Abstract
The paper considers a firm manufacturing several thousand different products, each requiring manufacturing time on several of its available production facilities. Production is scheduled according to a week-by-week rota, arranged so as to generate even average work loads on each production facility. Fluctuations in sales demands, however, lead to a discrepancy between the stock requirements in any section of the rota and the production capacities available. Consequently, to avoid the rota becoming disorganized, some form of production control must be applied. Since the production capacities are largely fixed, this must consist of adjusting the stock requirements. A method of computing adjusted requirements which minimizes a weighted sum of squares of the adjustments made is described. The method is based on Wolfe's technique for quadratic programming, modified to make it computationally feasible in the situation considered which involves several thousand non-negative and usually non-zero variables.