Pursuit of urine nitrifying granular sludge for decentralised nitrite production and sewer gas control

Chemical Engineering Journal 289 (2016) 17–27

Contents lists available at ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Pursuit of urine nitrifying granular sludge for decentralised nitrite production and sewer gas control H.R. Mackey a,b,⇑, G. Rey Morito a, T. Hao a, G.-H. Chen a,c a

Department of Civil & Environmental Engineering, The Hong Kong University of Science and Technology, Clear Water Bay, Kowloon, Hong Kong, China College of Science and Engineering, Hamad Bin Khalifa University, Education City, Doha, Qatar c SYSU-HKUST Joint Research Center for Innovative Environmental Technology, Sun Yat-sen University, Guangzhou, China b

h i g h l i g h t s

g r a p h i c a l a b s t r a c t


Nitrifying SBRs operated for high rate

urine oxidation with base addition.
Granular sludge development

explored through various operational parameters.
Small, compact and strong granulelike aggregates developed.
Granulation limited by pH/NH3 and opposing conditions for heterotrophs and nitrifiers.
High-rate stable nitritation achieved at volumetric conversion rate of 1.1 kg-N/m3 d.

a r t i c l e

i n f o

Article history: Received 24 August 2015 Received in revised form 18 December 2015 Accepted 21 December 2015 Available online 28 December 2015 Keywords: Nitrifying aerobic granules Selection pressure pH Free ammonia Urine source separation Microbial competition

a b s t r a c t This study aims to develop a high-rate reactor for decentralised urine nitrification using granular sludge, which allows high biomass retention in a small footprint installation. The system incorporated alkalinity dosing to ensure full conversion of ammonia to nitrite for subsequent control of sewer gas production. The key operational parameters for granule formation such as feeding and settling duration were tested as well as environmental conditions including pH. Pulse feeding was found to be instrumental to both the treatment performance and development of well settling granule-like aggregates. These aggregates showed granular characteristics with a compact and well defined shape, excellent settleability and high strength. However, pulse feeding also led to increased pH and free ammonia concentrations which reduced heterotroph activity and aggregate growth. Furthermore, in order to prevent nitrification collapse and high pH and free ammonia associated with such an event the reactor nitrogen loading was limited. Due to the low organic to nitrogen ratio of urine this restrained the organics loading of the dominant heterotrophic community further limiting granule growth. Nevertheless, the development of compact well settling aggregates under pulse feed operation allowed the objective of high-rate nitrification to be achieved with sustained nitritation rates up to 1.1 kg-N/m3 d at a urine dilution of approximately 25%. Ó 2015 Elsevier B.V. All rights reserved.

Abbreviations: AOB, ammonia oxidising bacteria; bCOD, biodegradable chemical oxygen demand; BSA, bovine serum albumin; COD, chemical oxygen demand; DO, dissolved oxygen; EPS, extracellular polymeric substances; FISH, fluorescent in situ hybridisation; MLVSS, mixed liquor volatile suspended solids; MLSS, mixed liquor suspended solids; N, nitrogen; NLR, nitrogen loading rate; NOB, nitrite oxidising bacteria; RTKN, removal rate of total Kjeldahl nitrogen (%); SVI, sludge volume index; TAN, total ammoniacal nitrogen; TKN, total Kjeldahl nitrogen; TN, total nitrogen; TNN, Total nitrite nitrogen; TSS, total suspended solids; VSS, volatile suspended solids. ⇑ Corresponding author at: College of Science and Engineering, Hamad Bin Khalifa University, Education City, Doha, Qatar. Tel.: +974 4454 5665. E-mail address: [email protected] (H.R. Mackey). http://dx.doi.org/10.1016/j.cej.2015.12.071 1385-8947/Ó 2015 Elsevier B.V. All rights reserved.

18

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

1. Introduction The discharge of nutrients from large urban centres into waterways has historically been one of the main concerns for environmental engineers. To avoid eutrophication of waterways the typical approach is to collect all domestic wastewater for treatment at centralised wastewater treatment plants. However this approach introduces the need for costly sewer networks that are used solely for transport as well as complex and inefficient treatment process configurations for dealing with the mixture of various diluted household waste streams [1]. As urine accounts for approximately 80% of the nitrogen and 50% of the phosphorus of the total nutrient loads present in domestic wastewater [2,3], the decentralised treatment of separated urine could provide an opportunity for efficient nutrient removal and recovery options. A number of processes already exist for separate urine treatment and nutrient recovery, as detailed in Maurer et al. [3]. The process selection depends heavily on a combination of environmental, social and economic aspects [4]. For instance, in areas with nearby agriculture partial-nitrification and vacuum-compression evaporation can be used for complete nutrient recovery as a fertiliser product [5]. In dense urban areas nutrient removal via struvite precipitation, urine nitrification and in-sewer denitrification is another alterative with the following advantages:

either growth inhibition or chemical changes to extracellular polymer structure which holds the granule together [15]. Such a high pH lies outside those previously investigated [13,16,17]. The only previously reported attempt to cultivate granular sludge directly for urine-stabilisation was on strongly diluted urine (200– 250 mg-N/L) by Sun et al. [18]. Direct cultivation was not successful although a reactor was successfully operated using pre-cultivated granules for 132 days at a nitrogen loading rate (NLR) of 0.5 g-N/L d [18]. Potential factors preventing granulation in the study of Sun et al. include a low biodegradable organic loading rate of 0.4 g-COD/L d as a result of the low COD:N ratio, nitrifier washout under the short settling time imposed and continuous feeding which would result in low bulk liquid organic concentrations and high concentration gradients within the biomass. Despite such potential difficulties the present study explores the feasibility of developing urine nitrifying granular sludge, driven both by scientific understanding of previous failure [18] and practical application for decentralised urine nitrification. Therefore, the goal of this study is to explore the feasibility of cultivating urine nitrifying granular sludge by focusing on the key operational parameters of settling time and feeding strategy and other possible influencing factors such as COD:N ratio, loading rate and pH. 2. Materials and methods 2.1. Reactor configuration and operation


Phosphorus removal via struvite precipitation using magnesium addition provides an alternative source of fertilisers [6].
The in situ urine nitrification allows the use of the sewer network as a post-anoxic bioreactor enabling more efficient use of organic carbon in denitrification [7].
In sewer denitrification allows possible simplification of centralised plants to attain high rate carbon removal [8]. In these and other systems urine nitrification plays the important role of converting ammonia, which is formed through the hydrolysis of urea, to nitrate/nitrite to reduce the bulk pH and prevent odour from ammonia gas release. Additionally, aerobic removal of organics is provided. For decentralised urine nitrification a high-rate compact bioreactor is highly desirable, which generally requires retention of high biomass concentrations in the system. Biofilm reactors are commonly used for nitrification systems due to their high degree of biomass retention allowing reduced footprint and resilience against cold climate. Nevertheless mass transfer issues generally limit the overall volumetric loading rates that can be achieved [5,9]. Granular sludge, as a self-immobilised and suspended type of biofilm, increases mass transfer of substrate through the biomass while its rapid settling allows excellent liquid–solid separation and high biomass concentrations. In addition the granular structure enhances nitrite oxidising bacteria (NOB) colonisation [10]. As a result granular sludge has already been demonstrated as an ideal system for high rate nitrification [11]. Furthermore, as granular sludge does not require biofilm support it is a more economical alternative for decentralised treatment. A sequencing batch reactor (SBR) is commonly applied for sludge granulation due to its unique capability in controlling substrate concentrations and sludge settling/decanting flexibility [12]. A few studies have applied granular sludge technology to wastewaters with total ammoniacal nitrogen (TAN) concentrations in excess of 500 mg-N/L [11,13,14], but have significant differences to the current study with regards to higher organic-nitrogen ratios, higher reactor temperatures or use of pre-cultivated granules. A high pH in hydrolysed urine of approximately 9.2, even following significant dilution, may also impact aerobic granulation through

