etude et modélisation des processus biologiques au Technol. 37 (9), 135–142.
cours du traitement aérobie du lisier de porcs en vue d’une optimisa- Henze, M., Grady, C.J., Gujer, W., Marais, G.V., Matsuo, T. 1987.
tion du procédé (modelling of biological processes during aerobic Activated sludge model N1. IWAPRC Scientific and Technical
treatment of pig slurry aiming at process optimisation). PhD thesis, Report N1, London.
Institut National des Sciences Appliquées de Toulouse, France. Henze, M., Gujer, W., Mino, T., van Loosdrecht, M.C.M., 2000.
Boursier, H., Béline, F., Paul, E., 2004. Activated sludge model no. 1 Activated sludge models ASM1, ASM2, AsM2d and ASM3. Tech.
calibration for piggery wastewater using respirometry. Water Sci. Rep. 9, IWA Scientific and Technical Report, London.
Technol. 49 (5–6), 389–396. Huang, J., Hao, O.J., 1996. Alternating aerobic-anoxic process for nitrogen
Boursier, H., Béline, F., Paul, E., 2005. Piggery wastewater characterisation removal: dynamic modeling. Water Environ. Res. 68 (1), 94–104.
for biological nitrogen removal process design. Bioresource Technol. Kim, J.H., Chen, M., Kishida, N., Sudo, R., 2004. Integrateds real-
96, 351–358. time control strategy for nitrogen removal in swine wastewater
known amount of piggery wastewater was added in the the reactor reached steady-state conditions (after more
reactor through a buffer tank where the quantity added than three hydraulic residence times), the characterisation
was weighed and recorded. The reactor was equipped with of the effluent was done (COD, soluble COD and total
a foam breaker working during the aerobic phases and a nitrogen) and was followed by a study of the evolution
fine bubbles diffuser fed with a compressor (2–5 m3 h1). of NHþ
4 , NO2 and NO3 in the reactor, during one
For the mixing, a centrifugal pump (30 m3 h1) circulated cycle.
the activated sludge, during the anoxic phases from the Immediately after run 1, a second study of the evolution

bottom to the top of the reactor. Water was continuously of NHþ 4 , NO2 and NO3 , during one cycle, was performed
added in the reactor (65 l per day) to compensate for the (run 2). At the beginning of this cycle, 1 l of ammonium
evaporation caused by the high temperature and the aera- carbonate solution (13.5 kg N m3) was added in the reac-
tion. At the end of each cycle, an amount of the activated tor during the feeding phase, together with the piggery
sludge equal to the piggery wastewater added at the begin- wastewater (PW1). For this experiment, the other experi-
ning of the cycle was withdrawn from the reactor. The bio- mental conditions were similar to run 1.
logical reactor was equipped with on-line measurements of For run 3, as for run 1, the reactor was initially filled
DO (WTW Oxi197 oxygen meter), temperature (Ponselle with 50 l of activated sludge coming from the same farm
PONAPF-TTRANS), oxidation–reduction potential (Pon- and 50 l of tap water. About 600 l of raw piggery waste-
selle PONAPF-EHTRANS) and pH (Ponselle PONAPF- water was collected in an experimental farm near Rennes
PHTRANS) allowing measurements at 10 min. intervals. (France) and was mechanically sieved (1 mm) using a
The system operated without sludge recirculation nor rotary screen. The liquid fraction called piggery wastewater
decantation in the reactor, leading to a solids retention 2 (PW2) was stored at 4 C and was used for run 3. For this
time equal to the hydraulic retention time. For each exper- run, the duration of a cycle was equal to 12 h, consisting of
iment, each whole-cycle of the modified SBR was scheduled 7 h of anoxic conditions and 5 h of aerobic conditions. At
as follows: the beginning of each cycle, 4.4 l of PW2 was added in the
reactor, leading to a working volume of 104.4 l and a HRT
• a feeding phase, equal to 11.9 days. At the end of each cycle, the same
• an anoxic phase, amount of activated sludge was withdrawn. A characterisa-
• an aerobic phase, tion of the effluent (COD, soluble COD and total nitrogen)

