Aeration strategies and total nitrogen removal in a hybrid aerated treatment wetland Open Access

Treatment wetlands (TWs) have gained considerable attention for wastewater treatment due to their robustness, low operation and maintenance requirements and ability to deliver high-quality effluent (Morvannou et al. 2015). Total nitrogen (TN) in wastewater is usually removed through the use of a traditional nitrification–denitrification pathway. The process requires two steps: aerobic autotrophic nitrification to transform ammonia into nitrates and anoxic heterotrophic denitrification, to transform the nitrates into nitrogen gas. Both phases demand specific environmental conditions, foremost the oxygen concentration and organic carbon source (Saeed & Sun 2012). Passive TWs usually require two or three treatment stages to address TN removal by implementing separate aerobic and anoxic filters. Therefore, depending on the design chosen, the land requirement can vary between 4 and 10 m²/PE, which is a disadvantage for those systems (Dotro et al. 2017) compared to conventional systems (0.12–0.3 m²/PE for activated sludge).

In addition, passive TWs have some limitations regarding pollutant removal; namely, horizontal-flow (HF) wetlands do not nitrify to a great extent, and vertical flow wetlands nitrify but are limited in their capacity to remove TN (Masih 2017; Zhuang et al. 2019; Cross et al. 2021). While vertical flow (VF) TWs exhibit almost complete ammonium removal (upwards of 90%), passive TWs (either vertical or horizontal) achieve only 20–50% TN removal (Cross et al. 2021). Consequently, considering the restrictions imposed on nutrient discharge and the problem of available land area in urban areas, intensified TW designs have been developed. The term ‘intensification’ implies the use of energetic, chemical or operational modifications to increase pollutant removal (Fonder & Headley 2010). By adopting an energy-oriented approach, it becomes possible to enhance the availability of oxygen within a TW, thereby promoting the removal of oxygen-requiring compounds such as carbon and ammonium.

Aeration in TW refers to the injection of air into the system, enabling aerobic conditions in a saturated zone. Aerated treatment wetland (TW) technology was created over 25 years ago (Wallace 2001), and its implementation has witnessed a notable surge over the past decade. Today, approximately 500 aerated TW systems are operational on a global scale (Nivala et al. 2020). The main advantage of aerated TWs is their ability to completely nitrify (Masih 2017; Zhuang et al. 2019; Cross et al. 2021), and their compactness compared to passive TWs (1 m²/population equivalent (PE) versus 4–5 m²/PE) (DWA 2017); consequently, rendering them more viable for implementation where land availability is limited, such as in urban areas. Despite the disadvantage of increased energy costs, those systems operate at low pressure and consume less energy than conventional intensive wastewater treatment plants (Rochard 2017).

Moreover, intermittent or controlled aeration strategies (compared to continuous 24/7 aeration) have demonstrated the potential to enhance denitrification processes in aerated wetlands (Dotro et al. 2017). Continuous aeration has been shown to achieve TN concentration removal ranging from 15 to 60% (Nivala et al. 2019; Cross et al. 2021) and intermittent aeration has been found to optimize TN removal, yielding removal efficiencies of up to 90% (Masih 2017).

Recent configurations and adaptations have been mixing the aeration and hybrid TW to go even further in the TN removal. One of these systems, known as Rhizosph'air^® (Troesch et al. 2020), allows the nitrification and denitrification processes in a single-stage TW. The aeration in this system provides increased oxygen transfer, resulting in the reduction of the system footprint from 2 m² PE⁻¹ (classical two-stage French vertical flow wetland) to 0.8–1.2 m² PE⁻¹ depending on the conditions (Dou et al. 2016; Cross et al. 2021). This hybrid system has been installed around France (18 plants) and one of them was monitored for 2 years, treating domestic wastewater for 1400 PE (Prost-Boucle & Molle 2021). Regardless of the tested aeration mode, the treatment performances for COD, BOD₅ and TSS consistently remained high and stable. However, under the aeration mode of 12 h per day divided into four cycles, nitrification was successfully achieved, while denitrification was limited due to insufficient carbon availability. To enhance TN removal, a reduction in the daily aeration duration proved beneficial. By adopting a four-cycle aeration mode with a total duration of 3 h per day, the system was capable of removing 97% COD concentration and 82% TN concentration, remaining 19 mg TN/L at the outlet.

However, they tested limited aeration strategies with high variability in inflow and outflow. This current research intends to go further by testing the difference between each experimental condition in a controlled system, allowing a comparison to be made. Thus, to guarantee a stable and optimized nitrogen removal, it is crucial to adapt aeration strategies to meet specific treatment goals, ranging from complete nitrification to nearly complete removal of TN. Aeration strategies should be based on carbon and nitrification requirements, considering the absence of oxygen for denitrification. Therefore, this research aims to investigate the impact of aeration strategy in a hybrid aerated vertical/horizontal flows TW on the performance of TN removal.