In this study, a three-stage experiment was designed. In Stage 1, a urine-nitrifying SBR, named R1, constructed from Plexiglas with an effective volume of 3.2 L (68 mm diameter, 875 mm high) and an exchange ratio of 26% was adopted as the first reactor to explore possible urine nitrifying granular sludge under trial conditions. Fine bubble aeration was provided at 2 L/min, which gave a superficial air upflow velocity of 0.92 cm/s and at the same time maintained the dissolved oxygen (DO) above 4 mg/L. This air upflow velocity was limited by foaming but was deemed sufficient for sludge granulation based on comparisons of the loading rate and substrate composition to similar studies [19,20]. Each operation cycle was set to 6 h. Aerobic conditions were maintained throughout except for the settling phase. The times assigned for feeding and settling were changed over a number of defined periods, as detailed in Table 1. Seed sludge was collected from a secondary sewage treatment plant in Hong Kong with an initial mixed liquor volatile suspended solids (MLVSS) concentration of 2130 mg/L. The reactor was operated at room temperature (22 ± 2 °C). Base (1 mol/L NaHCO3) was dosed to provide sufficient alkalinity for complete nitrification. The pH setpoint for base dosage was 7. Stage 2 coincided with the final period of operation in R1 when the pH controller was replaced by dosing the necessary amount of NaHCO3 (alkalinity) into the influent. This caused the influent pH to rise from around 9.2 to 9.5–9.6 with an increase in influent free ammonia of between 40% and 50% in the influent. This provided the opportunity to study the effects of pH and free ammonia on the sludge properties. Stage 3 was implemented in order to confirm some of the observations obtained from R1. For this stage sludge was diverted from R1 to two other identical reactors, named R2 and R3, for confirming the effect of feeding pattern and settling time as the two key operational parameters on urine nitrifying granular sludge development. The initial sludge concentration in these two reactors was set at 2550 (±40) mg-MLVSS/L. Both reactors were operated under the same conditions as R1 except that feeding occurred only once per cycle and the exchange volume was reduced slightly to 23%. Feeding in R2 was under a pulse regime and for R3 under a gradual-feed regime to directly compare the role of feeding on

19

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27 Table 1 Reactor operational parameters and test phases in this study.

a b c

Reactor

Stage / period

Day of operation (d)

Influent feeds/cycle

Influent feed duration/feed (min)

Aeration durationa (min)

Settling duration (min)

Critical settling velocity (m/h)

Alkalinity

NLRb (mg-N/L d)

OLRc (mg-COD/L d)

R1 R1 R1 R1 R1 R1 R2 R2 R3 R3

1 1 1 1 1 2 3 3 3 3

0–88 88–394 394–551 551–657 657–815 815–973 0–63 63–165 0–63 63–165

2 2 2 2 2 2 1 1 1 1

67 67 67 38 7 7 6 6 208 208

340 347 353 353 353 353 353 330 353 330

20 13 7 7 7 7 7 30 7 30

0.67 1.06 1.97 1.97 1.97 1.97 1.55 0.37 1.55 0.37

At pH 7 At pH 7 At pH 7 At pH 7 At pH 7 Influent At pH 7 At pH 7 At pH 7 At pH 7

540 1002 902 762 948 1155 464 420 464 330

618 1096 858 838 987 1019 282 293 282 218

i ii iii iv v vi i ii i ii

Includes feeding. Average nitrogen loading rate for period. Average organic loading rate for period.

sludge properties, while settling pressure was decreased in both reactors (a reversal of Stage 1 experiments) to confirm its role on sludge granulation. A summary of the key operational periods and parameters for these two reactors is given in Table 1. Urine used for the study was collected from 20 males and then stored in a 50 L container at 4 °C until use. Chemical characteristics of the fresh urine are provided in Table 2. The retention time of the container was between 1 and 3 weeks and hydrolysis was near complete with TAN typically more than 80% of the influent total nitrogen (TN). As the exchange volume in the reactors was fixed, the nitrogen loading rate (NLR) was controlled by adjusting the urine dilution rate. Urine was diluted with tap water with a maximum urine content used in this study of approximately 25% (1600 mg-N/L). The NLR was periodically adjusted to achieve the maximum loading while maintaining stable performance. Therefore, the NLR rate was increased whenever the reactors were consistently oxidising more than 99% of the original TAN concentration in the influent, and decreased if the oxidation performance was deteriorating.

2.2. Chemical analysis of reactor performance Aqueous chemical analysis was conducted in accordance with the Standard Methods for Water and Wastewater Examination [21] after filtering through 0.45 lm cellulose filters. TAN was measured by flow injection analysis (QuickChem 8500 Series II, Lachat, Milwaukee, WI, USA), TN was measured using a total carbon analyser with TN module (TOC-VCPH plus TNM-1, Shimadzu, Kyoto, Japan) and anions were measured by ion chromatography (LC20A Super with a conductivity detector and an IC-SA2 analytical column, Shimadzu, Kyoto, Japan). COD was measured with Hach

Table 2 Chemical characteristics of urine used in this study collected fresh from donors.

TKN Soluble TN Chloride Phosphate Total COD Soluble COD Sulphate Calcium Magnesium Potassium Sodium Salinity Conductivity

Unit

Mean

Min

Max

Std. dev.

Samples

mg-N/L mg-N/L mg-Cl/L mg-P/L mg-COD/L mg-COD/L mg-S/L mg-Ca/L mg-Mg/L mg-K/L mg-Na/L ppt mS

7300 6330 3120 331 6540 6280 1090 58 8.4 1360 1850 7.3 12.6

5590 2360 279 40 2230 2520 195 29 4.1 1050 757 0.7 1.3

9670 14,100 10,100 1230 15,100 14,700 4000 129 15.3 1890 2930 18.7 30.1

1300 2860 2060 259 3720 3560 806 48 6.1 458 1090 4.6 7.3

10 38 56 61 11 10 31 4 3 3 3 59 59

High Range test kits. DO and pH were measured using a multimeter (Multi3420 with FDO925-3 and Sentix 940-3 probes, WTW, Weilheim, Germany). Calculation of free ammonia and free nitrous acid conditions followed the method described in Udert et al. [22]. The ionic strength of undiluted urine was estimated at 0.51 eq/L and values for diluted urine were assumed proportional to the level of dilution. The pKa values for ammonia and nitrous acid at 22 °C were 9.34 and 3.39 [23,24]. Ionic activity was calculated by the Davies method. Calculation of biomass specific NLRs is described in the Supplementary data. Total Kjeldahl nitrogen (TKN) removal (RTKN) was calculated as a percentage using effluent TN, TNN and nitrate concentrations according to the following expression: RTKN = (TNN + NO3)/TN  100.

