Dynamic, Data-Driven Simulation In Construction Using Advanced Metadata Structures and Bayesian Inference

Abstract

Effective project control in construction requires the rapid identification and subsequent mitigation of deviations from planned baselines and schedules. Although simulation has been used to successfully plan projects in the pre-construction phase, the use of simulation for project control during execution remains limited. Current real-time simulation strategies have difficulty self-adapting in response to deviations from planned baselines, requiring experienced simulation experts to manually update the input parameters of simulation models. This study is proposing a dynamic, data-driven simulation environment that is capable of minimizing the manual intervention required to incorporate as-built construction data in real-time by coupling newly-developed metadata structures with Bayesian inference. Still in development, an overview of the proposed simulation environment is presented, details of the advanced data structures are discussed, and preliminary functionality of the environment is demonstrated.