Artificial intelligence control of a sequencing batch reactor for nitrogen removal via nitrite from landfill leachate

Leachate generated in old landfills is a high-strength wastewater, which is particularly difficult to treat owing to its low biochemical oxygen demand/total Kjeldahl nitrogen ratio. This paper seeks to demonstrate that reliable leachate treatment by means of sequencing batch reactors (SBRs) is indeed possible by means of the application of a smart control system. This study assesses the results of a computer-controlled bench-scale SBR treating raw sanitary landfill leachate to achieve nitrogen removal through the nitrite shortcut. Significant improvements have been obtained by introducing a fuzzy inferential system based on simple process measurements (i.e. dissolved oxygen, oxidation-reduction potential and pH). The paper analyzes the results of a test period of over 280 consecutive days of operation, during which the fuzzy control system correctly recognized over 97% of the SBR phase transitions and provided smart adjustments of the process operating conditions in terms of phase length and external COD addition. In spite of time-varying process conditions, the application of fuzzy logic provided stable nitrogen removal via nitrite through continuous adjustments of the main process parameters and resulted in a decreased hydraulic retention time, an increased loading rate, a saving in the external COD addition and considerable aeration energy conservation.