2.3. Sludge characterisation Total and volatile suspended solids measurements were determined in accordance with the Standard Methods [21]. Total and volatile solids samples from the mixed liquor (MLSS, MLVSS) were taken at the discharge port (‘‘upper”) and at the bottom of each reactor (‘‘lower”) as stratification is a commonly reported phenomenon in granular sludge reactors. Sludge volume index (SVI) and particle size were used as regular indicators of physical changes in sludge structure, indicating settleability and aggregate growth respectively. These two measurements, including the ratio of 5 and 30 min SVI could be used as a pseudo-indicator of granulation in the system [25]. SVI was determined at 5 and 30 min from relevant settling tests using an Imhoff cone. Particle size was determined with a laser diffraction particle size analyser (LS13-320, Beckman Coulter, Brea, CA, USA) for both lower and upper regions of the reactors. Microscope images were obtained from an inverted microscope (CKX-41, Olympus, Tokyo, Japan) and digital camera (EOS 550D, Canon, Tokyo, Japan) to regularly analyse morphological changes in sludge structure. Image analysis was performed using ImageJ software (National Institute of Health, Bethseda, Maryland; http://rsbweb.nih.gov/ij/index.html) on particles with a minimum plan area of 500 lm2, with at least 86 particles measured for each sample. Aspect ratio, circularity, roundness and solidity were determined by image analysis, and are explained in detail at http://rsb.info.nih.gov/ij/docs/menus/analyze.html#ap. Wet density was determined using a pyncometer with water (3 replicates). Extracellular polymeric substances (EPS) extraction was conducted by the formaldehyde-NaOH method following 2 min sonication [26]. Polysaccharides were determined using the Anthrone method [27] and proteins using a modified Lowry

20

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

method (DC Protein Assay, BioRad, Hercules, CA, USA) with glucose and bovine serum albumin (BSA) as the standards respectively. The shear strength of the sludge taken from R2 and R3 was measured to benchmark the physical structure against other reported granular sludge studies in both periods of operation. Shear tests were conducted following the method described in Wan et al. [28] whereby shear strength is measured by the particles ability to withstand significant changes in particle size following high speed mixing. The method also determines the aggregate’s flocculation behaviour with relation to the turbulence Kolmorogov scale. G values of 970 and 13,200 s1 were used for the two mixing speeds with further details of the test setup in the Supplementary data. 2.4. Biomass activity tests To assess the biomass activity under low and high pH, free ammonia and free nitrous acid concentrations a series of four batch tests were conducted on the sludge between days 706 and 733 (period v). Tests were conducted on washed sludge in a 250 mgN/L urine solution. Tests lasted 6 h with a biomass concentration of around 700 mg-VSS/L. Nineteen samples were taken at intervals through the test and linear regression used to determine maximum specific activity. Tests were conducted under the following conditions: (1) pH 7; (2) pH 7 spiked with NaNO2 to a concentration of approximately 800 mg-N/L; (3) pH 9; (4) pH 9 spiked with NH4Cl at an additional concentration of 450 mg-N/L. Further details are available in the Supplementary data. Maximum kinetic removal rates were determined using the ANOVA analysis package in Microsoft Excel and free ammonia and free nitrous acid concentrations were time-averaged over the test period. 2.5. Fluorescent in situ hybridization (FISH) Samples for FISH analysis were taken from R1 on day 853 from both upper and lower regions of the reactor to determine the fraction of nitrifying bacteria present in the urine-nitrifying aggregates. General fixation and hybridization procedures were in accordance with Fuchs et al. [29] and samples were cryosectioned into 14 lm sections after embedding in OCT compound (Tissue Tek OCT, Sakura-Finetek, Tokyo, Japan). The FISH oligonucleotide probes used in this study were purchased from Tech Dragon (Hong Kong, China) and are described in Table 3. Images were acquired on a confocal laser scanning microscope (LSM7 Duo, Zeiss, Oberkochen, Germany) using a 40 plan apochromat oil objective and z-stacks with 1–3 lm slices. Biovolume fractions were determined using the microbial image analysis software DAIME [34]. Congruency was maintained above 90% and at least 55 slices used for analysis of each sample. NOB were not targeted as effluent nitrate concentrations and FISH data from another fully nitrifying urine reactor in our laboratory indicated they would be below the detection limit if present. Preliminary FISH of unsectioned aggregates at

day 458 and sequencing data at day 651 also indicated NOB were absent or present at very low numbers. 3. Results 3.1. Treatment performance In Stage 1 period i–ii the main objective was to encourage biomass growth and granulation in R1 through rapidly increasing the loading rate [35,36]. During these two periods the loading was increased to 1.36 g bCOD/L d or 1.76 g-N/L d with a corresponding influent TKN concentration of 1670 mg-N/L. This was equivalent to a biomass loading rate of 760 mg-N/g-VSS d which resulted in nitrification failure. Initially in period i nitrate consisted of more than 95% of effluent TN between days 37 and 52 but was gradually lost as the NLR increased above 0.4 g-N/L d (310 mg-N/g-VSS d). During periods iii–vi the NLR was set at just over 800 mg-N/L d and gradually increased to 1150 mg-N/L d with equivalent TKN concentration of 1260 mg-N/L. Treatment was steady with TKN removal averaging 93%. Stable ammonia oxidation occurred at biomass specific NLRs below approximately 600 mg-N/g-VSS d while nitrite oxidation occurred at biomass specific NLRs below approximately 250 mg-N/g-VSS d (periods i and v). In Stage 2 (period vi) alkalinity addition to the influent resulted in a loss of nitrite oxidation despite a low biomass specific NLR. This was associated with a doubling in free ammonia post-feeding to 60–100 mg-N/L as a result of the rise in bulk liquid pH from 9 in period v to 9.4 in period vi. Such concentrations are 1–2 orders greater than the inhibition levels reported for NOB [24]. No noticeable impact was observed on ammonia oxidation. COD removal throughout R1 operation averaged 77 ± 12% (standard deviation) consistent with reports from other studies [22,37]. In Stage 3 reactors R2 (pulse feed) and R3 (gradual feed) were operated side by side. TKN removal reached 80–95% in both reactors during period i but midway through the operational period deteriorated due to sludge washout and the sludge retention times dropping below 10 d (Fig. 1c–f). In period ii, starting day 63, the settling time was increased from 7 to 30 min and MLVSS and sludge retention times responded positively in both reactors, albeit more rapidly in R2. TKN oxidation was stable from day 96 onwards in R2 at NLRs up to 550 mg-N/L d (410 mg-N/g-VSS d). Conversely, in R3 TKN removal remained poor (20–56%) and at day 81 the influent for the two reactors was separated to allow a lower NLR to R3. The TKN removal eventually stabilised in R3 from day 125 onwards with NLRs up to 350 mg-N/L d (260 mg-N/g-VSS d). Nitrite was the dominant nitrogen oxidation product in both reactors (>95%) and COD removal was similar to R1 averaging around 80%. 3.2. R1 sludge development The activated sludge used to seed R1 was flocculent (Fig. 2, day 0) with a relatively low SVI5 of 105 mL/g due to its high inert

Table 3 FISH oligonucleotide probes used in this study. Probe Name c

EUB EUB IIc EUB IIIc Nso1225 Nsv443 NEU Cte a b c

Sequence (50 -30 )

FmAa (%)

Binding positionb

Specificity

References

GCT GCC TCC CGT AGG AGT GCA GCC ACC CGT AGG TGT GCT GCC ACC CGT AGG TGT CGC CAT TGT ATT ACG TGT GA CCG TGA CCG TTT CGT TCC G CCC CTC TGC ACT CTA TTC CAT CCC CCT CTG CCG

0–50 0–50 0–50 35 30 40 40

338–355 338–355 338–355 1224–1243 444–462 653–670 659–676

Most bacteria Planctomycetales Verrucomirobiales All b-proteobacterial AOB Nitrosospira spp. Most halophilic/halotolerant Nitrosomonas spp. Competitor for NEU

[30] [31] [31] [32] [32] [33] [33]

FmA – formamide. According to Escherichia coli sequence. These probes were combined and referred to as EUBmix to target all bacteria in the system.