• a withdrawal phase. and a study of the evolution of NHþ 4 , NO2 and NO3 dur-
ing a cycle, were performed when more than 3 hydraulic
residence times had passed.
2.2. Experimental The main characteristics of each piggery wastewater
used in this study are presented in Table 2.
During this study, three experiments were carried out
using the pilot-scale modified SBR (Table 1).
Table 2
For run 1, the reactor was filled with 50 l of activated Characteristics of the piggery wastewaters
sludge coming from a farm equipped with a modified
Variable PW1 PW2
SBR at the beginning of the experiment. The reactor
3
was supplemented with 50 l of tap water. At the same TS, kg m 54.0 (0.4) 31.1 (–)
VS, kg m3 42.5 (1.0) 20.1 (–)
time, about 600 l of raw piggery wastewaters were col-
COD, kg O2 m3 50.6 (0.3) 39.4 (1.0)
lected and were manually sieved (5 mm mesh). The liquid CODs, kg O2 m3 21.8 (0.1) 14.8 (0.4)
fraction called piggery wastewater 1 (PW1) was stored VFA, kg O2 m3 10.7 (0.2) 6.2 (–)
3
at 4 C and was used for this experiment. For this run, NHþ 4 N (or SNH), kg N m 3.0 (0.1) 3.4 (0.1)

the duration of a whole-cycle was equal to 24 h consisting NO2 N (or SNOI), kg N m3 0 0
NO 3 N (or SNOA), kg N m
3
0 0
of 4 h of anoxic conditions and 20 h of aerobic conditions.
TN–N, kg N m3 4.6 (0.1) 4.3 (0.05)
At the beginning of each cycle, 10 l of PW1 were added SS, kg O2 m3 5.2 (0.5) 6.0 (1.0)
in the reactor, leading to a working volume of 110 l and XS, kg O2 m3 14.1 (3.6) 23.9 (1.4)
a HRT equal to 11 days. At the end of each cycle the Inert COD, kg O2 m3 31.3 (3.4) 9.5 (1.4)
same amount of activated sludge was withdrawn. When Standard errors shown in parenthesis, n = 5.
Table 1
Summary of the conditions during the three experiments carried out with the pilot-scale modified SBR
Runs PW HRT (days) P1 (h) P2 (h) P3 (h) P4 (h) Observations
2 1 11.0 0.1 3.9 19.9 0.1 Study of a cycle (ammonium carbonate was added)
PW: piggery wastewater, P1: feeding phase, P2: anoxic phase, P3: aerobic phase, P4: withdrawal phase.
A.2. Process stoichiometry for organic matter oxidation, nitrification and denitrification
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
SI SS XI XS XBHI XBAI XBAA XP SO SNOI SNOA SNH SND XND XNI
1 ð1Y H Þ
q1 YH 1 YH iBN
1 1Y HD
q2 Y HD 1 1:71Y HD
iBN
1 1Y HD 1Y HD
q3 Y HD 1 1:14Y HD
1:14Y HD
iBN
q4 1 3:43Y
Y AI
AI 1
Y AI iBN Y1AI
q5 1 1:14Y
Y AA
AA
Y 1AA 1
Y AA iBN
Æ iBN
Æ iBN
Æ iBN
q9 1 1
q10 1 1
q11 1 1
A.3. Default values at 20 C for the model and parameters used for simulating run 1, 2 and 3 (-: default value was used)

