Optimising and Predicting Performance of Industrial Filtrations using Process Data

Industrial cake filtration is non-trivial from an operational point of view. Discrete events such as the removal of filter cake occur on a frequent but irregular basis. These events tend to upset the steady state of the incorporating line, which may constrain plantwide optimisation. A case study has been carried out with an industrial partner where changes in the biological feedstock act as strong disturbances on a series of manually reinitialised dead-end pressure leaf filters.

This renders production planning a challenging task which,so far, is carried out by experienced operators. We look for shortcomings in the current, heuristically grown manner of operating the filters and present guidelines for a superior strategy. A predictive process model is required for a deterministic scheduling algorithm, and two approaches at modelling the filtrations are presented and compared.