Comparative study of anammox-mediated nitrogen removal in three reactor configurations

Abstract

Anaerobic ammonium oxidation (ANAMMOX) is an efficient and cost-effective process developed for biological nitrogen removal from wastewater. However, widespread application of the ANAMMOX process for wastewater treatment remains constrained due to the slow growth of ANAMMOX bacteria, propensity for out-competition by fast growing microbes, and its sensitivity to environmental and operational conditions. Consequently, understanding the influence of mixing conditions in different reactor configurations on this process is paramount in its improvement. This study focused on the comparative analysis of ANAMMOX-mediated nitrogen removal in a hybrid up-flow anaerobic sludge blanket reactor (H-UASB), moving bed biofilm reactor (MBBR) and a gas-lift reactor (GLR). The study involved experimental study of nitrogen removal, bacterial population dynamics and physical properties of the bacterial biomass within the reactors, as well as the description of process performance and the growth of nitrifying and ANAMMOX bacteria in the reactors using a calibrated mechanistic model. All the reactors were operated for 535 days using the same synthetic feed under anaerobic conditions. K1-type carrier materials were added to each reactor for biofilm development. The concentrations of ammonium (NH4 + ), nitrite (NO2 - ) and nitrate (NO3 - ) in the effluent from the reactors were determined colorimetrically. Among the three reactors, MBBR displayed the highest nitrogen removal efficiency (NRE) during the study (66±36%), and contained the lowest concentration of free ammonia (FA) (19±22 mg-N/L) and free nitrous acid (FNA) (0.001±0.001 mg-N/L). In comparison, the NRE and the concentrations of FA and FNA in H-UASB during the study were 63±28%, 91±41 mg-N/L and 0.006±0.004 mgN/L, respectively, while in the GLR, they were 54±39%, 28±29 mg-N/L and 0.002±0.002 mg-N/L, respectively. Based on the ratios of NO2 - consumed to NH4 + consumed, and the ratios of NO3 - produced to NH4 + consumed, the start-up of ANAMMOX process was faster in the MBBR (144 days) compared to H-UASB (193 days) and GLR (272 days). MBBR also displayed less fluctuations in the NREs and nitrogen removal rates (NRRs) during the study compared to H-UASB and GLR. The microbial communities in the suspended biomass in the reactors were characterised using high-throughput sequencing on an Illumina MiSeq platform on days 125, 192, 260, 309 and 535, while the microbial communities in the biofilms were only characterised on day 535 (last day) due to slow biofilm development. Gradual increases in the relative abundance of ANAMMOX bacteria were observed in the suspended biomass in all the reactors between days 125 and 309, which corroborated the observed increases in the NREs. The relative abundance of ANAMMOX bacteria remained consistently higher in H-UASB during the study than in MBBR and GLR. On the contrary, the highest relative abundance of ammonia oxidising bacteria (AOB) was observed in the suspended biomass in the MBBR on day 125 at approximately 38%, while the highest relative abundance of nitrite oxidising bacteria (NOB) and complete ammonia oxidising (COMAMMOX) bacteria was recorded in the suspended biomass in the MBBR at approximately 30% and 5%, respectively. In all the reactors, the relative abundance of AOB in the biofilms and the suspended biomass was comparable on day 535. In addition, on day 535, higher relative abundance of NOB was observed in the biofilms in both GLR and H-UASB at approximately 7% compared to the suspended biomass, while their abundance in the suspended biomass in the MBBR was comparable to that recorded in the biofilms. Furthermore, in both H-UASB and MBBR, higher relative abundance of ANAMMOX bacteria was observed in the suspended biomass compared to the biofilms on day 535, while comparable abundance was observed in the GLR. The highest total microbial diversity (Shannon and Simpson indices) and evenness (Pielou’s Evenness) was observed in the suspended biomass in the MBBR. Granulation of the suspended biomass was observed in both GLR and H-UASB, while the suspended biomass in the MBBR was flocculent. In the MBBR, the colour of the biomass had turned brown on day 125, while the biomass in H-UASB and GLR on this day was tawny and dark-tawny, respectively. However, on day 309, the biomass in all the reactors had turned red, corroborating the highest relative abundance of ANAMMOX bacteria observed during the study. Faster attachment of biomass on the carrier materials in MBBR was observed in the course of study compared to H-UASB and GLR. On the last day, the concentrations of the biomass on the carrier materials in the MBBR was also higher (12 mg/carrier) in the MBBR than in the H-UASB (8 mg/carrier) and GLR (10 mg/carrier). Activated sludge model 1 (ASM 1), which was modified by separating the activities of Nitrospira spp. from those of Nitrobacter spp. as well as by adding both ANAMMOX and COMAMMOX bacterial activities, was used to describe process performance in the reactors. The modified ASM 1 was able to predict the trends in the effluent concentrations of NH4 + , NO2 - and NO3 - in all the reactors. In addition, the correlation of the actual relative abundance of nitrifying and ANAMMOX bacteria, with the model-predicted relative abundance, was positive. The model also indicated higher heterotrophic activities in both GLR and MBBR compared to H-UASB, an indication that continuous mixing in MBBR and alternation of plug-flow conditions with internal gas circulation in GLR favoured heterotrophic bacterial growth. However, the model was limited in predicting the fluctuations in bacterial abundance and the fluctuations in the effluent concentrations of NH4 + , NO2 - and NO3 - in the reactors. The obtained results indicate that better-mixed conditions in the MBBR led to comparable relative abundance of nitrifying bacteria between the biofilms and the suspended biomass, while plug-flow conditions in the H-UASB favoured ANAMMOX bacterial growth in the suspended biomass and the nitrifying bacterial growth in the biofilms. The alternation of internal gas circulation with plug-flow conditions in the GLR also favoured the growth of nitrifying bacteria in the biofilms. Overall, nitrogen removal in H-UASB was likely dominated by ANAMMOX process, while nitrogen removal in MBBR and GLR was as a result of combined ANAMMOX and sequential nitrification-denitrification processes. The novelty of this study stem from the impact of mixing conditions on process performance and microbial ecology of ANAMMOX-mediated systems.