Analysis and comparison of single period single level and bilevel programming representations of a pre-existing timberlands supply chain with a new biorefinery facility

Abstract

This research investigates the economic impact of a new biorefinery on an established timberlands system. This paper proposes that a single level model does not adequately represent the interactions in the supply chain. The timberlands system has two major decision makers: harvesters and manufacturers. Bilevel models are two-level turn-based models with each level representing a decision maker. The two levels are interconnected, but neither has control over the decisions of the other. The leader makes an initial decision, and the follower reacts; the leader can anticipate the follower's reaction. To demonstrate the value of this more complex representation, a single period bilevel and a single period single level model were formulated for a representative case study. The separation of objective functions in the bilevel problem yielded decisions different from those of the single level formulation. It was concluded that the bilevel model may more accurately represent the real system.