Modular procedural control of chemical processes

Procedural control theory (PCT) encompasses chemical process modelling, specification, controller synthesis and analysis. In particular, techniques have been presented for designing sequential controllers for chemical systems which are guaranteed to generate a behaviour which conforms to a functional specification and terminates in a desired process state (non blocking). Until now, the techniques within PCT have been limited to simple chemical systems modelled by no more than a few thousand states and for which the specifications are relatively straightforward. The aim of this work is to apply the same techniques to a broader category of control problems in chemical batch processing, for which the size of the state models may be many orders of magnitude greater, and the specifications quite complex. In this work, size complexity is addressed using reduced models of the process which are easily calculable within the modular approach for process modelling in PCT. A complementary method for handling specification complexity is addressed by borrowing from AI the concept of subgoals. Both decomposition techniques serve to modularise the controller synthesis problem.