The industrial pilot-scale TW studied is part of the REFLET research platform (https://eng-reversaal.lyon-grenoble.hub.inrae.fr/equipment-platforms/reflet), located in Craponne, France, a commune situated in the Auvergne Rhône Alpes region. The pilot system had 1 month of startup before starting the sampling campaign. Its commission started in January 2019. In total, the pilot covers an area of 20 m² (8 × 2.5 m²). It replicates the Rhizosph'air system, which comprises a single-stage French vertical flow TW, receiving raw wastewater directly on the surface of the filter, above an aerated horizontal-flow TW. This arrangement enables the TW system to provide combined primary and secondary wastewater treatment.

The TW receives wastewater from around 4 m³/day (26 PE), collected by a combined sewer from the upstream sewage basin produced by the community. The domestic wastewater has a ratio of COD/BOD₅ of approximately 2.0, showing that is a typical municipal untreated wastewater (Kadlec & Wallace 2009).

The saturation layer (1 m depth) is shared between the two parallel filters. Aeration of the system is accomplished by employing 120 driplines positioned at the bottom of the saturation layer (60 diffuser/m²). The aeration process is facilitated by two side channel blowers (Becker SV 5.90/1). The estimated supply airflow rate for each blower is approximately 15 m³/h (equivalent to 1.5 m³/h/m²). It is worth noting that no gas flow meter or temperature sensor was installed and therefore, the airflow rate estimation is based on these specifications.

Aeration was based on time (‘T’ condition) for the first five experiments, e.g., hours of aeration per day spread across a number of aeration cycles. Initially, a low HLR (0.15 m/day) with the longest aeration time (9 h) was tested as a C1 condition to favor plant establishment as well as good aerobic performances: T–9 h–0.15. The aim of this condition was to not overload the system from the beginning and fasten nitrifying bacteria growth in winter. Then, the aeration time and the number of aeration cycles were decreased (C2: T–3 h–0.15). To better identify the treatment capacity of the system, the HLR was gradually increased (C3: T–3 h–0.25 and C4: T–3 h–0.35), and then the pollutant load was increased by adding ammonia to the influent (C5: T–3 h–0.35–NH₄). Since the influent comes from a combined sewer and, consequently, the inlet ammonium concentration was low, ammonium addition (the goal was to reach around 80 mg NH₄-N/L) allowed us to test the capacity of the system for more stringent wastewater concentrations (C5, C6 and C7 conditions).

To increase denitrification, the timing of the aeration was coordinated with pulse feeding on the PF (‘F’ condition), optimizing the use of the available carbon in raw wastewater. Additionally, the aeration was decreased on the SF (C6: F–3.3 h–1.7 h–0.35–NH₄). For the last condition (C7: F–2.5 h–0 h–0.35–NH₄) aeration was decreased on the PF (2.5 h) and completely stopped on an SF (0 h) to favor denitrification. In these cases, aeration was set to start 40 min before the batch and stop after 20 min for C6 and 15 min for C7 (time needed to degrade the oxygen inside the filter and remain in an anoxic environment).

The laboratory analysis occurred once per week, sampling the inlet and outlet through 24-h composite samples. Each week, the filter was sampled alternated (different inlet and outlet). The following parameters were analyzed: total chemical oxygen demand (COD), filtered COD (fCOD), total suspended solids (TSS), NH₄-N, NO₃-N, NO₂-N and Total Kjeldahl Nitrogen (TKN). The chemistry laboratory followed the European standard methods (NF EN 872, EN ISO 10304-1, NF EN ISO 14911, NF EN 25663, NF EN 6878, NF EN ISO 9963-1). TN was calculated as the sum of NO₃-N, NO₂-N and TKN concentrations.

Table 2 presents the treatment performance for the experimental conditions employing the timing aeration strategies (‘T’). Notably, the concentrations measured at the inlet exhibited variation across different conditions, evidenced by lower COD concentrations for C1 (T–9 h–0.15) (384 mg COD/L) and C4 (T–3 h–0.35) (280 mg COD/L), emphasizing the influence of seasonal factors. These periods corresponded to the winter season, characterized by wastewater dilution due to rainwater inflow and intrusion of a greater quantity of clear water in the combined sewer. Overall, the system demonstrated high performance in removing carbon and solids pollution regardless of the experimental conditions, with an average outlet concentration of less than 29 mg COD/L and 4 mg TSS/L. These results highlight the filtration capacity provided by the TW. Moreover, these findings align with previous studies on the treatment efficiency of similar full-scale systems (Petitjean et al. 2021), affirming the effectiveness and stability of Rhizosph'air^® in removing carbon and solids from raw wastewater.

