Modeling, Experimentation, and Control of Autotrophic Nitrogen Removal in Granular Sludge Systems

摘要

Complete autotrophic nitrogen removal (CANR) is a novel process that can increase the treatment capacity for wastewaters containing high concentrations of nitrogen and low organic carbon to nitrogen ratios, through an increase of the volumetric removal rate by approximately five times. This process is convenient for treating anaerobic digester liquor, landfill leachate, or special industrial wastewaters, because costs related to the need for aeration and carbon addition are lowered by 60% and 100%, respectively, compared to conventional nitrification denitrification treatment. Energy and capital costs can further be reduced by intensifying the process and performing it in a single reactor, where all processes take place simultaneously, e.g. in a granular sludge reactor, which was studied in this project. This process intensification means on the other hand an increased complexity from an operation and control perspective, due to the smaller number of actuators available.In this work, an integrated modeling and experimental approach was used to improve the understanding of the process, and subsequently use this understanding to design novel control strategies, providing alternatives to the current ones available. First, simulation studies showed that the best removal efficiency was almost linearly dependent on the volumetric oxygen to nitrogen loading ratio. This finding among others, along with experimental results from start-up of lab-scale reactors, served as the basis for development of three single-loop control strategies, having oxygen supply as the actuator and removal efficiency as the controlled variable. These were investigated through simulations of an experimentally calibrated and validated model. A feedforward-feedback control strategy was found to be the most versatile towards the disturbances at the expense of slightly slower dynamic responses and additional complexity of the control structure. The functionality of this strategy was tested experimentally in a lab-scale reactor, where it showed the ability to reject disturbances in the incoming ammonium concentrations. However, during high ammonium loadings, when the capacity of the present sludge was reached, an oscillatory response was observed. Proper tuning of the controller is therefore of essential importance.In this thesis, it was demonstrated that proactive use of model simulations, in an integrated methodology with experimentation, resulted in improved process understanding and novel control ideas. This will contribute to moving this promising technology from a case-by-case ad hoc approach to a more systematic knowledge based approach.