The objective of this study is to improve knowledge on the integrated fixed-film-activated sludge (IFAS) system designed for nitrogen removal. Biofilm growth and its contribution to nitrification were monitored under various operating conditions in a semi-industrial pilot-scale plant. Nitrification rates were observed in biofilms developed on free-floating media and in activated sludge operated under a low sludge retention time (4 days) and at an ammonia loading rate of 45–70 gNH4-N/kgMLVSS/d. Operational conditions, i.e. oxygen concentration, redox potential, suspended solids concentration, ammonium and nitrates, were monitored continuously in the reactors. High removal efficiencies were observed for carbon and ammonium at high-loading rate. The contribution of biofilm to nitrification was determined as 40–70% of total NOx-N production under the operating conditions tested. Optimal conditions to optimize process compacity were determined. The tested configuration responds especially well to winter and summer nitrification conditions. These results help provide a deeper understanding of how autotrophic biomass evolves through environmental and operational conditions in IFAS systems.
Over the past decades in Europe, the growth of urban area and the discharge limits imposed on nitrogen effluent concentrations have made it necessary to upgrade or retrofit many wastewater treatment plants. Many plants nitrify/denitrify municipal wastewater via activated sludge systems designed with a mixed liquor solid retention time (MLSRT) around 10–15 days at 10 °C (Henze et al. 1996). When footprint is limited, more compact processes are required, and among the options available, the integrated fixed-film-activated sludge (IFAS) system offers an interesting solution for retrofitting-activated sludge plants to achieve nitrification at affordable cost (Randall & Sen 1996; McQuarrie et al. 2004; Stricker et al. 2009). The process uses free-floating media fluidized in activated sludge mixed liquor. The IFAS process maintains suspended culture by recirculation from the clarifier. Nitrification at low MLSRT is made possible by the growth of autotrophic biomass in the biofilm that developed on the carriers (Sriwiriyarat et al. 2008; Onnis-Hayden et al. 2011). In fact, biofilm containing attached biomass has a much higher retention time than freely suspended cells in the mixed liquor, which prevents total washout of slow-growing biomass such as nitrifiers.
Increasing nitrification efficiency necessarily implies increasing denitrification as well, which is dependent on carbon source, the ability to maintain anoxic conditions, and recirculation flow (Naidoo et al. 1998).
Biofilm culture responds to specific conditions that affect its microbial community. Diffusional limitations mean that the mixed liquor requires higher oxygen and substrate levels to ensure sufficient concentrations in the biofilm. Even with a low SRT, biofilm detachment still feeds the mixed liquor with autotrophic biomass (Morgenroth & Wilderer 2000; Elenter et al. 2007). Compared to MBBR processes, IFAS presents the same biofilm characteristics, but the presence of the mixed liquor lends it specific behaviour in terms of substrate utilization, oxygen demand and kinetics (Germain et al. 2007).
Most IFAS processes are located in North America where IFAS upgrades have proved very compatible with their specific operating conditions, such as 4–5 day MLSRT, low mixed liquor total suspended solids (MLTSS) concentration (1.5–2.5 g/L) and high food to microorganism (F/M) ratio (0.25–0.40 kgBOD5/kgMLVSS/d) (MLVSS: mixed liquor volatile suspended solids; BOD5: 5-day biochemical oxygen demand). However, there is a need for further work on a wider range of operating conditions, since few studies have been performed at pilot-scale (Boltz et al. 2009; Di Trapani et al. 2011) to test IFAS process limits and nitrification kinetics.
There is little information available to describe how the biofilm contributes to nitrification depending on applied operating conditions such as temperature, SRT and loading rate. To provide an improved understanding of the process and guidance on upgrading-activated sludge plants to IFAS, semi-industrial pilot-scale experiments were conducted over 2 years. The aim of this study is to find the limits of the process, assess the contribution of biofilm and investigate the nitrification and denitrification rates.
MATERIAL AND METHODS
Description of pilot-scale plant
A semi-industrial IFAS pilot-plant (Figure 1) treating 5–9 m3/d was built by IRSTEA at the La Feyssine (Lyon, France) treatment facility. The plant consists of a pre-anoxic reactor of 1,150 L (41%), a first aerobic suspended-culture zone of only 766 L (27%), and an IFAS aerobic reactor of 930 L. This configuration was designed to promote autotrophic biomass growth on carriers by limiting competition with heterotrophic biomass in the last reactor due to carbon substrate removal in aerobic basin 1. The IFAS zone is filled with fluidized floating carrier BMX1, cylindrical shape (10 mm diameter, 7 mm length) with cross inside made of high-density polyethylene and with a specific area of 545 m²/m3. Fill fraction is about 43% of bulk media. For process aeration, the coarse bubble diffuser was set up in aerobic basin 2 and a fine bubble diffuser was set up in aerobic basin 1. The pilot was run for about 2 years.
