129
Improved nitrogen removal in a membrane-aerated biofilm reactor with advanced biofilm control
Nemeth, A.*, Heffernan, B.*, Massons, G.**, Koch, John**, Puig, D.**, Zardoya, D.** and Manzano, J.*
*OxyMem Ltd, Moydrum Business Park, Athlone, Co. Westemath, Ireland, N37 DR79
** DuPont Water Solutions, Autovia Tarragona-Salou s/n, 43006 Tarragona (Spain) Keywords: MABR; nitrogen removal; biofilm management
Summary of key findings
A demonstration scale membrane-aerated biofilm reactor (MABR) system containing two OxyFilm membrane modules was operated at Vila-seca WTTP in Spain for 22 months. Initially the system was operated with opeators interpreting the data and setting the scour frequency and duration. The
variation of influent quality and the time delay between samples being taken and the opperators receiving data resulted in fluctuactions in the biofilm thickness – evidenced by the biofilm weight and the pressure drop over the module. The development and implementation of a scour decision processor (SDP) allowed for real-time in-situ biofilm measurements and an automated control of biofilm
thickness. After the implementation of the SDP higher hydraulic and pollutant loads could be applied and the removal of both ammonium and total nitrogen increased substantially.
Background and relevance
The development and application of MABRs has gained traction in recent years, with multiple commercial projects deployed around the globe. While a significant part of this interest comes from the potential to integrate a counter-diffusional biofilm with activated sludge, the stand-alone application of the MABR has prospects as a highly energy efficient technology for COD and N removal. However, the management of the biofilm has been identified as a challenge by multiple authors (Syron & Casey, 2007) (Nerenberg, 2016).
OxyMem has developed a patented non-invasive biofilm thickness measurement technology based on the diffusion of an inert gas from the membrane lumen into the liquid (EU Patent No. EP2192379A1, 2010). By correlating the transfer of the inert gas with the weight of the accumulated biomass, a biofilm thickness index (BTI) was formulated. The BTI represents the relative thickness of the biofilm and provides the ability to monitor changes in the biofilm thickness over time and most importantly to control it. The thickness of the biofilm is controlled by coarse bubble scouring, which removes solids from the membrane. The amount of solids removed depends on both the duration and the frequency of the scour events. Typically, operators determine the scour duration and frequency however an operator is always operating with information that is 3 – 4 days old and thus the it can be difficult to control the system, especially if the influent load is highly variable. The implementation of the BTI to determine the biofilm thickness and the SDP to control the scour takes this difficult task away from the operator.
The SDP has been deployed at a demonstration MABR system (with a total membrane surface area of 4400 m2) treating primary effluent of a municipal wastewater treatment plant. Influent and effluent wastewater was sampled on weekdays with automatic samplers, the samples were analysed for tCOD, sCOD, BOD5, TOC, TN, NH4-N, NO2-N, NO3-N, TP, TSS and VSS. The flow rate, pressure,
temperature and oxygen content of the process gas were measured as well, allowing the calculation of oxygen flux across the membranes.
130
Figure 1 When the BTI (blue circles) goes out of the specified range (light blue box), the scour intensity (yellow dots) is adjusted to remove less or more biofilm.
Results
The influent flow rate was changed between 35 and 110 m3/d in the 600 days of operation, the tCOD load varied between 1 and 20 g m-2d-1 due to the fluctuations of inlet concentration. The ammonium-N load was more even, varying between 0.25 and 1.20 gN m-2d-1. Removal rates of tCOD and NH4-N averaged at 5.19 and 0.55 g m-2d-1, respectively, with revolval efficiencies of 73% and 83%.
Following the implementation of the SDP the influent flow rate could be increased to 180 m3d-1 and the tCOD loading could be maintained in the range of 13-16 g m-2d-1. The average tCOD and NH4-N removal rates increased to 7.93 and 1.18 g m-2d-1, respectively. The TN removal was boosted from 0.20 to 0.70 g m-2d-1, resulting in an increase of removal efficiency from 29% to 52%. The average effluent TN concentration decreased from 24.7 to 16.2 mg/L.
