Cutting paper in Richards Bay: Dynamic local or global optimization in the trim problem

Abstract

The cutting stock or trim problem arises when jumbo rolls of paper are slit into reels of various width ("deckled"). The problem was examined as early as 1954; it is now seen as a classical Linear Programming formulation. The problem may be easy to formulate, but it is difficult to solve for genuine data. This is so not only because of the computations required to find good cutting plans. The major problem is this: such plans are based on certain data, but the data change as the plan is being executed and the plan no longer applies. So here it may not be optimal to optimise globally, but to work on a local heuristic basis. We can propose a heuristic procedure that compares well with an absolute solution and that can be used on a local basis. This heuristic is in use at Mondi Paper in Richards Bay.