Mixed liquor separation is performed by a 350 L secondary settler equipped with a sludge return loop proportional to influent flow (150%). A sludge return loop from the IFAS aerobic reactor (200–250%) to pre-anoxic zone was integrated to perform denitrification.
From June 2012 to February 2014, proportional 24 h-flow composite samples were collected by refrigerated automatic samplers two times a week. Standard chemical analyses (chemical oxygen demand (COD), BOD5, , total Kjeldahl nitrogen (TKN), , , total suspended solids (TSS)) were carried out on dissolved and particulate fractions according to Standard Methods (APHA 2012). This approach was designed to characterize the influent and the global performances of the installation. Wastewater influent characteristics during the period were about 215 mgTSS/L (±54, 42 values), 433 mgCOD/L (±88, 42 values), 153 mgBOD5/L (±35, 38 values), 55 mgTKN/L (±13, 36 values), <0.45 mgNO3-N/L (69 values) and <0.06 mgNO2-N/L (69 values). A percentage of 4% of inert organic nitrogen has been estimated. The COD:TKN ratio was 8.1 gCOD/gN (±2.2, 36 values), which is below the typical French reference ratio of about 10 gCOD/g N, thus qualifying the influent as high nitrogen-loaded.
Two organic loading rate conditions were applied to the pilot as described in Table 1. The first period was obtained with an influent flow of 5 m3/d, which was increased in periods 2 and 3 to 9 m3/d. Organic loads during these three periods were 0.9 and 1.5 kgCOD/m3/d, respectively, corresponding to a 9 d and 4 d MLSRT. Periods 2 and 3 were operated in summer and winter, respectively, to observe the impact of temperature variation on high organic load and reach the autotrophic MLSRT limit by temperature change.
|Period 1||Period 2||Period 3|
|Unit||Nov. 2012–May 2013||June 2013–Nov. 2013||Dec. 2013–March 2014|
|Period 1||Period 2||Period 3|
|Unit||Nov. 2012–May 2013||June 2013–Nov. 2013||Dec. 2013–March 2014|
aSRT limit: threshold MLSRT value under which autotrophic biomass should be washed out. This value was calculated based on growth (μa,max = 0.8 d−1) and decay rates (ba = 0.15 d−1), assuming non-limiting substrate and oxygen concentrations.
The reactors were equipped with Hach Lange sensors to instantaneously record key data such as TSS, dissolved oxygen concentration, redox potential and air flow. All sensors were connected to a monitoring system. Continuous measurement of NO3-N and NH4-N in anoxic and aerobic tanks 1 and 2 were measured by AN-IS sc Hach Lange sensors calibrated every week. These measurements gave instantaneous dynamic readings of nitrate production (nitrification) or consumption (denitrification) rates and made it possible to assess nitrogen balance in each basin. A typical pattern of inflow rate was programmed to feed the reactor with dynamic hourly flow variation. Wastewater flow, return-activated sludge (RAS), mixed liquor circulation (MLC), and extraction flow were recorded instantaneously, with RAS and MLC being regulated.
Treatment performance monitoring
Mass balance to assess nitrification rate (in-situ determination)
In-situ nitrification was measured in the pilot under stabilized treatment conditions using a mass balance. Assimilation of nitrogen was assumed to be around 5% of the BOD5 removed (Henze et al. 1996).
Batch tests to assess nitrification rate (ex-situ determination)
Determination of ex-situ and in-situ nitrification rates allows estimation of the limitation of nitrification occurring in the process, i.e. substrate diffusion in the biofilm layers (a high difference between those two values means high-kinetics limitation).
The biomass attached on carriers was determined using a protocol inspired from Di Trapani et al. (2011) and was measured twice a month. Fifty carriers were removed from the IFAS tank, dried at 105 °C for 24 hours, and then weighed. Carriers were stripped of their biofilm by exposing them to ultrasound for 30 minutes, washing them with distilled water then re-drying at 105 °C, and then weighed.
