Production Switching Heuristics for the Aggregate Planning Problem

Abstract

A number of approaches to the aggregate planning problem have been proposed in the literature, yet experience suggests that industrial concerns seldom use these models in actual planning situations. This paper describes a modified random walk production-inventory heuristic for the problem which should appeal to managers on the basis of simplicity as well as efficiency. The proposed approach is contrasted with linear programming, parametric programming, and linear decision rule optimal and near optimal solutions for several well known production situations. The simple production switching heuristic produces schedules which exceed optimal schedules by only 1 to 2 percent of total production costs in all cases.planning, production/scheduling, inventory/production: production smoothing

The production switching heuristic: A practical revision

“…However, only a few of these (Bowman 1956, Bowman 1963, Goodman 1974,Hanssman and Hess 1960, Holt et al 1955,Jones 1967, Meij 1980, Mellichamp and Love 1978, Taubert 1968) are identified by their originality in approach. The rest, which differ in their orientation, scope and methodology, are primarily extensions of these models to incorporate finer aspects of the production system.…”

“…To achieve this the classical paint factory problem (Holt et al 1955) is solved again by the proposed model using the same cost data and the percentage deviation in total cost has been used as a measure of the effectiveness of the new model. Despite their methodological elegance, these models have not had a significant impact in the business environment (Lee and Khumawala 1974, Lee and Moore 1974,Mellichamp and Love 1978,Taubert 1968) due to their complexity, rigid assumptions, and unpracticable solutions.…”

“…The Production Switching Heuristic (Mellichamp and Love 1978) was developed 1864 S. Barman and E. E. Burch workforce sizes too frequently over the planning horizon. Mellichamp and Love (1978) use a simple modification to the random walk model (Orr 1962) to restrict the production and workforce sizes to three discrete levels.…”

“…Mellichamp and Love (1978) use a simple modification to the random walk model (Orr 1962) to restrict the production and workforce sizes to three discrete levels. Specifically, for an aggregate production planning problem, the solution of the model yields only three different production and corresponding workforce sizes; high, normal, and low, for all periods in the planning horizon.…”

“…On the other hand, as the number of levels increases, the frequency of switching and, therefore, the complexity of the production system also increases. The very advantage of a switching heuristic, less frequent rescheduling of production and workforce, is lost gradually as more levels are added (Eilon 1975, Mellichamp andLove 1978). It would appear worthwhile to investigate the cost penalty for choosing a two-level PSH over a three-level one, since the cost disadvantage may very easily be outweighed by the gain in ease of scheduling as well as reduced computation time.…”