value
YH Aerobic yield heterotrophic 0.6 – – – Boursier et al. (2004)
YHD Anoxic yield heterotrophic 0.53 – – – Boursier et al. (2004)
lH Maximum specific growth rate for 6 – – – Henze et al. (2000)
heterotrophic biomass (day1)
bH Decay coefficient for heterotrophic 0.62 – – – Henze et al. (2000)
biomass (day1)
KS SS half-saturation coefficient for 20 – – – Henze et al. (2000)
heterotrophic biomass (g COD m3)
KOH SO half-saturation coefficient for XBH 0.2 0.05 0.05 0.05 Henze et al. (2000)
(g COD m3)
KNH SNH half-saturation coefficient 0.05 – – – Henze et al. (2000)
for biomass (g N m3)
KNOI SNOI half-saturation coefficient for XBH 0.5 – – – Henze et al. (2000)
KNOA SNOA half-saturation coefficient for 0.5 – – – Henze et al. (2000)
gG Correction factor for lH under anoxic 0.8 – – – Henze et al. (2000)
condition (dimensionless)
gH Correction factor for hydrolysis under 0.4 – – – Henze et al. (2000)
anoxic condition (dimensionless)
KH Maximum specific hydrolysis rate 3 25 25 6 Boursier et al. (2004)
(g COD (g COD day)1)
KA Ammonification rate 0.08 – – – Henze et al. (2000)
(m3 COD (g day)1)
(continued on next page)
Appendix (continued)
value
fp Fraction of biomass leading to inert 0.08 – – – Henze et al. (2000)
particulate products (dimensionless)
iBN Mass of nitrogen per mass of 0.086 – – – Henze et al. (2000)
COD (g N g COD1)
YAI Yield for NHþ 4 oxidisers autotrophic 0.15 – – – Hao et al. (2002b);
lAI Maximum specific growth rate for NHþ 4 0.8 – – – Hao et al. (2002b);
bAI Decay coefficient for XBAI autotrophic 0.05 – – – Hao et al. (2002b);
KOAI SO half-saturation coefficient for XBAI 0.6 0.3 0.3 0.3 Hao et al. (2002b);
KNAI SNH half-saturation coefficient for XBAI 2.4 5 5 5 Hao et al. (2002b);
YAA Yield for NO 2 oxidisers autotrophic 0.041 – – – Hao et al. (2002b);
lAA Maximum specific growth rate for NO 2 0.79 – – – Hao et al. (2002b);
bAA Decay coefficient for XBAA autotrophic 0.033 – – – Hao et al. (2002b);
KOAA SO half-saturation coefficient for XBAA 2.2 1.1 1.1 1.1 Hao et al. (2002b);
KNAA SNOI half-saturation coefficient for XBAA 5 2.5 2.5 2.5 Hao et al. (2002b);
References Bremner, J.M., Keeney, D.R., 1965. Steam distillation methods for
determination of ammonium, nitrate and nitrite. Anal. Chim. Acta
Abeling, U., Seyfried, C.F., 1992. Anaerobic–aerobic treatment of high- 32, 485–495.
strength ammonium waste-water nitrogen removal via nitrite. Water Choi, E., Eum, Y., 2002. Strategy for nitrogen removal from piggery
Sci. Technol. 26 (5–6), 1007–1015. waste. Water Sci. Technol. 46 (6–7), 347–354.
Andreottola, G., Bortone, G., Tilche, A., 1997. Experimental validation of Coelho, M.A.Z., Russo, C., Araujo, O.Q.F., 2000. Optimization of a
a simulation and design model for nitrogen removal in sequencing sequencing batch reactor for biological nitrogen removal. Water Res.
batch reactors. Water Sci. Technol. 35 (1), 113–120. 34, 2809–2817.
Anthonisen, A.C., Loehr, R.C., Prakasam, T.B.S., Srinath, E.G., 1976. Gujer, W., Henze, M., Mino, T., van Loosdrecht, M.C.M., 1999.
Inhibition of nitrification by ammonia and nitrous acid. J. Water Activated sludge model N3. Water Sci. Technol. 39 (1), 183–193.
Pollut. Con. F. 48 (5), 835–852. Hao, X., Heijnen, J.J., Van Loosdrecht, M.C.M., 2002a. Model-based
APHA, 1992. Standard methods for the examination of water and evaluation of temperature and inflow variations on a partial nitrifica-
wastewater. 18th ed., American Public Health Association, Washing- tion-anammox biofilm process. Water Res. 36, 4839–4889.
ton DC, USA. Hao, X., Heijnen, J.J., van Loosdrecht, M.C.M., 2002b. Sensitivity
Béline, F., Daumer, M.L., Guiziou, F., 2004. Biological aerobic treatment analysis of a biofilm model describing a one-stage completely
of pig slurry in france: nutrients removal efficiency and separation autotrophic nitrogen removal (canon) process. Biotechnol. Bioeng.
performances. Trans. ASAE 47 (3), 857–864. 77 (3), 266–277.
Bernet, N., Dangcong, P., Delgenes, J.P., Moletta, R., 2001. Nitrification Hellinga, C., Schellen, A.A.J.C., Mulder, J.W., van Loosdrecht, M.C.M.,
at low oxygen concentration in biofilm reactor. J. Environ. Eng. ASCE Heijnen, J.J., 1998. The SHARON process: an innovative method
127 (3), 266–271. for nitrogen removal from ammonium-rich wastewater. Water Sci.