RESULTS AND DISCUSSION
Removal performance vs operating conditions
Figures 2(a) and 2(b) present applied and removed volumetric loading rates in the total reactor volume for total COD (TCOD) and TKN. The pilot showed high TCOD removal efficiencies (>90%) during period 1 when applying volumetric loading rates up to 1.0 kg TCOD/m3/d (Figure 2(a)) but lower TCOD removal (80%) at higher loading rates (periods 2 and 3). During periods 2 and 3, no difference was observed with regard to the temperature (from 16 °C in winter to 22 °C in summer) even if the applied organic loading rate was three times higher than typically applied in activated sludge processes. However, the pilot effluent showed high TCOD at concentrations between 18 and 26 mgBOD5/L during periods 2 and 3.
TKN volumetric loading rates on the total reactor volume are shown in Figure 2(b). Throughout the experiment, the pilot demonstrated high TKN removal, with low effluent concentrations of 2.0–5.0 mgTKN/L in period 1 (TKN removal >90%) and 4.0–10.0 mgTKN/L in periods 2 and 3 (TKN removal >85%). As the effluent showed no rapid increase of TKN, it was hypothesized that the process did not reach its SRT limit. This slow decrease in removal efficiency suggested that the hydraulic retention time (HRT) limit for pollutant treatment had been reached. The results from the Broomfield wastewater treatment plant (WWTP) are around 99% ammonium removal efficiencies for a nitrogen load applied of 90 to 120 gNH4-N/m3/d at temperatures varying from 13 to 22 °C (Rutt et al. 2006); this plant operates with similar carriers (specific area of 500 m2/m3) and a MLSRT of 4 days. Our study confirms these observations while investigating a higher ammonium load at 150 gNH4-N/m3/d with 85% ammonium removal efficiency.
Ammonium loads removed in the aerobic zone are presented in Figure 2(c) as a function of effluent concentration. The pattern of removal rates corresponded to a rapid increase up to an effluent concentration of 3 mgNH4-N/L. Above this value, removal rate seems to be relatively stable at around 180 gN/aerobic m3/d. This representation was made by Di Trapani et al. (2011) who observed a lower result of about 150 gN/aerobic m3/d at 11.5 °C. A more recent study investigated an ammonium load of 250 gNH4-N/aerobic m3/d at 14 °C with 95% of removal rate (Di Trapani et al. 2013).
Limitations of nitrification in aerobic reactors
The actual and maximal nitrification rates obtained with in-situ and ex-situ protocols, respectively, were determined on sludge sampled in aerobic tanks 1 and 2 during the three periods. These nitrification rates were compared to conclude on the kinetic limitations in the aerobic reactors. The in-situ and ex-situ nitrification rates over the three periods are presented in Figure 3. High oxygen was supplied to the tanks over the periods: concentration range was 2.5–6.0 mgO2/L for aerobic tank 1 and 3.4–5.5 mgO2/L for aerobic tank 2.
In period 1 (lowest loading rate), nitrification was mainly performed by the mixed liquor with 3.5 mgN/L/h nitrified in in-situ conditions. The process MLSRT was significantly higher (7–10 days) than the MLSRT limit and allowed autotrophic biomass to grow in the mixed liquor. There was a significant difference between in-situ and ex-situ nitrification in aeration tank 2, 4.7 mgN/L/h and 11.0 mgN/L/h, respectively. High kinetics limitations occur in this tank which penalize nitrification rate.
In period 2, nitrification was still dependent on mixed liquor. In-situ and ex-situ results were similar for tank 1, showing no limitation of nitrification, whereas in-situ removal rate was about 57% lower than ex-situ nitrification rate, showing limited nitrification in tank 2. Ex-situ nitrification rates obtained in tank 2 were high, at 13.6 mgNO3-N/L/h, which is consistent with Regmi et al. (2011) who measured 12.8 mgNO3-N/L/h at the Newport News facility. This plant was operated at 21 °C and a MLSRT of 4.8 days.
Nitrification rates in aerobic tank 2 are promoted by higher volumetric ammonium loading rates in period 3. In-situ nitrification in aerobic tank 2 increased significantly from 6.0 to 8.8 mgNO3-N/L/h, whereas in parallel, the ex-situ and in-situ nitrification rates in aerobic tank 1 decreased from 7.0 to 4.0 mgNO3-N/L/h. Ammonium concentrations were higher in aerobic reactor 2 due to temperature decrease (i.e. periods 1 and 2 involved very low ammonium concentrations at night). Duration of nitrification was thus about 5 and 8 hours per day for periods 1 and 2 compared to almost 19 hours in period 3.
