Overview
Traditional sales and operations planning (S&OP) in industrial production often relies on profitability metrics, such as the single-product gross profit margin. These metrics operate under the assumption that individual production lines can be committed independently, simplifying product prioritization through a per-ton ranking. However, this premise is challenged in facilities characterized by coupled production processes, particularly polymer plants. In these plants, parallel production lines frequently draw simultaneously from a single bulk feed, a condition referred to as forced co-production. This inherent coupling can compromise the effectiveness of standard profitability metrics, potentially leading to suboptimal and value-destroying production plans.
To address this challenge, a novel framework has been developed, centered on the concept of the Augmented Gross Profit for Product Clusters (AGPPC). This framework aims to transform the conventional per-ton ranking metric into a 'per-coupled-hour' ranking. A distinguishing feature of the AGPPC framework is its provision of a 'per-instance optimality certificate', indicating its capability to certify optimal production outcomes for specific operational scenarios.
Research Context
The standard S&OP toolkit employs metrics like single-product gross profit margin to prioritize products. This methodology presumes that production lines operate independently, allowing for straightforward per-ton ranking, which is computationally simple and easily applied by planners. This independence, however, is not universally applicable, especially within coupled plant environments. Polymer production plants exemplify this challenge, where multiple parallel lines share a common bulk feed, necessitating co-production. Such interdependence can invalidate the assumptions underlying independent line prioritization, leading to potentially inefficient or detrimental production strategies.
Approach
The proposed framework, built upon the Augmented Gross Profit for Product Clusters (AGPPC), converts the standard per-ton ranking metric into a per-coupled-hour ranking. This conversion is designed to explicitly account for the interdependencies present in coupled production environments. The theoretical foundation of AGPPC integrates two key components: the 'co-production column concept' and a 'fluid relaxation' of the planning problem. The co-production column concept likely represents a method to model the simultaneous production of multiple products linked by shared resources or processes.
In conjunction with the AGPPC framework, the research incorporates an integrated bilinear mixed integer program. This program is formulated to model the complexities of the plant's operational dynamics. To provide verifiable performance benchmarks, the framework also includes a McCormick-linearized baseline. This baseline is utilized to derive certified bounds for the bilinear mixed integer program, offering a measure against which the optimality of the AGPPC-derived plans can be assessed.
Findings
The effectiveness of the AGPPC framework and the associated optimization approach was demonstrated using real operational data. This data originated from an Indonesian polymer producer. The evaluation indicated that implementing the coupling-aware metric alone resulted in a 7.6% increase in operating profit when compared to existing operational practices. Furthermore, the application of the full optimization methodology, which combines the AGPPC framework with the integrated bilinear mixed integer program and McCormick-linearized baseline, yielded a more substantial increase. The full optimization achieved a 28% rise in operating profit relative to the producer's current practice.
Why This Matters
The findings indicate a substantial financial impact from implementing coupling-aware planning methodologies in industries with forced co-production. For the Indonesian polymer producer studied, the AGPPC framework alone delivered a 7.6% increase in operating profit, while the comprehensive optimization approach boosted operating profit by 28%. This highlights the potential for significant profitability improvements by moving beyond conventional, independent production line assumptions in S&OP.