However, when ammonium was added to the influent raw wastewater (C5: T–3 h–0.35–NH₄), removal efficiency decreased to 89.2 ± 12.5%, indicating that nitrification had reached its maximum. To further enhance ammonium removal, the possible strategies are to increase the number of aeration cycles per day, the duration of aeration for each cycle or to automate the aeration using online sensors. Nevertheless, the result reinforces previous observations about the ability of a hybrid aerated TW to achieve ammonium removal over 90% (Prost-Boucle & Molle 2021).

Comparing the C1 (T–9 h–0.15) and C2 (T–3 h–0.15) conditions (6 cycles versus 4 cycles of aeration/day), the C2 condition showed significantly higher denitrification, as indicated by the lower average outlet nitrate concentration (C1: 34 mg NO₃-N/L; C2: 25 mg NO₃-N/L), resulting in higher TN percent concentration removal for C2 of 47%. This can be attributed to the longer duration of the anoxic condition, with an average of 13.1 h per day in the PF and 3.2 h/day in the SF, which favored denitrifying bacteria activity. In Figure 5, for condition C2, whenever the filter is being fed, there is more carbon to degrade and the oxygen is rapidly consumed, while in the SF, the oxygen concentration increases, exhibiting longer aerobic conditions (3.2 h/day for the PF and 20.5 h/day for the SF).

When a similar aeration strategy (3 h/day) was applied but the HLR was increased from 0.15 m/day (C2) to 0.25 m/day (C3) and 0.35 m/day (C4), there was a decrease in the average effluent NO₃-N concentration (C2: 25.3 ± 6.2 mg/L; C3: 16.0 ± 2.4 mg/L; C4: 13.6 ± 0.2 mg/L). Consequently, TN removal increased from C2 (47.2 ± 19.6%) to C3 (62.6 ± 17.7%). The lower percent concentration TN removal in C4 (T–3 h–0.35) (42.8 ± 25.7%) was attributed to the lower inlet TN concentration and lower hydraulic retention time as the HLR was increased. During conditions C2 and C3, denitrification was more efficient as the HLR were higher and higher inlet COD concentrations (about twice as high) provided higher availability of organic carbon. In addition, for these three experimental conditions, the loads applied in NH₄-N (2.3–18 g/m²/day) and TN (2.4–21.7 g/m²/day) remained within the same range of values, and the treated loads (2.1–17.9 g NH₄-N/m²/day and 0.7–17.7 g TN/m²/day) did not differ among the experimental conditions (Figure 4). These treated loads are higher than aerated VF (5.2 g NH₄-N/m²/day) or aerated HF (7.3 g NH₄-N/m²/day) (Dotro et al. 2017).

For condition C5, the condition with an ammonium addition (C5: T–3 h–0.35–NH₄), the inlet average concentration was 108 ± 18 mg TN/L (between 9 and 19 g TN/m²/day of applied load), approximately 2–4 times higher than the inlet TN concentration in other conditions. This operational condition achieved a high TN removal, reaching 71%. However, effluent NO₃-N concentrations were still measurable, indicating limitations in the denitrification process. The oxygen sensors reported an average of 18.1 h/day of anoxic conditions in the PF and 9.4 h/day in the SF. As Figure 5 shows, it is the first time that the duration of an anoxic condition was constantly longer than aerobic condition. Therefore, with high nitrification and a long anoxic time during this experimental condition, it became clear that the limitation of denitrification is not due to a lack of nitrates or excessive oxygenation in the saturated zone, but probably to the lack of organic carbon in the saturated zone. This issue is commonly reported in the literature (Bezbaruah & Zhang 2003; Wang et al. 2022; Yao et al. 2023).

Based on the assumptions of Morvannou et al. (2015) (80% concentration removal for COD and 50% for TKN in the unsaturated zone of French VF TWs), the COD/NO₃-N ratio arriving in the saturated zone can be estimated. According to Morvannou et al. (2017), this ratio should be at least 2.0 at the entrance to the saturated zone (and a minimum residence time of 0.75 days) for efficient denitrification. Estimations in this study would indicate that only condition C5 (T–3 h–3.5–NH₄) lacked sufficient biodegradable carbon for denitrification. The deficit would be approximately between 40 and 120 mg COD/L (results are presented in Supplementary material).

It was not possible to match aeration requirements to batch feeding by controlling aeration strategies on a temporal basis. When a batch arrived during a period of high dissolved oxygen concentration in the saturated zone, oxygen was used for carbon degradation resulting in a lack of carbon for denitrification during anoxic periods. Under the conditions tested from C1 to C5, the balance between nitrification and denitrification periods was not ideal for effectively transforming nitrogen forms while maintaining a sufficient carbon source for the denitrification process. Thus, to remedy this deficiency and attempt to optimize denitrification, a new aeration strategy was tested that coordinated the aeration period and batch feeding.

