Proceedings

In this paper, the use of batch-wise multivariable monitoring and control techniques is considered for
the rapid sand filtration and dead-end ultra filtration processes used in the water industry. The
application of these modelling techniques is relatively new to this industry. In the first instance, rapid
sand filters for treating surface water in a fresh water treatment plant were considered. It was found
that the use of these modelling techniques can accurately predict the headloss trajectory of the filter in
the face of varying filter flow rates, and raw water quality. This information is then used in a
scheduling algorithm for optimising the run length of multiple filters.

This paper also looks at the possibilities for applying similar approaches to dead-end ultra-filtration
processes, where the batch cycles can be much more frequent, and can affect the stability of the
process to a much greater degree.