Computer aided process control systems synthesis using rule-based programming

Current steady-state process simulators have greatly increased the speed and efficiency of the development of Process Flow Diagrams. Chemical Engineers would benefit in the same way from a Computer Aided Design package to assist with generating completed Piping and Instrument Diagrams. Despite the many theoretical methods available in the control science area there is no single and complete available solution to the problem of synthesising control systems for whole chemical processes and therefore no concrete basis from which to develop a computer program. Design activities rely on a significant experience factor and this element has largely been ignored especially in control systems synthesis. The recent emergence of rule-based programming allows this "experience" dimension to be added to software. Although there is previous work in the literature on expert systems for distillation column control systems synthesis there is very little published on programs for other unit operations or the whole plant problem. In this project the problem of how to set up an expert system for whole plant control systems synthesis was addressed. As a preliminary step this required that expert systems for control systems synthesis for unit operations be written. The necessary knowledge to do this for distillation columns, heat exchangers and reactors was sourced from the literature and programs developed for each using a shell written in a version of Prolog. These programs were coordinated to work together and provide controllable solutions to whole process control problems using a matrix representation of the relationship between control objectives and manipulated variables developed in structural controllability analysis. This provided the framework for a prototype whole plant program. The operation of all the programs is illustrated using typical examples and their rule bases included in appendices to the thesis. The work demonstrated that, with more extensive rulebases than it was possible to develop in the time available for this project including access to theoretical methods when required, expert systems could provide a useful solution to both unit operation and whole plant control systems synthesis problems