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

21

Fig. 1. Operational performance of R1–R3 showing nitrogen loading, nitrogen oxidation performance, and sludge accumulation. (a) R1 nitrogen treatment; (b) R1 sludge retention; (c) R2 nitrogen treatment; (d) R2 sludge retention; (e) R3 nitrogen treatment; (f) R3 sludge retention.

content (30%). During the initial 60 days the inert content dropped and remained around 8% while the SVI5 deteriorated to over 500 mL/g (Fig. 3a) as the biomass acclimated to the new environmental conditions. The SVI5 improved towards the end of period i to 200 mL/g and only slightly thereafter during period ii. For this reason biomass retention was poor and only a slight increase in MLVSS to a maximum of 3300 mg /L was achieved over period i–ii. In period iii the settling time was decreased to promote floc washout and significantly improved the SVI5 from approximately 170 mL/g in the previous period to 70 mL/g. Nevertheless, the MLVSS continued to decrease after 157 days of operation due to high effluent solids (107 ± 30 mg/L). Furthermore, the otherwise dense and well defined aggregates possessed loose tentacle-like outgrowths (Fig. 2, day 542) and the ratio of SVI5/SVI30 ranged between 1.3 and 2.4 suggesting granulation had not occurred [25]. Period iv and v focused on the influence of feeding duration which was reduced from 67 to 38 min in period iv. This had a very positive impact on biomass retention which increased from 1500 to 7000 mg/L by early period vi (as measured at the effluent port height, Fig. 1). Due to stratification under the high biomass concen-

tration MLVSS at the bottom of the reactor exceeded 12,000 mg/L. SVI5 from period iv onwards was stable between 70 and 80 mL/g with the exception of two short fluctuations. These were thought to be associated with increased pH or free ammonia following feeding caused by operational changes for periods v and vi respectively. The brief disintegration of aggregates following these changes are shown in Fig. 2, days 675 and 905. Particle size throughout the majority of the study mimicked the biomass specific loading rate reaching a maximum of 425 lm in period iii (Fig. 3a). Due to the excellent biomass retention after pulse feeding biomass specific loading rates gradually declined in subsequent periods with particle size reaching 140 lm by end of period v and 100 lm at the end of period vi. 3.3. R2 and R3 sludge development In operational period i (settling time 7 min) the mean particle size and SVI5 values in R2 and R3 were similar (Fig. 3b). SVI5 measurements for the two reactors were 71 ± 13 and 85 ± 7 mL/g respectively, averaged between days 28 to 53. During period ii

22

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

Fig. 2. Optical microscope images showing development of sludge in different reactor operational periods. Scale bar = 200 lm.

Fig. 3. Changes in sludge settling properties and mean particle diameter for (a) R1 and (b) R2 plus R3. Particle diameters for R2 and R3 are the mean of lower and upper measurements for viewing simplicity.

(settling time 30 min) mean particle size and SVI5 values in the two reactors began to diverge. Particle size remained steady in R2 (pulse feed) averaging 159 ± 22 lm while in R3 (gradual feed) the particle size increased with a final value of 220 lm. In a similar manner the SVI5 deteriorated only slightly in R2 averaging 85 ± 9 mL/g while in in R3 the SVI5 deteriorated to 132 ± 1 mL/g. Sludge retention hovered between 5 and 15 days in both reactors in period i but increased over 25 days in both reactors in period

ii. Nevertheless MLVSS did not exceed 2500 mg/L in either reactor over the study period. 3.4. Aggregate structure Shear strength was measured for R2 and R3 as an indicator of cohesiveness and compactness. At the highest mixing intensity (G = 13,200 s1) similar changes in mean diameter were observed

23

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

in period i (day 17): 120–106 lm for R2, and 129–104 lm for R3. More substantial changes were obtained in period ii (day 137) following the increase in settling time: 143–114 lm for R2 and 222–122 lm for R3. Both reactors showed deterioration in shear strength during period ii but the pulse fed reactor (R2) demonstrated higher shear strength in both periods compared with the gradually fed reactor (R3). No significant changes were observed at the first lower mixing intensity of G = 970 s1 for any sample, nor significant reaggregation in the final mixing stage at the same intensity. Comprehensive changes to particle size distribution during the shear tests are provided in the Supplementary data, Fig. A2. Wet density measurements taken in period ii also supported the higher shear strength of the pulse feed operation with densities of 1025 ± 12 mg/mL for R2 and 1002 ± 10 mg/mL for R3. Image analysis of aggregates from R1 period vi and R2 and R3 during period ii demonstrated particles with distinctly granule rather than floc like morphology (Table 4). Values were similar amongst the reactors although aggregates from R1 operated at higher influent concentrations and loading were more elongated with slightly more regular or well defined surface. Despite the similarity of R3’s (gradual feed) sludge morphology to the two pulse feed reactors, microscopic analysis revealed the proliferation of flocculent material amongst the denser aggregates during period ii (Fig. 2, R3 day 119). These particles are typically smaller than the image analysis threshold of 22 lm ferret diameter and explain the lower density and weaker structure of R3 compared to its pulse fed counterpart R2.

3.5. Microbial community FISH analysis of the aggregates indicated a generally homogenous distribution of total bacteria throughout the aggregates using probe mix EUB I–III (Fig. 4). Halotolerant and halophilic Nitrosomonas species, targeted by probe NEU, were clustered into tight colonies [33,42] and distributed in both the inner and outer regions of the aggregates. Halotolerant and halophilic Nitrosomonas comprised 23.0 ± 0.7% and 22.2 ± 1.1% in the upper and lower regions of the sludge bed respectively. Therefore stratification of the biomass concentration did not influence the ammonia oxidising bacteria distribution. The Cy3 labelled Nso1225 probe had low fluorescence intensity so that quantification of the total ammonia oxidising bacteria (AOB) population was inconclusive, although binding to NEU but not Nso1225 has been observed elsewhere [42]. However, previous FISH analysis at days 277 and 458 using a Cy5 labelled version of Nso1225 on unsectioned aggregates gave similar distributions to NEU at the aggregates surface (Supplementary data, Fig. A3). No Nitrosospira were detected using Nsv443.

