Modelling nutrient removal in a sequencing batch reactor with respirometry

The main objectives of the present thesis can be summarized as: i) the development and validation of simplified mathematical models for activated sludge processes in an SBR treating real domestic wastewater; ii) the application of these simplified models for analysing the respirometric response and for obtaining information about the oxygen uptake for the different processes; iii) the application of the monitored respirometric values for model calibration and determination of parameter values, which are used to predict the processes in the next cycle; iv) the use of models as theoretical background for the development of control strategies for plug-flow systems and for SBR; v) relating the basic time scale for the models to the short term.The starting points for the model development and simplification were: i) the Activated Sludge Model No. 1, for carbon oxidation, nitrification and denitrification; and ii) the Activated Sludge Model No. 2, for biological phosphorus removal .In this study an SBR pilot plant was used and seen as a model for a plug-flow system. During the two and a half years of operation, the plant underwent three different technological phases. The first phase began with the removal of organics and nitrification. Denitrification was incorporated in the second phase. The last phase included biological phosphorus removal.In the first phase, two simplified mathematical activated sludge models are presented. The first model gives the response of the respiration rate in an SBR with nitrification, the oxidation of readily biodegradable matter, and endogenous respiration during one cycle. This model is used to predict the respiration rate during a complete SBR cycle. For this, it uses parameter values calibrated during the previous cycle, some default values and information about the ammonia concentration in the influent. The endogenous respiration rate is described with an exponential equation. The second model is used to predict the changes in nitrification capacity after a change in the loading rate and/or sludge wastage rate. For model calibration and validation, an SBR pilot plant receiving domestic wastewater was operated for nine weeks.In the second phase, a mathematical model is presented for the behaviour of the respiration rate and nitrate removal in an SBR with nitrification, denitrification and carbon oxidation. This model is based on the response of the respiration rate measured during nitrification and carbon oxidation and the nitrate removal rate during the post-denitrification period. For model calibration and validation, an SBR pilot plant receiving domestic wastewater was operated for three months. The respiration rate was used to calibrate several parameters of the model.In the third phase, a mathematical model for an activated sludge SBR with nitrification, denitrification, carbon oxidation and phosphate removal is presented. This model is based on the response of the respiration rate measured during nitrification, carbon oxidation and phosphate removal, together with the behaviour of phosphate and acetate as proposed in the Activated Sludge Model No. 2. For model calibration and validation, an SBR pilot plant receiving settled domestic wastewater plus acetic acid solution was operated for five months.In all the three phases the model for the respiration rate ( r ) in an SBR during one cycle, including nitrification, oxidation of readily biodegradable matter, endogenous respiration and a fraction for the respiration rate for phosphorus uptake, gives a good simulation of the measured respiration rate. A good prediction of the total oxygen consumption and distribution during one cycle is found from a simulation, using parameters calculated from the previous cycle together with the variables from the influent. Therefor this model can be used in control strategies as long as it is used for a short time- scale. During long-term operation, parameter variation is significant and too complex to be predicted. In the particular case of nitrification capacity variation in an SBR during long-term operation, the model can explain the variation trend but cannot explain the abrupt changes.Simplified mathematical models for the activated sludge process on the bases of the respiration rate are validated. On short-time scale, the models give a good response prediction of the activated sludge process feed with wastewater. The models are good tools for control strategies, however periodical parameter calibration is needed.