Modeling denitrifying sulfide removal process using artificial neural networks.

Area of Science:

• Environmental biotechnology
• Bioreactor engineering
• Wastewater treatment

Background:

• The denitrifying sulfide removal (DSR) process involves intricate interactions between autotrophic and heterotrophic denitrifers, complicating mechanistic modeling and control.
• Developing accurate predictive models for DSR bioreactors is challenging due to these complex microbial dynamics.

Purpose of the Study:

• To present a novel artificial neural network (ANN) model for predicting the steady-state performance of an expanded granular sludge bed (EGSB)-DSR bioreactor.
• To evaluate the ANN's capability in modeling both nitrite denitrification and the complete DSR process.

Main Methods:

• Utilized artificial neural networks (ANNs) to infer complex relationships between input and output variables in the DSR process.
• Developed and applied a novel ANN model to predict the steady-state performance of an EGSB-DSR bioreactor.

Main Results:

• The proposed ANN accurately predicted the steady-state performance of the EGSB-DSR bioreactor.
• The ANN model identified a critical threshold for hydraulic retention time (HRT) below 7 hours.
• Influent sulfide concentration was shown to significantly affect reactor performance at HRT < 7 hours.

Conclusions:

• ANNs offer a viable approach to model complex biological wastewater treatment processes like DSR without detailed mechanistic characterization.
• The study highlights the critical influence of influent sulfide concentration and HRT on DSR bioreactor efficiency.