Buffer management of critical chain projects based on activity dimension differences

As project complexity increases, stricter requirements are imposed on project buffer management. However, existing buffer management approaches primarily consider project attributes while neglecting variations in activity dimension attribute differences, making it difficult to address the differentiated buffering needs among activities and ultimately hindering overall project utility optimization. This study proposes a Dimension-Differentiated Buffer Management (DDBM) method, which refines buffer allocation and monitoring processes to enhance project performance. First, activity attribute indicators are constructed, and the K-means++ clustering algorithm is employed to categorize critical chain activities into subgroups based on dimensional differences, followed by buffer settings and buffer allocation for each subgroup. Second, considering subgroup-level differences, a comprehensive utility coefficient is established to determine buffer distribution among activities. Finally, a dual-level dynamic buffer monitoring mechanism is introduced, incorporating both activity-level and project-level monitoring, based on a qualitative analysis of subgroup-specific buffer demands. The Monte Carlo simulation results demonstrate that the proposed DDBM approach effectively reduces false schedule alarms and achieves an integrated optimization of cost and duration, thereby enhancing buffer monitoring performance. By incorporating activity dimensional differences into buffer management, the proposed model provides decision-makers with an effective tool for managing differentiated buffering needs in complex projects.随着项目复杂度的提升，对项目缓冲管理提出了更为严苛的要求。然而现有缓冲管理方法主要考量项目整体属性，却忽视了活动维度属性差异带来的变化，难以适配各活动间差异化的缓冲需求，最终制约了项目整体效用的优化。本研究提出一种维度差异化缓冲管理（Dimension-Differentiated Buffer Management, DDBM）方法，通过细化缓冲分配与监控流程以提升项目绩效。首先，构建活动属性指标体系，采用K均值++（K-means++）聚类算法，基于维度差异将关键链（Critical Chain）活动划分为不同子群，随后针对各子群开展缓冲设置与缓冲分配工作。其次，考虑子群层面的差异，建立综合效用系数以确定活动间的缓冲分布方案。最后，在对子群专属缓冲需求进行定性分析的基础上，引入兼顾活动层级与项目层级的双层动态缓冲监控机制。蒙特卡洛（Monte Carlo）仿真结果表明，所提DDBM方法可有效降低进度误报率，实现成本与工期的协同优化，进而提升缓冲监控性能。本模型将活动维度差异纳入缓冲管理范畴，可为复杂项目中差异化缓冲需求的决策提供有效工具。