THE INVESTIGATION OF LEAD-TIME BUFFERING UNDER UNCERTAINTY USING SIMULATION AND COST OPTIMIZATION

The impact of uncertainty and variability in the productivity of trades aggravates the problem of work interruptions and idle time within repetitive activities. To eliminate the interruptions and idle time in order to achieve smooth work flow of resources, activities are deliberately delayed from their early start date. However, this practice induces a problem of tradeoffs between project cost and duration. Many recent studies have suggested that different types of buffers can be used to absorb the impact of uncertainty and variability on production work flow and most studies focus on using buffer to determine “when to halt an on-going production line”. In contrast, this paper focuses on “when to start a production line so that there is no interruption”. Two different approaches to lead-time buffering, the sequence step algorithm (SQS-AL) and the completed unit algorithm (CU-AL), are investigated using STROBOSCOPE (a discreteevent simulation system) with a special search add-in that implements a genetic-algorithm (GA). The investigation reveals that applying lead-time buffer provides better work flow and greater project profit; however, these depend on the penalty cost of work flow interruption and indirect cost. Both algorithms have implications that translate to advantages and limitations depending on assumptions, simplicity of simulation model, project characteristics, and uncertainty.