Discussion
During the first operation period with manual biofilm management information was gathered on the BTI and various scour and sludge extraction strategies were used. The BTI was observed to vary between 0.92 and 1.75, and while likely impacted by the incoming organic load, the values could not be correlated directly to the daily tCOD loading rate. An average BTI value of 1.26 with a standard deviation of 0.23 resulted from the manual adjustment of scour intensity, which occurred at most at a frequency of once per week.
With the SDP the BTI was evaluated on a daily basis and the scour intensity was adjusted if the value was outside the specified range (Figure 1). This more frequent, automatic control reduced the standard deviation of BTI to 0.15 with an average of 1.24. The increased hydraulic load could be managed safely despite the high tCOD loads. The most remarkable difference was detected in the ammonium and total nitrogen removal. Despite the high COD and NH4-N load the ammonium removal e improved, and compared to earlier a significantly higher removal rate could be achieved at the same effluent concentrations (Figure 2). This indicates that the nitrifier population could be enhanced and mass transfer resistance was well-managed, not interfering with the removal.
131
Figure 2 With automatic biofilm control (SDP) enabled, the nitrifier population could be increased and higher removal rates were achieved at the same effluent ammonium concentrations
Total nitrogen removal also improved, probably because of the systems ability to handle larger COD loads and thus create an anoxic layer in the biofilm. It should be noted that for both periods, with and without, using the SDP TN removal efficiency and effluent TN concentration showed a dependency on the influent COD:N ratio as can be expected. Even so he average TN removal efficiency increased and the standard deviation decreased when automated biofilm management was implemented,the 90th percentile TN removal efficiency improved from 29% to 77%.
As mentioned the increased COD and nitrogen load received by the plant could contribute to the formation of more stable anoxic layers on the outside of the biofilm, where denitrification can take place. The results suggest that highly loaded biofilms can be more successful in achieving
simultaneous nitrification and denitrification, without using nitrate recirculation. A key element to realise this is to preserve an active nitrifier population under high COD loads.
This study highlights the contribution of the automated biofilm measurement and management system as an effective tool to unlock the potential of the stand-alone MABR in the treatment of municipal wastewater.
References
Heffernan, B., Casey, E., & Syron, E. (2010). EU Patent No. EP2192379A1.
Nerenberg, R. (2016). The membrane-biofilm reactor (MBfR) as a counter-diffusional biofilm process. Current Opinion in Biotechnology, 38:131-1366.
Syron, E., & Casey, E. (2007). Membrane-Aerated Biofilms for High Rate Biotreatment: Performance Appraisal,
Engineering Principles, Scale-up, and Development Requirements. Environmental Science & Technology, 1833- 1844.
132
3
3 3
NOB Suppression in a Single ZeeNAMMOXTM Biofilm with the Coexistence of AOB and Anammox
Long, Z.*, Houweling, D.*, Ireland, J.*, Peeters, J.*, Coutts, D.**, and Reeve, M.*
* SUEZ Water Technologies and Solutions. 3239 Dundas Street West. Oakville, ON, Canada L6M 4B2
** SUEZ Water Technologies and Solutions. Via Benigno Crespi No. 57, Milano MI, Italy 20159
Keywords: membrane aerated biofilm reactor; Partial nitritation and Anammox; NOB suppression Summary of key findings
ZeeNAMMOXTM takes advantages of the counter-diffussion benefits of membrane aerated biofilm reactor (MABR) technology, forming a single biofilm that carries out partial nitritation and Anammox (PN/A) process with the coexistence of ammonium oxidizing bacteria (AOB) and Anammox. Data from pilots under a variety of different operational conditions for the treatment of sidestream
wastewater demonstrated high ammonium oxidation rates (AOR) and total inorganic nitrogen removal rates (TINRR) with low and controlled NO --N generation rates (NO --N GR). Further data analysis found that the ratio of NO --N GR to the AOR had a decreasing trend with the increase in the TINRR, suggesting that nitrite oxidizing bacteria (NOB) might be gradually suppressed by the accumulation of Anammox in the biofilm. Anammox activity batch tests confirmed that the mature biofilms at steady state had more capacity for TIN removal than needed for the continuous operations. This might be the strategy for Anammox to outcompete NOB by minimizing the nitrite availability and occupying the growing space at the same time.
