Intertemporal stochastic sawmill planning: Modeling and managerial insights

• We study a sawmill planning problem which presents uncertainty in the log supply.
• We have different levels of information detail depending on the time span.
• Various two-stage stochastic models are presented that could simulate the problem.
• A Rolling Horizon scheme was used to test the models.
• The results show the behavior differences between the models and managerial insights.

Optimization models have long been used in the Forest Industry. Here, as well as in other areas, models are used in different time horizons to support planning and scheduling. Guaranteeing the consistency of the production policies in those different time periods is highly relevant for efficiency and demand fulfilment. This paper presents a set of Sawmill Planning Models that cover tactical planning, as well as operational planning. Aggregated planning decisions are modeled in order to determine the log supply for the sawmill. At the operational level, detailed weekly production plans are defined using the actual log supply, which might not be consistent with what was originally planned, due to several variabilities. We address the issue of coordinating short-term decisions with mid-term planning using a two-stage stochastic optimization formulation. Various models with certain variations are proposed in order to simulate all of the complexities that are present in the Sawmill Planning Problem. To test the models, we simulated a special Rolling Horizon method using different demand scenarios. Finally, we present results and managerial insights regarding the effects of uncertainty.