End-to-End Supply Chain Planning in the Paper Industry Via Column Generation and Benders Decomposition
Published in arXiv preprint arXiv:2607.16618, 2026
Changkun Guan, Amira Hijazi, Pascal Van Hentenryck. "End-to-End Supply Chain Planning in the Paper Industry Via Column Generation and Benders Decomposition." arXiv preprint arXiv:2607.16618, 2026.
Large-scale paper manufacturing requires coordinating production scheduling, trimming, vehicle loading, and multi-period demand fulfillment. Optimizing these stages sequentially creates inefficiencies, while full integration is computationally challenging. This paper develops an exact mathematical model of the integrated end-to-end planning problem and a hybrid BDCG-DP framework that combines column generation with dynamic programming for supply decisions and Benders decomposition for fulfillment, exploiting the structure that downstream decisions depend on upstream choices only through aggregate supply availability.
On proprietary industry instances, BDCG-DP achieves a 24.4% cost reduction compared to traditional CG-DP on eight-week problems, reduces four-week runtimes from five hours to under one hour, and delivers integer-feasible plans within 2.3–6 hours for complex scenarios — planning capability previously unavailable to practitioners.