These results suggest a limitation of basin 2-nitrification by the first aerobic basin. Compared to ex-situ nitrification rates, the in-situ nitrification rates in aerobic reactor 2 were just 43% in periods 1 and 2 and up to 65% in period 3 when there was higher ammonium input into aerobic tank 2. The same case was reported from the Broomfield facility, with ex-situ nitrification at 11.5 mgN/L/h in the first aerobic IFAS reactor which decreased to 5.3 mgN/L/h in the second aerobic IFAS basin due to substrate limitation (Onnis-Hayden et al. 2007).
Nitrification by the biofilm
Biofilm-driven nitrification rates at high loading rates ranged from 0.73 (period 2) to 0.91 gNO3-N/m2/d (period 3). Regmi et al. (2011) reported rates of 0.89 gN/m2/d and 66% of nitrates produced by biofilm on the Newport News facility at high ammonium load. Our results confirmed the results of Regmi et al. (2011) in period 3 at low temperature; however, more nitrification took place in the mixed liquor in the higher-temperature period 2. Biomass content in the carriers followed a seasonal pattern, 6.0 (period 2) to 8.3 gTSS/m2 (period 3), correlated with decrease of mixed liquor performance at low temperature.
In period 1, only 50% of nitrification was performed by biofilm, as autotrophic biomass was equally distributed between mixed liquor and biofilm. When loading rate increased, more nitrates were produced by biofilm than by mixed liquor, with biofilm contributing 60 and 67% nitrification in periods 2 and 3, respectively.
Increased duration of high ammonium concentration led to improved nitrification rate in the biofilm. Biofilm rates increased 1.20-fold in period 3, due to the ammonium input to reactor 2 in which duration of nitrification rose from 8 to 19 h/d.
All results showed high nitrate concentrations (i.e. 8–15 mgNO3-N/L) in the treated effluent. This study found no increase in denitrification rates; in fact, denitrification rate was not correlated to operating conditions such as temperature. Denitrification rates varied between 1.0 and 3.9 mgNO3-N/gMLVSS/h. Additional observations seemed to demonstrate that the anoxic basin is limited by poor soluble organics substrate from raw wastewater, since the redox sensor showed very low readings and no oxygen. Denitrification rate is highly correlated to influent content such as soluble organic compounds, as described by many authors at both lab-scale (Naidoo et al. 1998) and industrial scale (Onnis-Hayden et al. 2007). To confirm this finding, a linear mathematical relation between denitrification process and soluble organics load was tested. Denitrification removal efficiencies were investigated as a function of gFCOD (filtered COD) per gNO3-N applied in the anoxic basin. The optimal ratio was about 4.0 gFCOD/gNO3-N to remove 80% of nitrates produced in the reactor. Compared to a FCOD/NH4-N ratio in the influent of 3.0 gFCOD/gNH4-N, denitrification may be limited by poor organic substrates coming into the anoxic zone to reach the 80% denitrification removal rate.
This study provides significant data on the IFAS process:
The IFAS technology employed here enabled high removal efficiencies (>85%) for carbon and nitrogen at carbon loading rates of 0.3 kgBOD5/kgMLVSS/d, thus performing about three times better than conventional nitrifying/denitrifying-activated sludge plants.
Nitrification rates appeared to reach a stable value of 180gN/aerobic m3/d at high-loading rate of 70 gN/kgMLVSS/d at 16–22 °C.
The MLSRT limit was not reached, and the mixed liquor continued to achieve nitrification even at a 4-day SRT. Aerobic reactor 1 achieved 25 to 60% of nitrification during the three periods depending on winter and summer temperatures. Ex-situ tests showed that the mixed liquor contains a significant quantity of autotrophic biomass.
Autotrophic biomass is widely present in biofilm and contributes 67% of total nitrification rates. Biofilm performance varied from 0.62 to 0.92 gN/m2/d and is directly linked to the ammonium load applied to aerobic reactor 2.
Our pilot-scale configuration specifically designed for nitrogen removal points to the following conclusions:
The pre-anoxic tank-based denitrification process was not adapted to these high loads due to limited available soluble carbon in influent. Denitrification rates could be enhanced by adding an external carbon source, such as methanol.
This IFAS configuration is able to nitrify at high organic loads. Aeration in reactor 1 could be decreased to promote a higher biofilm contribution to nitrification. Process optimization depends on ammonium load in aerobic tank 2 and residual nitrification in aerobic tank 1.
The next step planned is to calibrate an IFAS model to predict removal efficiencies under different operating conditions.
The authors thank Région Rhône-Alpes for co-financing this research, and M. Masson, L. Richard, S. Plétan, M. Arhror and P. Le Pimpec for assistance with sampling and/or analysis.