Strategies and predictive models for reducing workflow variability in construction production systems

In this research, workflow reliability is defined as the extent to which a construction plan is an accurate forecast of future events. In other words, the difference between what was planned to get accomplished and what is actually completed. Because construction projects consist of a large number of interdependent tasks, having a highly-reliable workflow is important not only for the productivity of an individual task, but also for entire project performance. The primary goals for this research are to develop methods and predictive models to identify and measure information exchange effectiveness in planning meetings, to quantify and analyze the benefit and cost trade-off for developing a highly-reliable plan, and to develop recommendations to guide site managers on making-do decisions. Information theory was utilized to measure the amount of uncertainty reduction in the workflow reliability due to information gained in planning meetings. Game theory was used to develop a framework to quantify and demonstrate the benefit of subcontractors’ extra effort for highly-reliable planning. Factor analysis was used to compare the underlying structure of the causes for task starting time and duration delay in China and the U.S. The Mann-Whitey U test, Random Forest machine learning, and entropy-based decision tree were used to identify the most important factors influencing making-do decisions. This research fills in gaps in the body of knowledge pertaining to workflow reliability improvement in construction projects. There has been limited research on quantifying the benefit of subcontractors’ horizontal partnering in increasing their planning reliability. Little effort has been made to investigate the importance of constraints removal discussions with respect to the existing workflow reliability. It is still unclear whether delaying start time of activities that lack precondition readiness will result in less duration delay. In addition to addressing gaps in the body of knowledge, this research has a broader impact on the construction community. It provides a framework and repeatable analytical methods to identify, prioritize, and direct management’s effort for increasing workflow reliability in construction projects.