3.6. Activity inhibition by free ammonia, free nitrous acid and pH Results of activity tests are shown in Table 5. AOB were found to be 73% more active at pH 9, while heterotrophs were 64% more active at pH 7. A disparity was also observed for free nitrous acid with AOBs 42% more active at the higher free nitrous acid concentration of 0.15 mg-N/L while heterotrophs were 59% less active. At the very high free ammonia concentration tested, representing partial failure in the reactor, AOB activity was inhibited by over 80% while heterotroph activity was halved. 4. Discussion 4.1. Impact of pulse feeding on nitrification The results from all three reactors indicate pulse feeding was more favourable for high rate nitrification in urine nitrifying SBRs. Reactor R2 (pulse feed) demonstrated increased volumetric and biomass specific removal rates over its gradually fed counterpart, R3, and increased volumetric nitrification rates were also observed in R1 for periods iii–v following implementation of pulse feeding. Pulse feeding creates higher bulk liquid substrate concentrations and pH which leads to higher catabolic activity of AOB, as supported by the biomass specific rates achieved in each reactor and the results of the activity tests at pH 7 and pH 9. Biomass retention was also improved by pulse feed operation. Pulse feeding enhances the formation of discrete compact aggregates by enhancing substrate penetration into aggregates and reducing internal substrate gradients [43] as well as reducing growth gradients through pH inhibition of heterotrophs and by promoting substrate storage [43]. Consequently, pulse feed operation promotes both granulation processes and high nitrification activity. This enabled a treatment rate of 1.1 g-N/L d which is similar to the authors’ initial study with another SBR [7] and considerably higher than those reported in other studies with real urine [5,18,22,37]. 4.2. Structure analysis The primary characteristics of granules are a compact and distinct morphology, ability to resist deformation under physical stress and absence of flocculation under quiescent conditions [25]. Based on gathered literature values, roundness, aspect ratio and solidity demonstrated reasonable sensitivity to granule structure with clear distinction between granules and flocs (Table 4). For these three parameters the aggregates showed very strong granular characteristics, especially roundness and solidity. In combination these three parameters provide important information on

Table 4 Geometrical properties of aggregates >500 lm2 determined by image analysis. Sample

Plan area (lm2)

Circularity

Aspect ratio

Roundness

Solidity

R1(day 743) R2 (day 164) R3 (day 164) Granulea Granuleb Granulec Flocb Flocc Flocd Floce

5457 ± 2934 7343 ± 4598 7760 ± 4807

0.41 ± 0.12 0.41 ± 0.14 0.42 ± 0.14 0.37–0.54

1.45 ± 0.26 1.44 ± 0.26 1.42 ± 0.27 1.16–1.45 1.34–1.42

0.71 ± 0.12 0.72 ± 0.12 0.73 ± 0.12

0.85 ± 0.12 0.83 ± 0.18 0.82 ± 0.21

1.76

0.25

1.8 ± 0.5

0.43–0.56 0.46 ± 0.06

Mean ± standard deviation. a Beun et al. [12];

0.57–0.61 0.78–0.86

b

Su and Yu [38]; c Amaral [39];

d

0.57–0.77 0.7–0.76 0.73 ± 0.05

van den Broeck et al. (2009) in Van den Kerkhof et al. [40]; e Mesquita et al. [41].

24

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

Fig. 4. FISH images of two sectioned aggregates from R1 showing spatial distribution of total bacteria (green – EUBmix) and halotolerant Nitrosomonas (red – Nso1225). Dual staining shows as orange in the combined image. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 5 AOB and heterotroph activity under varying pH, free ammonia and free nitrous acid. Test No. 1 2 3 4

pH

Free ammonia mg-N/L

Free nitrous acid mg-N/L

AOB activity mg-N/g-VSS h

Heterotroph activity mg-COD/g-VSS h

7 7 9 9

1.1 1.0 52 243

0.0064 0.15 0.0001 0.0000

13.9 ± 0.4 19.8 ± 1.8 24.1 ± 0.6 4.6 ± 0.1

179 ± 31 74 ± 13 109 ± 10 57 ± 2

the elongation and shape irregularities/surface smoothness. Compactness of the pulse fed sludge from R2 in this study was also demonstrated by wet density measurements, with similar density to other granule studies despite a low inert content [35,44]. Aggregates, particularly those from conditions combining pulse feed and short settling, also demonstrated high resistance to deformation despite the maximum shear rate tested being over 40 times higher than that caused by aeration in the reactors (300 s1, see Supplementary data, Table A1). Additionally little to no reaggregation was observed under reduced shear such that the aggregates in this study fulfilled all the previously mentioned characteristics of granular sludge. Nevertheless the ratio of SVI5/SVI30 which is typically used to monitor the degree of granulation did not approach 1 [25]. The primary reason for failing to satisfy this general indicator was the small size of the particles in this study which was strongly correlated with the loading (see Fig. 3a and Supplementary data, Fig. A6). As the nominal size for granules is also loosely set at 200 lm [25] one may consider the sludge aggregates developed in the study granule-like or transitional granular sludge.

4.3. Influence of feeding operation and settling pressure on granulation Both pulse feeding and short settling time have been demonstrated as key factors in promoting aerobic granulation [12,19,25,45]. In this study both factors played an important role in developing granular like aggregates as demonstrated by SVI5, morphology, density and shear strength measurements. Neverthe-

less, the influence of pulse feeding was more pronounced based on the comparison between R2 and R3 (Stage 3). When settling pressure was relaxed in phase ii shear strength and SVI5 in the pulse fed reactor (R2) only deteriorated slightly relative to the gradual feed reactor (R3). This indicates pulse feeding could maintain a relatively granular structure without the need for settling selection pressure. In contrast a reduction in the settling time from 13 to 7 min in R1 Stage 1 (period iii) was accompanied by a significant decrease in SVI5 but failed to successfully retain sludge in the reactor. Only after a pulse feed was also implemented did the biomass concentrations increase (period iv–v). Furthermore, despite a reduction in particle diameter to less than a third of the previous size after pulse feeding was implemented the SVI5 values before and after were similar, indicating pulse feeding led to significantly more compact aggregates. As urine is comprised of predominantly organic acids which are very readily degraded pulse feeding could play an instrumental role in reducing internal substrate gradients to promote compact aggregates. Unlike wastewaters with lower COD:N ratios the daily hydraulic loading of urine SBRs is low and therefore the role of settling or hydraulic washout pressure on granulation is reduced. Settling time should also be considered a secondary selection pressure due to its negative impact on operational stability in a urine nitrifying system. Despite the maximum critical settling velocity implemented (2.0 m/h) being at the lower limit reported to promote granulation [20,36] the sludge retention time in R1 remained just above an approximate 10 d threshold to ensure reactor stability (Fig. 1a and b, period iii and Supplementary data, Tables A2 and A3). Similarly in R2 and R3 (Stage 3) unstable performance from a sludge retention time below the approximate 10 day threshold was only recovered after the critical settling velocity was reduced from 1.6 to 0.4 m/h. In granulation of low to moderate strength wastewaters nitrification performance is often sacrificed during start-up to achieve a sufficient hydraulic selection pressure and organic loading rate, with nitrifying communities developing once granulation has stabilized [46]. However, in a urine-fed system nitrification failure results in free ammonia concentrations sufficient to prevent nitrification returning and is therefore not a viable strategy (see the Supplementary data, Table A5).

