A Study on Critical Chain Buffer Sizing

• Main Authors: Chien, Wan-Ling, 簡菀伶
• Other Authors: Lin, Kao-Cheng
• Format: Others
• Language: zh-TW
• Published: 2010
• Online Access: http://ndltd.ncl.edu.tw/handle/90723196119352525395

碩士 === 南台科技大學 === 工業管理研究所 === 98 === In the theory of Critical Chain/Buffer Management (CC/BM), the safety associated with the critical chain tasks is shifted to the end of the critical chain in the form of a project buffer, the aim of which is to protect the project due-date promised to the customer from variation in the critical chain tasks. Feeding buffers are placed whenever a non-critical chain activity joins the critical chain. Both kinds of buffers are pseudo-activities added to protect the critical chain from disruptions on the activities feeding it, and to allow critical chain activities to start early in case things go well. In the literature, the most widely used buffer sizing methods are the Cut and Paste Method and the Root Square Error Method (RSEM). Herroelen and Leus (2001) performed a full factorial experiment on a set of benchmark problems to test the CC/BM scheduling mechanism, and then reported that the Cut and Paste Method may lead to a serious overestimation of the required buffer protection. From the results of this experiment, it can be also found that the project makespan highly depends on the average resource utilization and the order strength. Therefore, Tukel et al. (2006) took these two factors into consideration, and then proposed two adjustment factors for the RSEM. However, their adjustment factors may lead to a significant and unnecessary expansion. From the statistical theory of confidence intervals, when the project network is a chain, the buffer sizes determined by using RSEM are usually large enough to integrate the uncertainty in activity durations. Therefore, in this study, we give some revision to their adjustment factors. To avoid unnecessary expansion, our adjustment factors take the value 1 in cases that the resource is very sufficient, and the network is a chain. From the result of numerical examples, we find that the Adaptive Procedure with Density method proposed by Tukel et al. (2006) usually produces a schedule very close to the earliest schedule. This will increase the WIP and fall into the vicious cycle of traditional scheduling approaches. In addition, since a path that is not critical in the traditional scheduling approach, but has a highly uncertain total duration, may be the path that determines the makespan in practice. Therefore, taking the uncertainty in activity durations into consideration is important, when determining which path is critical. From the numerical example, we find that this may also reduce the possibility that the length of buffer added can not be as long as we wish. We also note that when the critical path found using this method is not a path with the longest total average duration, there exist some activities in the critical path which have nonzero safety floats. These float times will reduce the impact of the uncertainty of their precedent activities’ durations.