Dynamic benchmarks for operations evaluation using soft computing methods

New dynamic benchmarks are developed as performance metrics for operational activities in the ingot mill of an aluminium smelter. This new type of high-level performance measure enables both a systematic and a holistic appraisal of operations performance that is not possible at present. The benchmarks also allow changes in performance of each distinct resource group to be clearly identified. The method for computing such measures is based on a wide range of shop-floor data, and possibility theory is used to model the soft characteristics of much of the input. Fuzzy performance estimates are dynamically updated with special attention to combining the different forms of uncertainty that exist in the input information. Although these procedures are site specific, the methodology is applicable to other types of manufacturing systems.