Application of Bayesian updating for real-time schedule estimation in a concreting operation

Abstract

Ready-Mixed Concrete (RMC) delivery schedules should be carefully planned due to the limited time span between mixing and placing. The scheduling of RMC supply is challenging to both contractors and plant managers because of multiple operational issues such as traffic conditions and unloading delays. Simulation has been applied to address this challenge; however, the unique and complex nature of construction environments raises a validity issue in the use of previous datasets as input for the simulation of a new operation. Based on this recognition, the idea of using real-time data has been proposed to address the input data validity issue and to enhance the predictability of the simulation model. One important concern that has not yet been fully addressed regarding this issue is the amount of data collection required to yield a valid sample such that the output is sufficiently reliable. To answer this question, this paper uses a Bayesian updating technique with data collected from a concreting operation in a high-rise building project in Abu Dhabi. Through the case study it is concluded that the Bayesian forecasting technique has great potential for providing a better statistical representation of operational data and improving the predictability of the simulation model.