Background and relevance
Deammonification via PN/A process removes total nitrogen (TN) at minimal requirements for aeration, sludge disposal and carbon source. However, difficulties exist in limiting NOB while proliferating AOB, and in growing and retaining the slow growing Anammox bacteria. Full-scale experience of the existing PN/A technologies has indicated that process disruptions frequently
occurred and are mainly associated with the failure of control over critical process parameters, such as DO and/or pH, biomass loss due to high total suspended solids (TSS) in the feed and/or bad
performance of solids separation units. (Lackner et al., 2014).
ZeeNAMMOXTM is a novel application of MABR technology for side-stream nutrients removal.
Bubbleless aeration and complete control over O2 supply in MABR technology enables ZeeNAMMOXTM to achieve stable, high-rate PN/A at low energy consumption (Coutts et al.,
2020a,b). In this paper, experimental results from four ZeeNAMMOXTM pilots are discussed, with the focus on the understanding of NOB suppression in the biofilm.
Results and discussion
Operations: Four ZeeNAMMOXTM pilots have been operated from more than 6 months to about 2 years. All pilots were seeded with nitrifying sludge during startups and then went through stages of high-rate partial nitritation followed by deterioration of partial nitritation. The high-rate partial nitritation after startup is expected due to the higher growth rate of AOB than NOB at elevated temperatures. The deterioration of partial nitritation indicated that NOB gained ground in the biofilm over time. The extend and rate of the deterioration was depending on the operational conditions, highlighting the importance of process optimization for the suppression of NOB so that the complete nitrification can be controlled within acceptable levels. Different control strategies, including
133
Anammox seeding, oxygen limitation, and intermittent process air supply, were examined and
implemented individually or in combination for this purpose. Table 1 summarizes the main operational conditions and pilot performance during steady state after the implementation of the control strategies.
Performance: As can be seen from Table 1, the volumetric AORs, which were determined by the product of the specific AOR from the pilots and the membrane packing density of ZeeNAMMOXTM product, ranged from 0.63 to1.1kgN/m3/d, which, by comparing with the volumetric loading rates of full-scale applications (Lackner et al., 2014), are equivalent to or higher than those of the SBR-type technologies such as DEMON and the biofilm-type technologies such as ANITAMoxTM. While this is not the apples to apples comparison, the point can be made that ZeeNAMMOXTM system will be very compact, resulting in saving for footprint. Moreover, the majority of the ammonium oxidized was completely removed from the wastewater as indicated by the high ratios of TINRR to AOR (from 0.76 to 0.98). The rest of the ammonium oxidized was mainly converted into NO3--N, as can be seen from the specific and volumetric NO3--N generate rates (NO3--N GR) in the table. The TIN removal was mainly carried out by Anammox bacteria, which was confirmed by biofilm qPCR tests and Anammox activity batch tests using wastewater containing NH4+-N and NO2--N. In some cases, denitrification, denitritation and even partial denitrification might have also contributed to the TIN removal. The NO3
--N might be generated from PN/A process and/or nitrification process, and could be removed by processes such as denitrification and/or partial denitrification. The theoretical values of the two ratios for the PN/A process are around 0.89 and 0.11, respectively (Strous et al., 1998). Affecting factors, such as denitrification, sources of nitrite from the feed or generated by partial denitrification, might have resulted in a higher ratio of TINRR to AOR and/or a lower ratio of NO3--N GR to AOR. By contrast, a lower ratio of TINRR to AOR and a higher ratio of NO3--N GR to AOR might indicate there were accumulation of nitrite and/or nitrate for some reasons.
