OPTIMAL CONTROL OF ACTIVATED SLUDGE SYSTEM:
LABORATORY MODEL STUDY
A. B. Shah Alam, Graduate Student
D. I. Angelbeck, Assistant Professor
Department of Civil Engineering
The University of Toledo
Toledo, Ohio 43606
INTRODUCTION
Presently, the efficiency of the activated sludge waste treatment process has been
determined quite precisely under steady-state conditions.   However, steady-state conditions (i.e., constant waste character and hydraulic flow) have a limited value when compared to the transient character of most liquid waste streams.   In order to assess the
manner in which the biological waste treatment system operates in the field, the effect
of variable or transient waste character on the performance of biological treatment
systems must be precisely understood.
Previous studies describing the activated sludge process under transient loading have
utilized various unsteady-system models to simulate process performances [1-7].   If a
process is developed which accurately describes transient patterns of effluent quality
parameters under unsteady conditions, operational system control strategies can and
should be defined.   The use of these strategies can offer many benefits including:
(a) improved effluent quality, (b) improved monitoring and reliability, (c) process
stability, (d) reduced capital requirements, and (e) reduced operating and maintenance
costs.   The degree to which any of these benefits are realized is certainly dependent
upon the level of control applied.   Such control strategies would minimize the effluent
quality deviations due to unsteady process operation.   The principle control actions
currently utilized in the activated sludge process are sludge recycle rate, waste sludge
flow rate and aeration rate.   Other control variables which have been considered mainly
for laboratory and computer simulation are aerator retention time [8-11] and food-to-
microorganism ratio (12).
Several control systems such as "feed back" [6,13], "feed forward" [6] and
utilization of Pontryagin's optimal control theory [14] have been theoretically shown
to have potential for controlling the activated sludge process. The computer simulation
of the optimal control policies for an activated sludge process consisting of a complete-mix
reactor and a clarifier (Figure 1), using variational calculus, has been demonstrated by
the authors [1].   The simplified model included:   two state variables, aerator MLVSS
level and aerator substrate level; two control variables, dilution factor and sludge
wastage rate; and two disturbance variables, influent microorganism level and influent
substrate level.   It was demonstrated through computer simulation that a substantial
reduction in the effluent quality variation caused by an influent substrate disturbance
Figure I.  A complete-mix activated
sludge process.
713