CAPACITY OPTIMIZATION OF CNC MACHINING CENTERS IN HIGH-MIX LOW-VOLUME MANUFACTURING: AN INTEGRATED SIMULATION AND SCHEDULING APPROACH

High-mix low-volume (HMLV) manufacturing environments face persistent challenges in achieving high utilisation of CNC machining centres while simultaneously meeting delivery due-dates for diverse, low-batch-size orders. This dissertation presents an integrated framework combining discrete-event simulation (DES), analytical capacity modelling, and a modified Genetic Algorithm (GA) scheduling optimiser to maximise throughput and minimise makespan in a real-world HMLV job shop equipped with seven CNC machining centres of varying capability tiers. A capacity model based on Overall Equipment Effectiveness (OEE) and utilisation rate analysis was developed and calibrated against 18 months of production data from a precision parts manufacturer in Tashkent, Uzbekistan. Bottleneck machines were identified using the Theory of Constraints (TOC) methodology, revealing that two 5-axis centres (OEE = 61.3%) were limiting shop-floor throughput by an estimated 23%. A DES model built in Siemens Tecnomatix Plant Simulation was validated against historical key performance indicators (KPIs) within ±4.8% for throughput and ±6.1% for average flow time. The GA scheduler, operating on a 48- hour rolling horizon, reduced average job tardiness by 38.4%, increased average machining centre utilisation from 67.2% to 81.9%, and improved OEE of bottleneck centres from 61.3% to 76.8% through optimised preventive maintenance windows and setup-time reduction strategies. Annual capacity gain equivalent to 1.7 additional machine-shifts per day was demonstrated without capital investment. The framework provides a transferable decision-support tool for HMLV manufacturers seeking productivity improvement through data-driven scheduling.多品种小批量（High-mix low-volume, HMLV）制造环境长期面临双重挑战：既要实现数控加工中心的高利用率，又需同时满足多样化小批量订单的交付工期要求。本论文提出一套集成框架，整合离散事件仿真（discrete-event simulation, DES）、分析性产能建模与改进型遗传算法（Genetic Algorithm, GA）调度优化器，旨在为配备7台不同性能层级数控加工中心的真实HMLV车间最大化吞吐量并缩短制造周期。 基于设备综合效率（Overall Equipment Effectiveness, OEE）与利用率分析构建了产能模型，并借助乌兹别克斯坦塔什干一家精密零件制造商18个月的生产数据完成模型校准。采用约束理论（Theory of Constraints, TOC）方法识别瓶颈机床，结果显示两台OEE为61.3%的五轴加工中心预估会使车间吞吐量降低23%。 基于西门子Tecnomatix Plant Simulation搭建的DES模型，经历史关键绩效指标（key performance indicators, KPIs）验证，吞吐量误差在±4.8%以内，平均流动时间误差在±6.1%以内。采用48小时滚动时域运行的GA调度器，通过优化预防性维护窗口与换产时间缩减策略，使工件平均拖期降低38.4%，数控加工中心平均利用率从67.2%提升至81.9%，瓶颈机床的OEE从61.3%提升至76.8%。 无需新增资本投入即可实现相当于每日额外1.7个机床班次的年度产能提升。本框架可为寻求通过数据驱动调度提升生产力的HMLV制造商提供可迁移的决策支持工具。