It can also be seen from Table 1 that the pilot performance was not adversely impacted by the bulk NH4+-N concentration in the range from 37.5mgN/L to 225.2mgN/L. How low the bulk NH4+-N concentration can go without compromising the PN/A performance is one of the critical operational conditions, since this lowest bulk NH4+-N concentration will determine the maximum NH4+-N removal that can be achieved in the sidestream treatment by ZeeNAMMOXTM process. For example, the maximum NH4+-N removal would be around 90% at a feed NH4+-N concentration of 375mgN/L.
Because the feed NH4+-N concentration in sidestream wastewater is often higher than 375mgN/L, NH4+-N removal of greater than 90% can be easily achieved by ZeeNAMMOXTM process. Therefore, it should be understood that the NH4+-N removals in Table 1 were low mainly because impacts of the bulk NH4+-N concentration, rather than higher removal, were the focus of the studies.
NOB suppression: The microorganisms of interest in ZeeNAMMOXTM biofilm might include AOB, Anammox, denitrifiers, NOB and comammox (Figure 1). We will focus on AOB, Anammox and NOB for the discussion. Since all these pilots were recovered from different levels of NOB proliferation, the control strategies must have caused NOB suppression. It is therefore hypothesized that, while NOB was suppressed, the competitive microorganisms of NOB would gain ground, which would further facilitate NOB suppression. This hypothesis is supported by the plot of the ratio of NO3--N GR to AOR versus the TINRR using the data from all pilots (Figure 2). In Figure 2, the ratio of NO3--N GR to AOR was trending down with the increase in TINRR. In addition, The ratio varied in a wide range when TINRR was below 2.5gTIN/m2/d and it became lower and relatively constant when TINRR was between 2.5 to 5.5 gTIN/m2/d. Once the TINRR was above 5.5 gTIN/m2/d, the ratio was getting close to the theoretical value of 0.11 for the PN/A process. This downtrend might indicate that the
combination of AOB and Anammox might have gradually outcompeted the combination of AOB and NOB. Both AOB and Anammox could benefit from NOB suppression since NOB competes against AOB for O2 and against Anammox for NO2--N. Due to the counter-diffusion in ZeeNAMMOXTM biofilm, the active aerobic microorganisms (AOB, NOB and comammox) would be mainly located in the inner layers of biofilm adjacent to the membrane where O2 is supplied while the active anoxic or anaerobic microorganisms (Anammox and denitrifiers) would be mainly located in the outer layers of the biofilm. When NOB is suppressed by some control strategies, AOB and Anammox might form a new biofilm community that would limit the expansion of NOB as a result of the limited space and limited substrates (O2 and NO2--N) in the inner layers of the biofilm. It seems that Anammox was doing so as indicated by the TINRRs obtained from the Anammox activity batch tests, which were
134
always higher than those from the continuous operations, up to 2 times higher in some cases.
