Proceeding Articles

I should like to amplify the title of this session . The papers we shall hear deal with the principles of the connection between measurement on the one  hand and computer control on the other and the assumption that the measuring  subsystems are accurate and reliable . The programme committee realises  the importance of measurement per se, but has considered the subject too  large to be included in this symposium.

It is often the case that the output of a measuring subsystem is a fairly strong function of variables other than the one that the system is intended to measure, a fact that should be kept in mind.

This paper reviews some of the problems in the interpretation of measurements of papermaking process variables . It is directed to papermakers or scientists who are not specialists in this field. The viewpoint is that of a statistician.

The main problem discussed is the analysis of continuous signals from process instruments, in so far as they are affected by the papermaking system . Specific hardware problems are not considered. Sampling problems for control are also discussed.

Sampled data can arise in several ways-for example, from manual samples taken from the process at reel change, from discontinuous instruments such as scanning basis weight gauges, also from digital computer control. Sampled data in contrast with continuous data have the problem of how sampling  should be carried out. This is discussed in relation to the spectral characteristics of the variable. The superiority of sampled data control for dead time processes and the relation between sampled data control and statistical quality control are mentioned.

Sampled data control system analysis and synthesis is introduced. The z-transform and modified z-transform are included.

Direct digital control is introduced, using sampled data forms as well as PID analog controller replacement. The relative features of pulse amplitude, pulse width and of velocity and positional algorithms are reviewed, also the selection of the sampling interval. A few of the applications to pulp and paper are reviewed.

Should paper be tested in-line or off-line? No one will argue that in-line continuous measuring of paper quality is the ultimate solution . Much progress has also been made in this direction, but we are still not there. When we do get there, we shall have to design systems for sampling, treatment and use of the in-line information before we can discard routine, off-line paper testing.

In the meantime, paper testing goes on in the routine control laboratories much as it always has. This, of course, is not necessary. When we look at routine paper testing, we ought to distinguish between the limitations inherent in the process and those that are there because of our neglect. If we do, we shall find that there is a lot more to off-line paper testing than is generally believed. If we feel a need to justify economically such a reappraisal of established methods, we can examine the cost of present paper testing. A survey in Sweden a couple of years ago showed that the operating cost of the control departments (wages, social security, testing implements) amounted to 5-6 dollars per 1 000 dollars of product. Considering that this activity has felt little impact of modern technology, it would not be unreasonable to assume substantial economic gains from a more up-to-date technique.

Yesterday, we were grilled by the mathematics of the control engineers; today, it is the turn of the process engineers at least, to begin with. They will take you right out into the bush, then bring you back to the machine floor, for the session deals with growing plant to pulp.

As papermakers, we all know that it is important for the pulp to arrive at the papermill in the right quantity and quality at the right time. It is equally important for the pulpmill to receive the wood, thus the integrated system starts in the forest.

We will take the discussion in two parts one dealing with the wood system, the chemical pulping system and the interface between wood and pulp, the second dealing with computer applications to chemical and mechanical pulping and the interface between pulping and papermaking.

Operations involved in the delivery of pulpwood from the standing tree to the consuming pulpmill are considered as a subsystem of the whole papermaking system . Pulpwood may be produced by any one of four different systems, namely, shortwood, tree length, full tree and remote chipping. Control in woodlands operations is management control and is primarily a function of administration. The principle is identical to that in process control. Control is not accounting in the conventional sense, but is involved with improved planning and the conservation of resources.

Co-ordination between the production of pulpwood and the requirements of the consuming mill is essential to ensure continuity of mill operation, at the same time ensuring minimum pulpwood inventories . The pulp and paper industry, for the most part, carries its inventory in raw material rather than in the finished product. Optimum co-ordination of pulpwood deliveries is rendered difficult owing to seasonal variation in mill consumption along with seasonal constraints on pulpwood deliveries attributable principally to climatic conditions. Proper control at this point in operations can result in great savings in pulpwood inventories . With pulpwood constituting 40-50 percent of the cost of the final product and with the absolute cost of pulpwood at an all-time high value, control of pulpwood production and delivery is critical at this time.

The paper deals mainly with the control of α kraft mill for either unbleached or bleached pulp, with special emphasis on control of the quality and quantity of the pulp produced. Differences in control strategies and objectives for market pulp mills and pulp mills in integrated systems are elucidated.

In an integrated pulp and papermill, all subprocesses are interrelated. This means that they cannot always be run at the optimum production level, since the result as a whole is decisive. The production managers have for many years successfully carried out the difficult scheduling of an entire mill. With increasing complexity of a modern mill, it is desirable to help production managers more systematically to utilise existing storage capacities. Thus, a production control system has been developed for the Gruvön mill. The mathematical formulation and solution of the scheduling problem is based upon optimum control theory.

In order to perform unavoidable production changes with the minimum of disturbances, the process control must not only produce uniform quality during steady state conditions, but whenever possible carry out production changes without introducing disturbances in the quality of the product.

In order to implement computerised production and process control systems, good human relations must exist between computer staff and production personnel. This, together with good technical solutions, including the man/machine interface, will guarantee success.

At a newsprint mill, the running of the groundwood min is connected with the running of the papermill in several ways by the pulp and whitewater systems and by the supply of electric power. For controlling the groundwood mill in a way that a high and uniform pulp quality is maintained and an optimum use of machinery and resources is made, the groundwood mill crew must continuously supervise this complex system . This is beyond human ability with the limited implements that the crew often has at its disposal . Studies performed at the Hallstavik newsprint mill have shown the limits of human ability in this respect.

In order to give the crew an improved implement, an on-line process computer has therefore been installed in the groundwood mill at Hallsta papermill. The objects of the computer system are planning the production of groundwood pulp and consumption of electric power for a coming weekly period, supervising and controlling the pulp and whitewater system for the entire mill and, finally, reporting.

Owing to these points, the main intention was to reach a more steady running condition of the groundwood mill and improved utilisation of the contract of purchased electric power.

The computer installation, including programs, was completed in autumn 1968, after which test runs with the process were started.

The hierarchical structure of pulpmill control is defined in terms of six levels-planning, scheduling, supervisory, mufti-variable, direct and data acquisition. The functions of each of these levels for a continuous digester are discussed. The design of the scheduling level is investigated for the case of a single pulpmill that must produce two species sequentially. It is shown to be primarily the problem of designing a feedforward control system for a process with a major dead time.

The relationship of plant design to controllability is discussed, using the example of determining the size of the vuln storage tanks.