Modelling and control studies of the activated sludge process at Norwich Sewage Works

Abstract

In general, studies of the control of the activated sludge process have focused on the design of single-loop controllers able to suppress higher-frequency disturbances and provide minimum-variance performance about a precisely-defined, constant set-point. This paper addresses some of the complementary problems of control. It considers the slower dynamical components of activated sludge process behaviour, the detection of and response to conditions of process failure, control within imprecisely-defined 'domains' of performance, and (by implication) the concept of switching performance from one such domain to another. For these purposes the paper discusses the construction of a relatively complex microbiological model for the activated sludge process. The results of using this model in simulation studies of a fuzzy controller are then presented. These preliminary results are encouraging and suggest that, in spite of the substantial current limitations on process knowledge, there are many novel aspects of activated sludge process control yet to be explored.