Table 1. Summary of main operational conditions and performance of pilots at steady state
Paramete Unit Pilot A Pilot B Pilot C Pilot D
main operational conditions
wastewater N/A synthetic synthetic centrate diluted centrate
bulk T °C 35.0 34.0 31.5 34.9
bulk pH unitless 7.8 8.3 7.5 6.7
Bulk NH4 + -N mgN/L 37.5 138.2 152.1 225.2
Specific ALR[1] gN/m2/d 9.7 22.8 5.9 8.7
Performance summary
Specific AOR[2] gN/m2/d 4.9 5.6 4.2 3.30
Specific TINRR[3] gN/m2/d 4.2 5.5 3.20 2.60
- [4]
Specific NO3 -N GR gN/m2/d 1.4 0.80 1.00 0.29
Volumetric AOR[2] kgN/m3/d 0.93 1.1 0.80 0.63
Volumetric TINRR[3] kgN/m3/d 0.80 1.05 0.61 0.49
NH4 + -N removal % 51 25 71 38
NO3 - -N GR/AOR ratio unitless 0.29 0.14 0.24 0.09
TINRR/AOR ratio unitless 0.86 0.98 0.76 0.79
Figure 1. main substrates, microorganisms and processes of interest in ZeeNAMMOXTM biofilm
Figure 2. NO3--N reduction caused by increase in Anammox TIN removal
135 References
Coutts, D., Long, Z., Peeters, J., Houweling, D., DiPofi, M., 2020(a). Side Stream Treatment with Membrane Aerated Biofilm Reactors - no Carbon, no Alkalinity and no Bubbles. IWA World Water Congress &
Exhibition 2020.
Coutts, D., Di Pofi, M., Baumgarten, S., Guglielmi, G., Peeters, J., Houweling, D., 2020(b). Side-Stream treatment with Membrane Aerated Biofilm Reactors – the Simple, Robust and Energy Efficient Path. IWA Nutrient Removal and Recovery Conference 1 - 3 September 2020, Helsinki, Finland
Lackner, S., Gilbert, E.M., Vlaeminck, S.E., Joss, A., Horn, H. and van Loosdrecht, M.C.M. (2014) Full-scale partial nitritation/anammox experiences – An application survey, Water Research, 55, 292-303
Strous, M., Heijnen, J.J., Kuenen, J.G., Jetten, M. S. M., 1998. The sequencing batch reactor as a powerful tool for the study of slowly growing anaerobic ammonium-oxidizing microorganisms. Appl Microbiol Biotechnol (1998) 50: 589-596.
Presenting Author Dr. Long
My position: Lead Process Engineer
My affiliation: Suez Water technologies and Solutions
Is the presenting author an IWA Young Water Professional? Y/N (i.e. an IWA member under 35 years of age) No.
Bio:
Zebo Long, Lead Process Engineer working for Suez Water Technologies &
Solutions based in Oakville, Ontario Canada. Zebo has over 10 years of working experience in water and wastewater treatment, with expertise in new process &
product development, system simulations, and microbial characterizations. As a key member of a team, Zebo had developed ZeeLung membrane aerated biofilm (ZeeLung MABR) technology, which is widely accepted as a disruptive process due to its high energy efficiency, low footprint, and reliable performance. Zebo is
currently focusing on the development of ZeeNAMMOXTM technology, an innovative sidestream partial nitritation and Anammox (PN/A) process using ZeeLung MABR technology.
136
Multi parameter testing of MABR on several full-scale pilot plants
Nathan, N., Shefer, I., Dagai, L. and Shechter, R.
Fluence corporation, Caesarea, Israel
Keywords: membrane aerated biofilm reactors; biofilm processes; BNR;
Summary of key findings
This work reviews the testing of differtent MABR process schemes and configurations on large scale in one site over a period of about 3 years. Results from all systems combined, show that MABR nitrification rate correlates to ammonia concentration and loading rate. Results also show that mixing frequency influences MABR nitrification rate and can be used in design and operation of MABR systems. OTR (permeation) follows nitrification rate, indicating that
ammonia is limiting the rate and that oxygen does not pass through the biofilm and into the water.
Background and relevance
Membrane Aerated Biofilm Reactors (MABR) are gradually becoming an established technology with first large scale retrofit projects reported in 2019 and multiple package plants sold in East Asia. The promise of MABR relates to lower energy consumption for aeration as well as total nitrogen removal through simultaneous nitrification and denitrification (SND) [Shechter et al, 2019].
