Monte Carlo Methods in Heavy Industry

Abstract

While the expanding application of operational research demonstrates that executive boards are aware of its effectiveness as an aid to scientific management, the writers wish to direct attention to the merits of one aspect of operational research whereby accurate forecasts can be made offuture conditions in batch processes in which the duration of certain operational stages may vary with deleterious effects on the overall efficiency. The same method may be used to predict the benefit arising from a suggested remedy. Before authorizing capital expenditure on extension and modification of equipment, or before making drastic changes in the existing practice of an established shop, the discriminating manager needs to estimate as exactly as possible the full implications of the proposed changes. This may present little difficulty when the process is under the precise control of the management and no operational problems arise which cannot be solved by successive application of the known logistics of the process. With certain processes, however, the mantle of prophet is less comfortable. If, for instance, the raw materials are subject to fluctuations outside the control of the management, a variable is immediately introduced which makes accurate forecasting more difficult. If, in addition, the process requires physical conditions such as a high temperature close to the working limit of the containing vessel, uncontrolled erosion and corrosion can lead to undesirable reactions detrimental to the process, and forecasting becomes even more involved. If, to these, the limitations of the mechanical auxiliary appliances and of any subsidiary processes are added, the problem should be worthy of a scientifically obtained solution. Undoubtedly, inspired guesswork has on occasions been brilliantly successful; its many failures have ranged from the forgotten to the disastrous. What can be done to give a scientific forecast? A statistical picture or model can be built to describe existing conditions and, from this, the general principles governing the time characteristics of the processes may be extracted. The effects of proposed changes