A hybrid decision-support framework for cost-efficient manufacturing system design and operations under demand uncertainty
提出一个五步混合决策框架，结合混合整数非线性规划与离散事件仿真，在需求不确定性下优化单元制造系统的成本、资源利用率和在制品库存，并通过敏感性分析验证了设计的稳健性。
Uncertainty in supply chains amplifies the challenges of manufacturing system design, where strategic and operational decisions must be aligned to avoid inefficiencies such as underutilised capacity and excess Work-In-Process (WIP), which compromise cost efficiency. Although prior studies have examined cost, utilisation, and WIP, these factors are typically addressed separately, with limited attention to their combined trade-offs in system design. This study develops a five-step hybrid decision-making framework for designing and evaluating a cost-efficient layered Cellular Manufacturing System (CMS). At its core, a Mixed-Integer Nonlinear Programming (MINLP) model is proposed to minimize total cost while accounting for resource utilisation and queueing-based WIP estimation under uncertainty. Discrete Event Simulation (DES) evaluates the design, with flow time and WIP linked through Little’s Law to throughput as indicators of steady-state system. Regression analysis examines how KPI variation influences total cost, demonstrating robustness. A desirability-based method is then used to synthesize cost and KPIs to select the optimal configuration. Results show that the proposed design outperforms classical and layered CMS benchmarks by lowering total cost and stabilising utilisation and WIP. Moreover, sensitivity analysis on demand variation shows that the optimal cell configuration is preserved across all scenarios, indicating structural robustness with only modest cost variation.