Traditional denitrification process consumes external organic carbon leading to an increase in treatment costs and carbon emission. A novel sulfide-driven denitrification methane oxidation (SDMO) system, which could simultaneously utilize CH4 and H2S in biogas as electron donors for denitrification to reduce the cost and carbon emission, has been successfully operated in two kinds of reactors SBR-MBfR and EGSB-MBfR for ∼ 200 days in this study. The performance and microbial community were explored meanwhile machine learning approach have been used for predicting the complex effects on SDMO’s denitrification efficiency. Result shows SBR-MBfR’s superior performance with the maximum nitrate removal efficiency > 90 % while the denitrification rates and biological activity were also better than EGSB-MBfR. The autotrophic denitrification (Auto-D) contributed 5 % much more for nitrate removal in SBR-MBfR than EGSB-MBfR while denitrification anaerobic methane oxidation (DAMO) was more prominent in EGSB-MBfR. Nineteen machine learning models were used for operation data training and random forest regressor was demonstrated to be the optimal model for predicting nitrate removal efficiency in SDMO. According to random forest regressor’s analysis, HRT presented the highest importance in affecting SDMO nitrate removal efficiency. After the introduction of biogas for domestication, Auto-D bacteria Thiobacillus was the dominant bacteria in both systems occupying > 20 % and the abundance would increase gradually from ∼ 20 % to ∼ 60 % as H2S content rising. A hidden positive correlation between DAMO bacteria Candidatus Methylomirabilis and Thiobacillus was uncovered by mantel test analysis here, implying the integrating of Auto-D and DAMO process for nitrate removal. This study not only provides insights into the novel SDMO technology but also offers a potential advancement in simultaneously low-carbon wastewater treatment and biogas utilization.
Read full abstract