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

4.4. Impact of free ammonia and pH Free ammonia and pH are closely linked and both may play a role in disrupting granulation. Free ammonia has previously been reported to inhibit granulation and EPS production with a threshold concentration of 20 mg-N/L being identified in two separate studies [47,48]. In contrast, pH strongly influences microbial growth and at elevated levels over 9 has been reported to solubilise granular EPS [15], which would result in reduced cohesion and granule growth. In R1 the particle size showed strong correlation with the loading rate which is a known factor to influence particle diameter [35]. Particle size also showed negative correlation with post-feeding pH and free ammonia concentrations, including period vi when particle size did not correlate with biomass loading (Fig. 3a, Fig. 5). Moreover, particle size was clearly not influenced by free nitrous acid concentrations (Supplementary data, Fig. A4). Therefore, elevated pH or free ammonia was at least partly responsible for the observed decreases in aggregate size and a contributing factor limiting full granule development. The mechanisms could be related to impacts on heterotrophic growth and EPS production. Maximum heterotrophic activity following feeding in a urine system coincides with maximum pH and free ammonia concentrations, both of which were shown to be detrimental to heterotrophs in the activity batch tests. The impacts of pH and/or free ammonia were also supported in R2 and R3. Both reactors showed a reduction in particle size during days 40–60 when nitrification deteriorated and following stable performance at the end of period ii the higher loaded and pulse fed reactor R2 had smaller particle diameter than R3. 4.5. COD:N ratio and organic loading rate Urine contains an almost equal mass concentration of COD and TKN (both soluble) with the majority of organics being rapidly biodegradable. Consequently, the microbial community contained 23% halotolerant Nitrosomonas sp. as demonstrated by FISH. This number was further supported by comparison of estimated biomass yields for heterotrophs and nitrifiers from influent organics and nitrogen respectively. Heterotroph true yield on urine organics was estimated at 0.62 g COD/g COD based on thermodynamic calculations. Using reported yields for Nitrosomonas and ignoring decay for both heterotrophs and nitrifiers a 28% nitrifier fraction was estimated (Supplementary data). The majority of remaining organisms are likely to be heterotrophs due to aerobic conditions and high organic concentrations. This leads to a system that is still dependent on heterotrophic organisms for sufficient biomass and EPS production to achieve granulation. The maximum organic loading rate achieved in the study, corresponding to the highest stable NLR, was 0.96 g biodegradable-COD/L d. This is in comparison with other aerobic granulation studies that indicate a minimum organic loading rate in the range of 1.5–2.0 g COD/L d if phosphate removR1 Day 421 (iii) R1 Day 616 (iv) R1 Day 684 (v) R2 Day 162 (ii) R3 Day 163 (ii)

9

pH

8.5 8 7.5 7

0

60

120

180

Time into cycle (min)

240

300

360

Fig. 5. Changes in pH during reactor cycles showing influence of feeding. Note R1 feeding is at t = 0 and 150 min, for R2 and R3 at t = 0 min only. Feeding durations are provided in Table 1. R1 (vi) not available but pH post feeding was 9.4.

25

ing organisms and anaerobic upflow feeding through the sludge bed are not utilised [29,41]. Reduced polysaccharide and protein production in all three reactors (18.9–28.3 mg-glucose/g-VSS and 75.7–99.5 mg-BSA/g-VSS) compared with the seed sludge (32.5 mg-glucose/g-VSS and 133.6 mg-BSA/g-VSS) also supported the importance of organic loading rate limitation on the system (Supplementary data, Fig. A5). While nitrification is typically sacrificed during granulation [47] this is not feasible with such a high TKN–high pH wastewater as previously discussed. One possible way to overcome irrecoverable nitrification failure would be to initially provide two-way pH control, limiting the upper pH in the reactor to somewhere between 8 and 8.5 during the granulation process. With the urine dilution used in this study and a pH control point of 8.5 free ammonia concentrations would be reduced by 70% under nitrification failure and allow nitrification to re-establish if the SRT is greater than 4 days (Supplementary information, Table A6). In this way high loading rates (ie short cycle time) and short settling times can be applied to promote granulation initially. Following granulation and biomass accumulation significantly beyond the required SRT the upper pH control could be removed and a longer cycle time used to allow complete nitrification of influent TKN. 4.6. Comparison to floccular based urine nitrifying SBRs In our previous study [7] a urine nitrifying reactor was successfully operated with alkaline dosing for 85 days. The reactor achieved complete nitrification to nitrate at influent nitrogen concentration of approximately 1800 mg-N/L and a NLR of 1.1 mg-N/L d which was very similar to this study with the exception of nitrate rather than nitrite as the product. The lack of improved performance using the granular system can be attributed to the small aggregate diameter achieved. This limited the biomass retention and MLVSS to a maximum of approximately 9000 mg/L considering stratification, which was similar to that in the flocculent system using a longer settling phase. Both studies achieved slightly higher loading rates than studies without alkaline dosing with a half nitrite-half ammonia effluent [22,37]. Differences in the effluent product can be attributed to three key factors. Firstly feeding in the previous study [7] consisted of three 70-min feeding periods distributed through the 12 h cycle resulting in lower pH and free ammonia concentrations in the reactor. Secondly the seed sludge was an enriched nitrifying culture and was loaded gradually to acclimate biomass rather than an unacclimated culture loaded quickly in this study in an effort to force granulation. Finally, the previous study had an SRT of 40–50 days as opposed to an SRT of as little as 10 d in this study associated with a short settling time to promote floc washout. 4.7. Nitrite vs nitrate production Nitrate and nitrite both have their advantages and disadvantages as a final product for sewer discharge. Nitrate provides increased capacity for hydrogen sulphide, methane and organics oxidation in the sewer and reduces the impact of nitrous oxide emissions during heterotrophic denitrification. Nevertheless, nitrite has been demonstrated as more effective in the control of sulphate reducing bacteria and methanogens in sewer networks [49]. In addition, readily biodegradable organics may be limited by the rate of hydrolysis in the sewer and therefore shortcut denitrification using nitrite enhances the potential to fully remove nitrogen before the treatment plant. One concern from nitrite is the production of nitrous oxide through both heterotrophic and autotrophic pathways. As emissions from AOB are generally significantly greater than from heterotrophs in wastewater systems the greatest concern for emissions is

