Data-driven decision-making models for prefabricated construction supply chain management

Abstract

Prefabrication has emerged as a promising approach in construction, offering substantial benefits in terms of productivity improvement and environmental impact reduction. However, the successful delivery of prefabricated constructions is subject to various constraints, rendering the prefabricated construction supply chain (PCSC) more complex and challenging to control. Given the significance of the PCSC, proper management becomes paramount for achieving successful prefabricated constructions. Nevertheless, PCSC management poses significant challenges, particularly the lack of an integrated framework. Accordingly, this thesis investigates relevant technologies and decision-making models in PCSC management and proposes an integrated framework to enhance the performance of the PCSC. To begin with, a general hierarchical digital twin-enabled framework for prefabricated construction supply chain management is proposed. This framework incorporates state-of-the-art technologies to enable visual project management in prefabricated construction. It encompasses key features such as supply chain and logistics visualization, material visibility and traceability, and on-site installation scheduling. To address the logistics collaboration challenges within a prefabricated supply chain, a three-tier supply chain structure involving multiple factories is introduced. A mathematical model considering critical factors such as factory production capacity, production cost and time variations, and material requirements prioritization is developed. Furthermore, a data-driven analytic framework is presented to understand the factors impacting supply chain on-site stage equipment utilization. The significance of influencing factors is examined using the Generalized Linear Model (GLM). Lastly, the proposed systemic digital twin-enabled framework is practically implemented in collaboration with a Hong Kong company to validate its practicality and rationality. The thesis makes several primary contributions. Firstly, a novel service-oriented 5D digital twin-enabled platform is proposed for visual project management in prefabricated construction. Complex event processing (CEP) technology and an improved hierarchical finite state machine are utilized to express the logical model of the prefabricated construction process and manage state transitions of events hierarchically. Secondly, a buffer layer is introduced in the prefabricated supply chain structure to achieve synchronization between supply and demand. The optimization model considers the site’s preferences for material arrival sequence and determines optimal production and transportation plans, considering the impact of construction schedule changes. The output of this model outperforms production plans based on the traditional Earliest Due Date (EDD) method. Thirdly, a GLM-based data analytic approach is proposed to improve equipment utilization. It establishes the relationship between different influencing factors and the number of idle equipment, facilitating proper resource allocation on the construction site and ultimately improving project performance. Lastly, the DT-enabled framework is implemented in a prefabricated pump room project in Hong Kong, leveraging traditional data sources and digital twin data to support hierarchical prefabricated construction supply chain management while minimizing disruption to existing workflows. The comprehensive framework holds promise for practitioners and scholars to address similar real-world challenges.