Advanced Process Control

The road from theory to practice takes many steps, often over or around considerable obstacles. This certainly holds for the process industries, where actual conditions are rather unfavourable for control complexity and sophistication: process behaviour varies significantly and sometimes even qualitatively due to feed and product changes, unavailability of process equipment, lack of manpower for model building and updating, while operator training and motivation often receives insufficient attention. Still it is worthwhile to explore opportunities for applying modern control theory. Here a necessary step is to present methodsin a heuristic way and to demonstrate how they can be numerically worked out for realistic examples. Harmon Ray has done just that for distributed parameter systems, which are quite common in the process industries. The core of his book is a survey of the present state-of-theart, to which he and his co-workers have contributed in many respects. The following topics are covered: processes characterized by ordinary, parabolic and hyperbolic partial differential equations, and as a special case, pure time delays; optimal deterministic control; linear and nonlinear feedback control; deterministic and stochastic state estimation; optimal stochastic control; multivariable Smith predictor; choice of measuring points for distributed parameter processes. All methods are illustrated by typical processes, such as heat exchangers, steel slab (re)heating, packed-bed and stirred chemical reactors, distillation columns. In most cases, a good compromise has been reached between didactic simplicity and realistic complexity. The book gives little direct advice to engineers faced with the implementation of advanced control algorithms in industry. The latter requires insight into the sensitivity to changes in process behaviour, the effects of instrumentation failures, recipes for tuning parameters, and guidance for cost/benefit analysis. However, these shortcomings pertain rather generally to the literature as it is now. In the introduction no distinction is made between measured input disturbances and unmeasured ones. This distinction is important for state estimation and for feedforward control (which is not discussed either). The second chapter deals with on-line data acquisition and computer control, which (inevitably) has already been outpaced by the microcomputer revolution in process instrumentation. Too little is told about the latter, and too much about laboratory instrumentation. Some examples (3.2.1. on page 43; 3.2.8. on page 62) are