Key points are not available for this paper at this time.
In order to develop a realistic simulation model, it is critical to provide the model with factual input data based on the interactions and events that take place between real entities. However, the existing trend in simulation of construction fleet activities is based on estimating input parameters such as activity durations using expert judgments and assumptions. Not only may such estimations not be precise, but project dynamics can influence model parameters beyond expectation. Therefore, the simulation model may not be a proper and reliable representation of the real engineering system. In order to alleviate these issues and improve the current practice of construction simulation, a thorough approach is needed that enables the integration of field data into simulation modeling and systematic refinement of the resulting models. This paper describes the latest efforts by authors to design and test a novel methodology for multimodal-process data capturing, fusion, and mining that provides a solid basis for automated generation and refinement of simulation models that realistically represent construction fleet operations. Different modes of operational data are collected and fused to facilitate the discovery of operational knowledge required to create realistic simulation models. The developed algorithms are validated using laboratory scale experiments and analytical results are also provided. The main contribution of this research to the body of knowledge is that it lays the foundation to systematically investigate whether it is possible to robustly discover computer-interpretable knowledge patterns from heterogeneous field data in order to create or refine realistic simulation models from complex, unstructured, and evolving operations such as heavy construction and infrastructure projects.
Akhavian et al. (Mon,) studied this question.