26

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

from the nitrification reactor [50]. However, recent research by Law et al. [51] revealed that, while nitrous oxide emissions are increased at low nitrite concentrations (especially < 50 mg-N/L), at high nitrite concentrations (500–1000 mg-N/L) comparable to those in this study nitrous oxide emissions decrease and are comparable to fully nitrifying systems. Considering also that nitrous oxide emissions are increased under transient nitrite conditions [52] emissions from a nitrate producing SBR reactor could potentially be comparable. Due to the complex pathways and impacts of nitrite, DO and pH on nitrous oxide emissions further study of the greenhouse gas emissions from this system are required to make a sound assessment of the systems emission impacts. The presence of nitrite also results in nitrous acid production at lower pH which can cause AOB and NOB inhibition. At the highest effluent nitrite concentration recorded of 1266 mg-N/L the corresponding free nitrous acid concentration at the pH 7 set-point was 0.19 mg-N/L. This is well within reported inhibitory ranges for both AOB and NOB [24]. Although the loss of NOB from the system was associated with increased free ammonia and pH during rapid loading in period i, the presence of high nitrous acid concentrations thereafter may have contributed to the delayed return of NOB over phases ii–v. Nevertheless, some studies have shown that certain NOB are not inhibited at free nitrous acid concentrations well above those in this study [53,54]. Free nitrous acid was not inhibitive on AOB activity at a tested concentration of 0.15 mg-N/L (Table 5) although the impacts on AOB growth may be more significant (Supplementary data, Figs. A5 and A6). 5. Conclusion Urine is a unique substrate containing high TAN and readily biodegradable organic concentrations, high pH and low COD:N ratio. These factors make it a challenging and scientifically interesting substrate for granulation due to competing requirements to force granulation of the predominantly heterotrophic biomass (organic loading rate and settling) while maintaining nitrification to reduce pH and free ammonia for system stability. In this study it was found that pulse feeding was instrumental in the development of compact granule-like aggregates with low SVI5 (70 mL/g), compact structure and high cohesion strength. Nevertheless, pulse feeding also resulted in higher pH and free ammonia concentrations following feeding which impact heterotroph activity and contributed to reductions in particle diameter. Application of short settling times was also beneficial to sludge structure, but of less significance with the associated risk of sludge washout and system failure. It was found that application of critical settling velocities in the range of 2 m/h was near the limit of system stability, just below commonly reported values to initiate granulation in strongly nitrifying systems. Pulse feeding was also beneficial to treatment performance, increasing both volumetric and biomass specific TKN removal rates and allowing high rate conversion of 1.1 kg-N/m3 d and 600 mg-N/g-VSS d respectively to realise a compact decentralised nitrification reactor. Acknowledgement This project was partly supported by the Hong Kong Research Grants Council (611607). The authors would also like to acknowledge Prof. Mark van Loosdrecht for his constructive comments during the early stages of manuscript preparation. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.cej.2015.12.071.

References [1] R. Otterpohl, U. Braun, M. Oldenburg, Innovative technologies for decentralised water-, wastewater and biowaste management in urban and peri-urban areas, Water Sci. Technol. 48 (11–12) (2003) 23–32. [2] J. Hanæus, D. Hellström, E. Johansson, A study of a urine separation system in an ecological village in Northern Sweden, Water Sci. Technol. 35 (9) (1997) 153–160. [3] M. Maurer, W. Pronk, T.A. Larsen, Treatment processes for source-separated urine, Water Res. 40 (2006) 3151–3166. [4] T.A. Larsen, M. Maurer, R.I.L. Eggen, W. Pronk, J. Lienert, Decision support in urban water management based on generic scenarios: the example of NoMix technology, J. Environ. Manage. 91 (2010) 2676–2687. [5] K.M. Udert, M. Wächter, Complete nutrient recovery from source-separated urine by nitrification and distillation, Water Res. 46 (2012) 453–464. [6] J. Dai, W.-T. Tang, Y.-S. Zheng, H.R. Mackey, H.K. Chui, M.C.M. van Loosdrecht, G.-H. Chen, An exploratory study on seawater-catalysed urine phosphorus recovery (SUPR), Water Res. 66 (2014) 75–84. [7] F. Jiang, Y. Chen, H.R. Mackey, G.H. Chen, M.C.M. van Loosdrecht, Urine nitrification and sewer discharge to realize in-sewer denitrification to simplify sewage treatment in Hong Kong, Water Sci. Technol. 64 (3) (2011) 618–626. [8] J. Wilsenach, M. van Loosdrecht, Impact of separate urine collection on wastewater treatment systems, Water Sci. Technol. 48 (1) (2003) 103–110. [9] E. Morgenroth, Biofilm reactors, in: M. Henze, M.C.M. van Loosdrecht, G.A. Ekama, D. Brdjanovic (Eds.), Biological Wastewater Treatment: Principles, Modelling and Design, IWA Publishing, London, UK, 2008, pp. 493–511. [10] M.K.H. Winkler, J.P. Bassin, R. Kleerebezem, D.Y. Sorokin, M.C.M. van Loosdrecht, Unravelling the reasons for disproportion in the ratio of AOB and NOB in aerobic granular sludge, Appl. Microbiol. Biotechnol. 94 (2012) 1657–1666. [11] S. Tsuneda, M. Ogiwara, Y. Ejiri, A. Hirata, High-rate nitrification using aerobic granular sludge, Water Sci. Technol. 53 (3) (2006) 147–154. [12] J.J. Beun, A. Hendriks, M.C.M. van Loosdrecht, E. Morgenroth, P.A. Wilderer, J.J. Heijnen, Aerobic granulation in a sequencing batch reactor, Water Res. 33 (1999) 2283–2290. [13] N. Kishida, S. Tsuneda, J.H. Kim, R. Sudo, Simultaneous nitrogen and phosphorus removal from high-strength industrial wastewater using aerobic granular sludge, J. Environ. Eng. ASCE 135 (3) (2009) 153–158. [14] J.A. Torà, E. Moliné, J. Carrera, J. Pérez, Efficient and automated start-up of a pilot reactor for nitritation of reject water: from batch granulation to high rate continuous operation, Chem. Eng. J. 226 (2013) 319–325. [15] T. Seviour, M. Pijuan, T. Nicholson, J. Keller, Z. Yuan, Gel-forming exopolysaccharides explain basic differences between structures of aerobic granular sludge granules and floccular sludges, Water Res. 43 (2009) 4469– 4478. [16] S.F. Yang, X.Y. Li, H.Q. Yu, Formation and characterisation of fungal and bacterial granules under different feeding alkalinity and pH conditions, Proc. Biochem. 43 (2008) 8–14. [17] C. Wan, X. Yang, D.-J. Lee, Q. Zhang, J. Li, X. Liu, Formation of filamentous aerobic granules: role of pH and mechanism, Appl. Microbiol. Biotechnol. 98 (2014) 8389–8397. [18] F.-Y. Sun, Y.-J. Yang, W.-Y. Dong, J. Li, Granulation of nitrifying bacteria in a sequencing batch reactor for biological stabilisation of source-separated urine, Appl. Biochem. Biotechnol. 166 (2012) 2114–2126. [19] L. Qin, J.-H. Tay, Y. Liu, Selection pressure is a driving force of aerobic granulation in sequencing batch reactors, Proc. Biochem. 39 (2004) 579–584. [20] L. Qin, Y. Liu, J.H. Tay, Effect of settling time on aerobic granulation in a sequencing batch reactor, Biochem. Eng. J. 21 (2004) 47–52. [21] APHA, AWWA, WEF, Standard Methods for the Examination of Water and Wastewater, 21st ed., American Public Health Association, Washington, DC, 2005. [22] K.M. Udert, C. Fux, M. Münster, T.A. Larsen, H. Siegrist, W. Gujer, Nitrification and autotrophic denitrification of source-separated urine, Water Sci. Technol. 48 (1) (2003) 119–130. [23] R.G. Bates, G.D. Pinching, Acidic dissociation constant of ammonia ion at 0° to 50 °C, and the base strength of ammonia, Research Paper RP1982, J. Res. Natl. Bur. Stand. 42 (1949) 419–430. [24] A.C. Anthonisen, R.C. Loehr, T.B.S. Prakasam, E.G. Srinath, Inhibition of nitrification by ammonia and nitrous acid, J. Water Pollut. Con. Fed. 48 (1976) 835–852. [25] M.K. de Kreuk, N. Kishida, M.C.M. van Loosdrecht, Aerobic granular sludge – state of the art, Water Sci. Technol. 55 (8–9) (2007) 75–81. [26] H. Liu, H.H.P. Fang, Extraction of extracellular polymeric substances (EPS) of sludges, J. Biotechnol. 95 (2002) 249–256. [27] A.F. Gaudy Jnr, Colorimetric determination of protein and carbohydrate, Ind. Water Wastes 7 (1962) 17–22. [28] J. Wan, I. Mozo, A. Filali, A. Liné, Y. Bessière, M. Spérandio, Evolution of bioaggregate strength during aerobic granular sludge formation, Biochem. Eng. J. 58–59 (2011) 69–78. [29] B.M. Fuchs, J. Pernthaler, R. Amann, Single cell identification by fluorescence in situ hybridization, in: C.A. Reddy, T.J. Beveridge, J.A. Breznak, G. Marzluf, T. M. Schmidt, L.R. Snyder (Eds.), Methods for General and Molecular Microbiology, 3rd ed., ASM Press, Washington DC, 2007, pp. 886–896. [30] R.I. Amann, B.J. Binder, R.J. Olson, S.W. Chisholm, R. Devereux, D.A. Stahl, Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry

