Digital twin-enabled on-site synchronization for prefabricated construction management

Abstract

Prefabricated construction is an industrialized approach to produce prefabricated components in off-site factories and deliver these components for on-site assembly. On-site assembly is one of the most uncertain and complex stages in precast construction scenario, due to high variability of outdoor conditions, organization of multi-contractors, and geographic dispersion of activities. To handle the complexity and uncertainty in precast on-site assembly, multi-operations and logistics-operation synchronizations are essential for resource allocation and task scheduling. Supported by the cutting-edge information technologies, digital twins are generated by integrating geometric and semantic information with real-time updating data, which can be utilized to simplify computational models and convert stochastic factors into deterministic parameters. However, several challenges exist. Foremost, current on-site resources couldn’t be efficiently and consistently digitalized, and the cyber-physical interoperation is fragmented and out-of-date. Moreover, how to fully utilize the real-time data to generate coordinated decisions that integrate the operation and material supply in a synchronized manner under uncertain site conditions is also a challenge. Additionally, traditional optimization models can offer theoretically optimal solutions, but most solutions are not effective to handle uncertainties and always consume huge computational resources. Therefore, this thesis aims to propose a digital twin-enabled real-time synchronization mechanism with associated decision-making models for precast on-site assembly. At the first place, a digital twin-enabled systematic framework is proposed for smart precast construction sites, focusing on collaborative decision-making and daily operations. Cloud-based services can facilitate managers to monitor the real-time statuses of on-site resources through high-fidelity virtual models, and support operators to execute daily tasks with automatic navigations. A desktop robotic demonstration is conducted for performance evaluation. For digital twins’ generation and operation, a Ubiquitous Digital Twin (UDT) model is formulated for information management in precast on-site assembly. This reference model enhances domain-focused knowledge transfer and provides a unified approach for digitalized on-site assembly. Two cases for the transdisciplinary management of a precast nuclear plant are hierarchically instantiated based on the proposed UDT model. Moreover, a digital twin-enabled real-time synchronization mechanism (DT-SYNC) is formulated for precast on-site assembly, which facilitates collaborative decision-making and dynamic control through real-time data-driven digital twins. Horizontal and vertical synchronizations ensure coordinated operations, supported by a ticket-based decision-making model for logistics-assembly operations. For initial planning, a multi-objective optimization model is developed for logistics-assembly operations management in on-site fit-out assembly projects. Digital Twin as a Service, harnesses real-time data about resources and operations to drive a rolling-horizon-based optimization model. A computational experiment is conducted using Non-dominated Sorting Genetic Algorithm-II (NSGA-II) for model testing. For dynamic scheduling and execution, a real-time-data-driven Out-of-Order (OoO) model is formulated for fit-out construction planning, scheduling, and execution. OoO model, inspired by the CPU mechanisms, enhances visibility and traceability for dynamic decision-making and operator guidance. Stochastic computational experiments demonstrate the effectiveness under different uncertainty scenarios. Overall, this thesis leverages the potentials of digital twin to reshape the management mode of precast on-site assembly with enhanced synchronicity, resilience, and flexibility to handle complexity and uncertainty under stochastic site environments.