Sludge bulking is a problem widespread in the operation of activated s
ludge processes. Over the past fifteen years much research has been ap
plied to identifying the nature of the microbial species responsible f
or bulking and to developing an understanding of their population dyna
mics. While some isolated attempts have been made at developing a unit
process model capable of simulating the behaviour of bulking, none ha
ve been extensively evaluated against field data from full scale plant
s, This paper describes the development of a multiple-species model of
the activated sludge process (for both the reactor and settler), its
application to the assessment of various operational strategies for th
e control of bulking, and its simplification for incorporation into an
on-line estimation scheme using a Kalman filter. Routine operating da
ta from thee Davyhulme Wastewater Treatment Works in Manchester are us
ed for identification (calibration) of the model. Simulation studies o
f control strategies are based on the same Works. By using the Kalman
filter to reconstruct real-time estimates of the ''unmeasurable'' stat
es of the process model, the paper also explores the extent to which t
his filter-based control can bring improvements over similar control b
ased entirely upon conventionally measured operating variables.