H.R. Mackey et al. / Chemical Engineering Journal 289 (2016) 17–27

[31]

[32]

[33]

[34] [35]

[36] [37]

[38]

[39]

[40]

[41]

[42]

for analyzing mixed microbial populations, Appl. Environ. Microbiol. 56 (1990) 1919–1925. H. Daims, A. Brühl, R. Amann, K.-H. Schleifer, M. Wagner, The domain-specific probe EUB338 is insufficient for the detection of all bacteria: development and evaluation of a more comprehensive probe set, Syst. Appl. Microbiol. 22 (1999) 434–444. B.K. Mobarry, M. Wagner, V. Urbain, B.E. Rittmann, D.A. Stahl, Phylogenetic probes for analyzing abundance and spatial organization of nitrifying bacteria, Appl. Environ. Microbiol. 62 (1996) 2156–2162. Erratum published 63 (1997) 815. M. Wagner, G. Rath, R. Amman, H.-P. Koops, K.-H. Schleifer, In situ identification of ammonia-oxidizing bacteria, Syst. Appl. Microbiol. 18 (1995) 251–264. H. Daims, S. Lücker, M. Wagner, Daime, a novel image analysis program for microbial ecology and biofilm research, Environ. Microbiol. 8 (2006) 200–213. J.-H. Tay, S. Pan, Y. He, S.T.L. Tay, Effect of organic loading rate on aerobic granulation. II. Characteristics of aerobic granules, J. Environ. Eng. ASCE 130 (2004) 1102–1109. M.K. de Kreuk, M.C.M. van Loosdrecht, Formation of aerobic granules with domestic sewage, J. Environ. Eng. 132 (2006) 694–697. F.-Y. Sun, W.Y. Dong, M.F. Shao, J. Li, L.Y. Peng, Stabilization of sourceseparated urine by biological nitrification process: treatment performance and nitrite accumulation, Water Sci. Technol. 66 (7) (2012) 1491–1497. K.-Z. Su, H.-Q. Yu, Formation and characterization of aerobic granules in a sequencing batch reactor treating soybean-processing wastewater, Environ. Sci. Technol. 39 (2005) 2818–2827. A.L.P.D. Amaral, Image analysis in biotechnological processes: applications to wastewater treatment (Ph.D. dissertation), Department of Biological Engineering, University of Minho, Braga, Portugal, 2003. P. Van den Kerkhof, G. Gins, R. Van den Broeck, J.F.M. Van Impe, Multivariate assessment of activated sludge stability in lab-scale experiments, Proc. Biochem. 48 (11) (2013) 1789–1793. D.P. Mesquita, A.L. Amaral, E.C. Ferreira, Characterization of activated sludge abnormalities by image analysis and chemometric techniques, Anal. Chim. Acta 705 (2011) 235–242. G. Carvalho, R.L. Meyer, Z. Yuan, J. Keller, Differential distribution of ammoniaand nitrite-oxidising bacteria in flocs and granules from a

[43]

[44]

[45] [46]

[47] [48]

[49]

[50] [51]

[52]

[53]

[54]

27

nitrifying/denitrifying sequencing batch reactor, Enzyme Microb. Technol. 39 (2006) 1392–1398. A.M.P. Martins, J.J. Heijnen, M.C.M. van Loosdrecht, Effect of feeding pattern and storage on the sludge settleability under aerobic conditions, Water Res. 37 (2003) 2555–2570. Y. Liu, J.-H. Tay, The essential role of hydrodynamic shear force in the formation of biofilm and granular sludge, Water Res. 36 (2002) 1653– 1665. B.S. McSwain, R.L. Irvine, P.A. Wilderer, The effect of intermittent feeding on aerobic granule structure, Water Sci. Technol. 49 (11) (2004) 11–12. D.G. Weissbrodt, S. Lochmatter, S. Ebrahimi, P. Rossi, J. Maillard, C. Holliger, Bacterial selection during the formation of early-stage aerobic granules in wastewater treatment systems operated under wash-out dynamics, Front. Microbiol. 3 (2012) 332, http://dx.doi.org/10.3389/fmicb.2012.00332. S.-F. Yang, J.-H. Tay, Y. Liu, Inhibition of free ammonia to the formation of aerobic granules, Biochem. Eng. J. 17 (2004) 41–48. X. Zheng, P. Sun, J. Lou, J. Cai, Y. Song, S. Yu, X. Lu, Inhibition of free ammonia to the granule-based enhanced biological phosphorus removal system and the recoverability, Bioresource Technol. 148 (2013) 343–351. G. Jiang, O. Gutierrez, K.R. Sharma, Z. Yuan, Effects of nitrite concentration and exposure time on sulfide and methane production in sewer systems, Water Res. 44 (2010) 4241–4251. Y. Law, L. Ye, Y. Pan, Z. Yuan, Nitrous oxide emissions from wastewater treatment processes, Philos. Trans. R. Soc. B 367 (2012) 1265–1277. Y. Law, P.A. Lant, Z. Yuan, The confounding effect of nitrite on N2O production by an enriched ammonia-oxidising culture, Environ. Sci. Technol. 47 (2013) 7186–7194. M.J. Kampschreur, N.C.G. Tan, R. Kleerebezem, C. Picioreanu, M.M. Jetten, M. C.M. van Loosdrecht, Effect of dynamic process conditions on nitrogen oxides emission from a nitrifying culture, Environ. Sci. Technol. 42 (2008) 429–435. J.H. Hunik, H.J.G. Meijer, J. Tramper, Kinetics of Nitrobacter agilis at extreme substrate, product and salt concentrations, Appl. Microbiol. Biotechnol. 40 (1993) 442–448. U. Sudarno, J. Winter, C. Gallert, Effect of varying salinity, temperature, ammonia and nitrous acid concentrations on nitrification of saline wastewater in fixed-bed reactors, Bioresource Technol. 102 (2011) 5665–5673.