Advantages of Integrated and Long Term Aluminum Recycling Batch Planning in a Constrained Secondary Material Market

"Increased demand resulting from the economic and environmental advantages of recycling secondary materials has constrained secondary material supply. As a result of constrained supply, the conventional strategy to maximize recycled content in an individual aluminum alloy must be expanded to maximizing total recycled content in a given production period. This work compares the performance over a two week period of an actual remelter for two strategies; a sequential charge-by-charge optimization framework and an integrated approach that explicitly incorporates the production schedule. Optimizing across the entire two week production is found to increase recycled content and minimize primary and alloying material additions. It is expected that as the scrap material market continues to be constrained by high demand, longer term batch planning strategies that can optimize material allocation will deliver improved performance.IntroductionIntensifying economic and environmental pressure to incorporate secondary aluminum into products is pushing remelters to explore additional strategies and tools to increase recycled content. Aluminum remelters are exploring alternative sources of secondary materials because of constrained secondary material supply and relatively high primary and secondary material cost [1]. Computational models, such as blending models are a tool that could help industrial remelters mitigate supply and cost challenges and thereby increase recycled content. In particular, computational tools can help manage the incorporation of new raw materials of variable quality. Blending models find many industrial applications and have been explored for over sixty years originating with the classic nut-mix problem dating back to 1950’s [2]. Blending models optimize the raw material input to make a desired product subject to a series of constraints such as product specifications and material availability. Deterministic and linear blending model formulations are presently implemented in some industrial aluminum remelters [3]. The primary reluctance of aluminum remelters against computational model implementation is the incapability of their blending models to reflect the intricacy of industrial remelting conditions [4]. For example, to reduce calculation time, industrial aluminum blending models usually optimize a batch plan for a single alloy and are not formulated to optimize several products simultaneously. Since such a model does not include knowledge of subsequent charges in the optimized batch plan, the resulting batch plans tend to deplete the “cleanest” secondary materials leaving the more uncertain and compositionally challenging materials for later charges. As a result, a remelter may override the calculated batch plan or leave certain scrap materials out of the material input to conserve these materials for alternative products scheduled for later production. This work explores an alternative aluminum charge planning tool that can optimize across an extended production schedule and compares it to a heuristic sequential optimization strategy reflecting an industrial remelter."