The techniques for economic assessment that are described in this chapter have each been found useful in particular industrial situations. Attention is concentrated on quantifying the benefits of control. The low cost of microprocessor boards does not, in general, mean that low-cost process control can be achieved, since sensors, cabling, displays, software and engineering effort all need to be taken into account. The performance of a process depends on its design, on operating practices and on the level of control that can be achieved. The three areas interact strongly in their effect on performance. An economic assessment of a process will often show that improvements are called for in all three areas (process design, operating practice, control system) if maximum benefit is to be obtained from a particular level of investment.

This chapter is concerned with the application of traditional 'off-the-shelf controllers. Even complex industrial processes often can be controlled adequately by such controllers whose attraction lies in their ready availability, ease of installation and simple maintenance. Controller commissioning consists in allocating numerical values to a few parameters. A number of different methods, starting with the purely empirical, are put forward for selecting the parameters.

This chapter is concerned with single-input single-output processes and the solution of their control problems by relatively straightforward methods. Much of the ma terial can be found in classical control texts but supplementary material based on experience has been added. Most of the chapter deals with process understanding, since control design for well-understood processes is relatively straightforward.

In developing a computer control scheme for a large process with many loops, the main problem confronting the system designer is not that of control algorithms. More important in general are the problems of establishing a reliable database to include a wide variety of information, not all of it emanating from sensors, an efficient means of man/machine interaction and the incorporation of reliability/self diagnosis facilities. Heavy interaction is likely to interconnect groups of up to five or six individual loops and algorithm design for these groups is fairly easily achieved using some of the methods to be outlined in this chapter. A physically-based simulation in which the model parameters correspond with actual process parameters is a great help in deciding on the best configuration for the control system and investigating details that defy analytic solution.

A few years ago, control applications were largely restricted to the traditional areas: Aerospace, Chemicals, Petroleum, Power, Metals and the like. The range of applications has recently increased enormously to embrace such diverse areas as: control systems for car engines, systems for automatic unloading of oil tankers, energy management systems for large buildings, sophisticated mechanical handling systems and automatic exposure and focussing for cameras. The newer application areas have not, from the point of view of control theory, posed particularly novel or illuminating problems. Accordingly, the applications to be described in this chapter are largely from the traditional areas.

Robots are becoming increasingly important across a wide range of applications. An extensive literature exists covering for instance: industrial potential, particular applications, detailed design configurations and topics such as robot vision. Although some of these topics may be touched upon in the following treatment, we shall in general avoid detail in favour of general principles. Our aim then is to introduce the control engineering point of view. Seen in this way, a robotics problem is simply just another electromechanical (or hydraulic or pneumatic) problem that might come the way of a control system's designer, paralleling (for instance) the problem of designing a ship's-mast-mounted radar antenna with flexible mast, out-of-balance rotating forces and non-orthogonal control axes.