In view of the growing interest and in order to promote wider acceptance and availability of this emerging technology, more data on process performance is required as basis for design and evaluation. Efforts to provide this data are made by technology providers and researchers in pilot plants and laboratories [Sathyamoorthy et al, 2019]. Other work is conducted in order to gain fundamental understanding of biofilm population and the biochemical mechanisms involved in the unique process of cross directional diffusion to the biofilm [Kim et al, 2019].
One of the main parameters that characterized MABR as a biofilm process using an installed surface area is the nitrification rate, which is influenced by temperature, loading rate and other parameters.
This work maps some of the parameters influencing the nitrification rate in MABR through testing in several different large scale systems of different process configurations, but using the same type of spirally wound membrane module.
Description of the Pilot Plants
The R&D site at the Mayan Zvi facility on the Carmel Coat in Israel hosts 8 full scale MABR pilot and demo systems of different sizes, testing process configurations, operating conditions and module structures. This work will focus on results from 4 different systems running at the site, dealing with process and operating parameters as described in table 1 below.
Table 1: Specifications of the different pilot systems System name Tank vol.
m3 Process
structure MABR %
of volume Feed flow
m3/d Flow variation
Old configuration 15 2 stages 100% 30-36 Constant
R&D container 8 2 stages 100% 7-12 Constant
ASPIRAL pilot 70 4 (note 1) 60% 80-150 Constant
SUBRE 2900 3 (note 2) 20% 4000-5000 Diurnal
Notes to table 1:
(1) 3 MABR stages in series followed by an aerated volume
(2) Anaerobic, anoxic (installed with MABR modules), and most of the volume aerobic
All systems were fed with wastewater after pretreatment through a 1-1.5 mm fine screen. Each system had its own secondary clarifier and RAS circulation, however the SUBRE system is a retrofit of 1 of 2 lines of the Mayan Zvi plant, sharing the same secondary clarifiers.
All MABR stages in all systems were mixed periodical a few times per hour for 30±10 seconds, by turning on the aeration to a grid of coarse bubble diffusers laid beneath the MABR modules. The aeration is turned on for 4%-8% of the time and is considered to influence nitrification rate through
137 mixing mass transfer.
Results and Discussion
Nitrification rate and the influence of different parameters on it are of primary interest in MABR.
Therefore, these results are presented first for the different systems in figure 1, showing data for the nitrification rate in the first MABR stage of each system. The median rates are in the range 1.5-2.2 g/d/m2 and the average rates are in the range 1.4-2.5 g/d/m2. These correspond to the NH4-N median and average concentrations both shown in figure 2 to be in the range 10-23 mg/l. It can be generalized that the rates were higher in systems where the concentrations were higher and vice versa.
Figure 1: Statistical presentation of nitrification rates by system
Figure 2: Statistical presentation of ammonia concentration by system
The differences in ammonia concentration in MABR stage 1 of all the systems plotted above, and the corresponding nitrification rate, seem to be related to the different process configurations.
Specifically, when MABR stage 1 occupies a smaller volume fraction of the entire process, its loading rate is higher, and as a result the % removal is lower, the rate is higher and the outlet concentration is higher, and vice versa. For example - in the SUBRE MZ and ASPIRAL systems MABR stage 1 occupies about 20% of the aeration tank volume, and thus rates and outlet
concentrations are both higher than in the Old Configuration and R&D Container systems where it is about 50% of the total volume.
The influence of the mixing frequency on the nitrification rate as shown in figure 3, was mainly tested in the ASPIRAL and the Old Configuration systems, at levels of 10 times per hour (every 6 minutes) and 5 times per hour (every 12 minutes). It is seen that the mixing frequency has a strong influence on the nitrification rate, increasing in average from about 1.5 g/d/m2 at 5 times per hour to 2.3 g/d/m2 at 10 times per hour. The relatively wide distribution of rates at each level of mixing is mostly attributed to the different system structures and the loading implications as mentioned above. The conclusion is that mixing frequency is associated with both nitrification rate and energy consumption.