Control of an alternating aerobic-anoxic activated sludge system - Part 2: Optimization using a linearized model

Complex models, such as activated sludge model No. 1 (ASM1), have rarely been used in practice for process control or optimization. One major reason for this is the computational effort demanded by these models for both parameter estimation and simulation. Therefore, a linearized version of the ASM1 model is developed and applied to the control and optimization of a bench-scale alternating aerobic/anoxic activated sludge system. The model prediction was used to optimize the aeration time by manipulating t_c (total cycle time) and f_a (fraction of aeration cycle time) while meeting the permit requirement of the effluent ammonia concentration. The linear nature of the model facilitates its use for on-line calculations, and error feedback is used to counteract problems of model inaccuracy. The model was applied to two cases in which the influent compositions were either known currently or only the previous day's information was available. The average optimized f_a was found to be 0.30 for the first case and 0.37 for the second case, when the maximum effluent NH₄⁺-N concentration was set at 1 mg/l. The efficiency of nitrogen removal was 76% for both test cases since nitrate could almost completely be removed through adequate anoxic cycle ratios resulting from the f_a optimization.