Lessons Learned from Schedule Estimation Using Real-Time Data in a Concreting Operation

The unique, complex and interdependent nature of modern construction operations means that the use of previous datasets in simulating a new project may raise validity issues. To address this problem, the use of real-time data in simulation studies has been introduced and practiced in recent years. However, little is known about the details of how to use real-time data to better estimate the performance of an operation. In particular, there is little awareness of the timing of data collection required to yield a valid sample on which to base subsequent plans and schedules for the entire operation. Using data from a concreting operation in a Dubai construction project, this paper examines the validity and reliability of forecasted results using different segments of the available data. For this purpose, the datasets are statistically analyzed and then segmented based on the percentage of the operation completed. A discrete event simulation model is developed to estimate the schedule performance using each cumulative data segment and the simulation results are compared to actual project data. Through the case study, it was found that the statistical distribution identified at an early stage in the operation may not successfully represent the statistical properties of the whole dataset due to local extreme variation; and that it is important to identify the bottleneck resource (e.g., pumps in this study) and pay particular attention to variation in its cycle time in order to successfully estimate and control the performance of an operation.