Table 3 presents the treatment performance results for the experimental conditions employing the aeration strategy according to feeding time (‘F’). The first condition C6 (F–3.3 h–1.7 h–0.175–NH₄) had similar influent concentrations to condition C4 (T–3 h–0.175), characterized by diluted wastewater and subsequently, lower COD and NH₄-N influent concentrations. Despite this dilution, the system demonstrated high efficiency in removing those pollutants, even in wet weather conditions (refer to the Supplementary material for the local rainfall data), demonstrating some flexibility in the performance of the system.

In contrast, condition C7 (F–2.5 h–0 h–0.175–NH₄) exhibited an opposite trend compared to the other conditions: the majority of the day, the saturated zone remained anoxic (17.1 h/day in the PF and 22.9 h/day in the SF), leading to an almost complete denitrification process at the saturated zone, with 30.3 ± 3.1 mg NO₃-N/L effluent concentration. However, as the aerobic environment decreased, 5.7 h/day in PF and 0.9 h/day in SF, it also reduced the concentration removal of NH₄-N (80.0 ± 5.3%). The dissolved oxygen concentration observed for the C7 condition (supplementary material) was similar to those observed by Uggetti et al. (2016), who also found an average removal of 66% for TN removal.

Despite the extended anoxic periods, denitrification was again hindered by organic carbon limitation. The C/N ratios for C6 and C7 averaged 1.9 and 1.3, respectively. Thus, the carbon source from the batch was not sufficient to meet the denitrification requirement, causing nitrate accumulation (Wang et al. 2020). To overcome this, it would be necessary to increase the carbon source by using an external chemical product that is readily biodegradable, such as methanol or acetic acid (Fu et al. 2022).

The Rhizosph'air hybrid aerated TW demonstrated high removal of carbon and suspended solids removal for all seven experimental conditions tested. The conditions varied with influent concentration, duration of aeration, timing of influent loading and aeration, different hydraulic loading rates and addition of ammonium in the influent. TN removal was directly impacted by each operational change. In the experimental conditions with time-based aeration (C1–C5), percent ammonium concentration removal exceeded 90%. Aerating the system for 9 h total per day (6 cycles of 150 min aeration per cycle) resulted in the lowest TN removal (5%). Increasing the HLR (from 0.15 to 0.25 m/day) with 3 h aeration per day, the TN removal improved from 47 to 62%. This is the result of a better balance between oxygen demand and supply. When ammonium was added to the influent (C5: T–3 h–0.35–NH₄), nitrification was preserved while denitrification increased, reaching 71% TN concentration removal.

It is important to note the seasonal effect, particularly during winter when the influent on the pilot in this study was diluted by clear water intrusion (C4: T–3 h–0.35 and C6: F–3.3 h–1.7 h–0.35–NH₄), resulting in lower influent COD and NH₄-N concentrations. Consequently, this had an impact on the efficiency of the system, by decreasing the TN removal, while preserving low outlet TN concentration (around 15 mgN/L). However, despite these effects, complete nitrification still occurs due to efficient oxygen transfer.

When aeration was set according to the batch feeding (C6: F–3.3 h–1.7 h–0.35–NH₄ and C7: F–2.5 h–0 h–0.35–NH₄), the subsurface conditions favored denitrification with longer periods of anoxic conditions and dissolved oxygen concentrations of approximately 0.2 and 0.9 mg O₂/L. The condition C7 (F–2.5h –0 h–0.35–NH₄) removed 71% of TN, similar to the performance of condition C5 (T–3 h–0.35–NH₄) (71%). The change in aeration strategy from time-based to batch time provided an instantaneous source of carbon for denitrifying bacteria and consequently resulted in lower NO₃-N concentrations at the outlet. Nevertheless, this condition reached its maximum denitrification rates but showed an increase in outlet ammonium concentrations (18.3 ± 7.1 mg NH₄-N/L effluent concentration). This underlines the importance of both oxygen supply and sequencing.

Therefore, to optimize the hybrid aerated TW system in this study and decrease effluent nitrate and TN concentrations without adding carbon, the following modifications are suggested:

• Optimize the use of carbon for denitrification by stopping aeration earlier before the influent batch feeding to ensure anoxic environmental conditions when the system is fed.

• Use online sensors (pH, redox or O₂) to identify bending points or patterns related to nitrification and denitrification dynamics. This information could be used to control the operation of the aeration system based on the influent pollutant loads (e.g., treatment on demand).

The authors would like to thank the Office Français de la Biodiversité, Rhône–Méditerranée–Corse and Adour-Garrone water agencies as well as EUR H2O'Lyon for its financial support to the TONIC research project. In addition, we would like to thank Olivier Garcia for his contribution to the field.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.