THEORETICAL AND PRACTICAL IDENTIFIABILITY OF A REDUCED-ORDER MODEL INAN ACTIVATED-SLUDGE PROCESS DOING NITRIFICATION AND DENITRIFICATION

This paper deals with the structural identifiability and the identific ation of the parameters of a reduced order model used for control of a single reactor activated sludge process doing nitrification and denit rification. This reduced order model is splitted into two submodels, o ne 3-dimensional state submodel in aerobic conditions and one 2-dimens ional state submodel in anoxic conditions. The identifiability analysi s is based on on-line oxygen and nitrate concentrations data. It has b een shown that the reduced-order model is structurally identifiable. T he parameter identification has been carried out by using the simplex method of Nelder and Mead. Simulation results performed over a range o f six hours (two aerobic/anoxic cycles), show that there exists a good fit between the simulated solution and the actual behavior of a lab s cale pilot plant. (C) 1998 Published by Elsevier Science Ltd. All righ ts reserved.