Boursier, H., 2003. Etude et modélisation des processus biologiques au Technol. 37 (9), 135–142.
cours du traitement aérobie du lisier de porcs en vue d’une optimisa- Henze, M., Grady, C.J., Gujer, W., Marais, G.V., Matsuo, T. 1987.
tion du procédé (modelling of biological processes during aerobic Activated sludge model N1. IWAPRC Scientific and Technical
treatment of pig slurry aiming at process optimisation). PhD thesis, Report N1, London.
Institut National des Sciences Appliquées de Toulouse, France. Henze, M., Gujer, W., Mino, T., van Loosdrecht, M.C.M., 2000.
Boursier, H., Béline, F., Paul, E., 2004. Activated sludge model no. 1 Activated sludge models ASM1, ASM2, AsM2d and ASM3. Tech.
calibration for piggery wastewater using respirometry. Water Sci. Rep. 9, IWA Scientific and Technical Report, London.
Technol. 49 (5–6), 389–396. Huang, J., Hao, O.J., 1996. Alternating aerobic-anoxic process for nitrogen
Boursier, H., Béline, F., Paul, E., 2005. Piggery wastewater characterisation removal: dynamic modeling. Water Environ. Res. 68 (1), 94–104.
for biological nitrogen removal process design. Bioresource Technol. Kim, J.H., Chen, M., Kishida, N., Sudo, R., 2004. Integrateds real-
96, 351–358. time control strategy for nitrogen removal in swine wastewater
Fig. 5. Experimental and simulated nitrogen concentrations during the cycles of run 1, 2 and 3.
modification of the parameters was necessary for the 1 and 2, could influence slightly the nitrogen transforma-
modelling of run 2 (Fig. 5). In contrast to run 1 and 2, tions. On the other hand, for run 1, the Ss fraction was equal
a different piggery wastewater composition was used for to 5 g O2/l whereas the VFA were equal to 10 g O2/l. This
run 3. For this run, a reduction of the maximum specific result indicates that a part of the VFA was probably stored
hydrolysis rate from 25 to 6 g COD (g COD day)1 was as internal polymer by the micro-organisms. The effect of
necessary for the modelling (Fig. 5). Similar variations of storage and the internal polymer consumption is lumped
the maximum specific hydrolysis rate were previously in the hydrolysis process, leading to an underestimation of
observed between influents (Orhon, 1998) and could be the Ss fraction. The storage of the organic matter as internal
related to the characteristics of the organic matter. Except polymer was included in the ASM3 model (Gujer et al.,
this value, all other parameters were identical for the mod- 1999). This process could be included in the present model
elling of all runs and are summarized in Appendix A.3. to improve simulation. Adsorption of NHþ 4 to the flocs was
treatment using sequencing batch reactors. Water Res. 38, 3340– Peu, P., Béline, F., Martinez, J., 2004. Volatile fatty acids analysis from
3348. pig slurry using high performance liquid chromatography. Int. J.
Lee, S.M., Jung, J.Y., Chung, Y.C., 2000. Measurement of ammonia Environ. An. Ch. 84 (13), 1017–1022.
inhibition of microbial activity in biological wastewater treatment Sollfrank, U., Gujer, W., 1991. Characterisation of domestic wastewater
process using dehydrogenase assay. Biotechnol. Lett. 22, 991–994. for mathematical modelling of the activated sludge process. Water Sci.
Nielsen, P.H., 1996. Adsorption of ammonium to activated sludge. Water Technol. 23, 1057–1066.
Res. 30 (3), 762–764. Villaverde, S., Fdz-Polanco, F., Garcia, P.A., 2000. Nitrifying biofilm
Oh, J., Silverstein, J., 1999. Acetate limitation and nitrite accumulation acclimatation to free ammonia in submerged biofilters: start-up
during denitrification. J. Environ. Eng. ASCE 125 (3), 234–242. influence. Water Res. 34, 602–610.
Orhon, D., 1998. Evaluation of industrial biological treatment design on Wiesmann, U., 1994. Biological nitrogen removal from wastewater.
the basis of process modelling. Water Sci. Technol. 38 (4–5), 1–8. Adv. Biochem. Eng. Biot. 51, 113–154.
Orhon, D., Okutman, D., 2003. Respirometric assessment of residual organic Wilderer, P.A., Jones, W.L., Dau, U., 1987. Competition in denitrification
matter for domestic sewage. Enzyme Microb. Tech. 32 (5), 560–566. systems affecting reduction rate and accumulation of nitrite. Water
Orhon, D., Sozen, S., Artan, N., 1996. The effect of heterotrophic yield on Res. 21, 239–245.
the assessment of the correction factor for anoxic growth. Water Sci.
Technol. 34 (5–6), 67–74.

