MODEL CALIBRATION FOR THE HIGH-PURITY OXYGEN ACTIVATED-SLUDGE PROCESS- ALGORITHM DEVELOPMENT AND EVALUATION

A model calibration algorithm is developed for the high-purity oxygen activated sludge process (HPO-ASP). The algorithm is evaluated under d ifferent conditions to determine the effect of the following factors o n the performance of the algorithm: data quality, number of observatio ns, and number of parameters to be estimated. The process model used i n this investigation is the first HPO-ASP model based upon the IAWQ (f ormerly IAWPRC) Activated Sludge Model No. 1. The objective function i s formulated as a relative least-squares function and the non-linear, constrained minimization problem is solved by the Complex method. The stoichiometric and kinetic coefficients of the IAWQ activated sludge m odel are the parameters focused on in this investigation. Observations used are generated numerically but are made close to the observations from a full-scale high-purity oxygen treatment plant. The calibration algorithm is capable of correctly estimating model parameters even if the observations are severely noise-corrupted. The accuracy of estima tion deteriorates gradually with the increase of observation errors. T he accuracy of calibration improves when the number of observations (n ) increases, but the improvement becomes insignificant when n>96. It i s also found that there exists an optimal number of parameters that ca n be rigorously estimated from a given set of information/data. A sens itivity analysis is conducted to determine what parameters to estimate and to evaluate the potential benefits resulted from collecting addit ional measurements.