Meeting One-off Orders from Stock

Abstract

A CONSIDERABLE volume of literature has appeared in recent years on stock control models designed to deal with large numbers of identical parts. A problem in stock control also exists at the other end of the scale where each part has characteristics which make it almost unique. Such stock can arise in a variety of ways. A tailor-made piece may have the order cancelled, the tailoring may not have been perfect, some defects in restricted parts of the material may have been revealed during manufacture or the manufacture of this single piece as a surplus might have yielded a better balance of production. To explore the market with a single piece would probably not secure the best price since the practice savours of second-hand goods. Why not then examine all incoming orders against these unique stock items? This seems so obvious a procedure that it must immediately be explained that since the stock item is unique, the odds against a perfect match are so considerable that it is worth designing a drill, first of all to find to which incoming orders the stock could be applied in such a way as to satisfy the customer and secondly to ascertain which of these allocations give the best return to the manufacturing company both in respect of the best yield from the volume of the material and also in relation to the original price of the stock item. Translating the general problem into the particular, the authors are concerned with steel plates which can be ordered by the customer as single items in any dimensions up to 15 in. thick, 84 ft long and with a maximum weight of about 81 tons. There are also about 400 specifications against which the customer might order the quality of the steel. The possible combinations are thus enormous and some system of relaxation from perfect matches is essential if stock plates are to be sold at market price. The extremes of size sell at slightly more than the average: so when a large stock plate is available for some reason, the steel manufacturer is reluctant to cut it into two smaller pieces because even for the same total weight a somewhat lower selling price would be obtained. Likewise, in the matter of quality it is often possible to meet the customers' specification by supplying a higher grade of steel than ordered (and will be paid for), but again the steel manufacturer is reluctant to follow this procedure if it can be avoided. With this background information three main objectives in building the model can be defined. (1) To design an efficient search system; in practice, because of the complex technical considerations which need to be incorporated in the model, this amounts to discarding the unlikely plates as early as possible.