Thing Teen818 AnMBBR Reactor 2024

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Thing Teen818 AnMBBR Reactor 2024

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Thing Teen818 AnMBBR Reactor 2024

Uploaded by

Copyright:

Available Formats

Abstract: An anaerobic continuously stirred tank reactor (CSTR) integrated with an external tubular nanofiltration membrane module was

Introduction anaerobic systems such as high concentration of biomass retention

© ASCE 04019018-1 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

Fig. 1. Simplified PFD of experimental setup.

© ASCE 04019018-2 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-4 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-6 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-7 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

Fig. 1. Simplified PFD of experimental setup.

© ASCE 04019018-2 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-3 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-4 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-6 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-8 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-9 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

U. Abeling and C. F. Seyfried

Institut for Siedlungswasserwirtscha/t und Abfalltechnik (Water Supply and Treatment of

Potato-starch wastewater; anaerobic-aerobic treatment; high-strength ammonium wastewater;

Minimizin� the Eneuy Consumption

TREATMENT OF POTATO-STARCH WASTEWATER

14 YKw + 10pll 14 K · 10pll

"t, .. ioniz.tion constant of the _i. equilibrhlD equation K

TABLE 5. Chemical Characterization of Potato Starch Wastewater

ra.. wastewater 17,000 10,000 870 150 190 11.5 19.5

CQntrol 2f the DitritaUIlD

TABLE 6. Results - Nitro!:en Remoyal - Ayera" values <April-June 1991)

COO [mg/l] 1,600 80.0 90.0 70.0

Introduction anaerobic systems such as high concentration of biomass retention

© ASCE 04019018-1 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

Fig. 1. Simplified PFD of experimental setup.

© ASCE 04019018-2 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

© ASCE 04019018-9 J. Environ. Eng.

J. Environ. Eng., 2019, 145(5): 04019018

U. Abeling and C. F. Seyfried

Institut for Siedlungswasserwirtscha/t und Abfalltechnik (Water Supply and Treatment of

Potato-starch wastewater; anaerobic-aerobic treatment; high-strength ammonium wastewater;

Minimizin� the Eneuy Consumption

TREATMENT OF POTATO-STARCH WASTEWATER

t:----I:;;.;;;,;::;;;;�)_---__1 transport and

starch/potato - protein -coagulation )..-----l potato juice

S��t-----1 process water

Figure I. Flow diagram of potatO staJeh production

TABLE 1. Potato-Starch Production - Wastewater Characteristics and Specific volumes

washing water process water potato juice

m3/t potatoes 0. 3 - 1.0 0.6 - 1.3 0.8 - 1.0