Proceeding Articles

The design of more responsive control systems for papermachine dryer sections requires a knowledge of the coupling between web moisture content and changes in basis weight and machine speed. The interactive effect of changes in basis weight and machine speed on the measured moisture content must be known to allow proper compensation.

An estimate of the steady state coupling coefficients was made by forming an approximate mathematical model and calculating the steady state response surfaces for average web moisture content versus machine speed and basis weight. The model was a realistic representation of the partial differential equations relating heat conduction and vapour diffusion within the web. Approximate, constant values of web properties and transport coefficients and the alternating boundary conditions characteristic of conventional cylinder drying were used . Basis weight of 26-90 lb/1 000 ft2 and speeds of 800-1800 ft/min were studied.

The calculated coefficients showed that, for typical production conditions of 421b board at 1 400 ft/min, a 1 lb change in basis weight would be expected to produce a 2 percent change in web moisture content; a speed change of 100 ft/min would produce a 5 percent change in moisture content. The dependence of the coupling coefficients on the level of basis weight and speed was also demonstrated. The coupling coefficients for basis weight and for speed both increased approximately linearly with the operating speed, whereas the increases in both coefficients through basis weight changes were less significant. These results are of use in the design of non-interactive control systems for papermachine dryer sections.

Although the calculated results were restricted to the steady state gains for this system, an indication is given of the possible use of a more complex model in studying the dynamic behaviour of the drying system.

Control systems, whereby a final customer may obtain a satisfactory product after it has passed through several stages of manufacturing, are characterised by considerable time delays in both feedforward and feedback paths. These systems are examples of the industrial dynamics problems investigated by Forrester and others. The matter under discussion in this paper is concerned with the establishment of sensible control systems and the maintenance of their reliability and accuracy. It specifically considers paper products used in data processing as punched cards and forms for optical character recognition.

Models of the behaviour of national economies have been developed and popularised for some time. The economic behaviour of whole industries has also been examined and statistical analysis has uncovered at least the main features of the working of a number of industries. There is no difficulty in principle in carrying this process of analysis on to the examination of individual products . The paper  discusses these topics and illustrations are provided of national and industrial economic models and the behaviour of several product groups in the U.K. paper industry are examined. The results show that statistical analysis can reveal important economic influences at work, but that it is necessary to beware of special factors, technical, political and social, which can be of overriding importance for particular products at particular times.

Control of a complex organisation requires that its response to decision inputs be predictable . Traditionally, this has been made possible by almost breaking down the organisation into very simply coupled subsystems. This allows decision-makers to use a relatively simple and usually informal model.

Size and environmental rate of change are now tending to make the decoupling hard to make effective. Industrial dynamics provides both a theoretical structure and a computer simulation technique designed to provide decision-makers with a richer model of the dynamic feedback processes generally involved in industrial operations. The structure defines a system in terms of a closed boundary containing feedback loops, which in turn are formed by the relation between level and rate variables . The DYNAMO compiler allows models of systems conceived on these lines to be programmed in a straightforward way so as to allow experimental investigation of the system variables. The paper discusses case studies and concludes with notes on model validity and the problem of data.

Up to a few years ago, all industrial processes were either batch processes or were operated on a steady state basis, although some (like papermills, automatic looms and steel-rolling mills) operated in steady state only for a limited period until the specification or grade of the output had to be changed. Thus, industrial processes can be divided into three classes:

1. Batch processes.
2. Continuous processes.
3. Quasi-batch processes.

History and philosophy of control systems

The past decade has witnessed a major advance in the use of process control systems by the paper industry. Before that time, very few variables were controlled automatically, many instruments were either indicating or else, if controlling, were operated in the manual mode much of the time and the degree of instrument reliability and accuracy in the average mill did not promote operator confidence or use . Pneumatic controllers performing simple control loop strategies were the norm. Making paper was an art and frequently different papermakers would operate a machine in quite different ways to make the same given grade of paper. Fairly broad tolerances in paper specifications were accepted in the trade, since few machines could achieve or maintain very stable operations.

There is no difference, at least in principle, between control of administrative systems and control of other systems. The purpose of a control procedure is to specify the way in which the behaviour of a system can be affected. The controller of the system scrutinises information about the performance

of the system and every so often he has to make a decision, namely, to choose between several courses of action open to him. This decision is transmitted to the system with the expectation that it will react in a certain way. This cycle of events monitoring and evaluation of the system’s behaviour, followed by a decision for corrective action is the essence of the control process, irrespective of whether it is control of inanimate systems or managerial control of industrial enterprises.