Multicriteria Evaluation of Lot Sizing Techniques as a Function of Demand Pattern, Time Between Orders, and Demand Variability.

The interactive effects of demand pattern classified according to forecasting difficulty, cost structure represented by time between orders, and demand variability on the performance of five lot sizing techniques Eisenhut Part Period Balance, EOQ, Silver-Meal, Tsado, and Wagner-Whitin heuristic were evaluated using empirical demand data from local businesses. Three performance criteria were used inventory cost relative to the Wagner- Whitin optimal, number of stockouts, and percentage short per stockout. The results indicate that selecting the best algorithm depends on the performance criterion selected. EOQ, Tsado, and Silver-Meal performed best with respect to relative inventory cost, while Eisenhut and Wagner-Whitin heuristic performed best with respect to the shortage criteria. Percentage short per stockout was affected by differences in forecasting difficulty. The effect was constant across all techniques. Significant differences among all three classes of demand variability were found for all three criteria. Level of forecasting difficulty and variability do not affect the optimal choice of lot sizing technique if focus forecasting is used.