A two-stage robust optimisation model for multi-project scheduling and material ordering under demand uncertainty
针对按订单设计和制造系统中需求不确定、供应商产能和交货期变化的问题，提出了一个两阶段鲁棒优化模型，并用精确算法求解，发现列与约束生成法比Benders分解法快90%以上，同时分析了应对不确定性的策略组合。
Managing uncertain material demands in engineer-to-order (ETO) and make-to-order (MTO) systems is critical for preventing resource mismatches. This study addresses the integrated multi-project scheduling and material ordering problem under demand uncertainty, time-varying supplier capacities and lead times. We formulate a two-stage robust optimisation model and develop exact algorithms—specifically column-and-constraint generation (C&CG) and Benders decomposition—enhanced by valid inequalities. Computational experiments demonstrate C&CG’s superiority, achieving a 90.45% reduction in solution time and 96.67% optimality convergence compared to Benders decomposition. Furthermore, sensitivity analysis reveals that while minor fluctuations are managed via order adjustments, significant uncertainty necessitates combining inventory buffers, backup suppliers and schedule revisions. This research provides a robust decision-support paradigm for mitigating volatility in project-driven operations.