Optimization of spatial layouts for deep underground infrastructure in central business districts based on a multi-agent system model

• Published in: Tunnelling and underground space technology Vol. 135; p. 105046
• Main Authors: Zhang, Chenhao, Zhao, Ziwei, Guo, Dongjun, Gong, Dandan, Chen, Yulu
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2023
• Subjects: Central business districts; Deep underground infrastructure; Deep underground spaces; Multi-agent system; Deep underground spaces; Deep underground infrastructure; Multi-agent system; Central business districts
• ISSN: 0886-7798; 1878-4364
• DOI: 10.1016/j.tust.2023.105046

•Provided a bottom-up approach for urban deep underground space planning.•Defined the factors that affect the layout of deep underground infrastructure (DUI)•Built a DUI layout optimization model based on multi-agent system.•Analyzed the spatial utilization form and distribution of deep underground space.•Discussed the layout type and phased development strategy of DUI. In central business districts (CBDs), shallow underground spaces are being rapidly saturated owing to the increasing scale of urban underground development. Compared to shallow underground spaces, deep underground spaces are better suited for the installation of infrastructural facilities, and therefore, they are expected to become important in future urban space development. However, deep underground infrastructure (DUI) planning in CBDs is a complex and dynamic process involving numerous players because of the different service objectives and layout-related requirements of each facility. Thus, the requirements and rules of DUI layouts need to be investigated for the orderly development of underground spaces. This paper proposes a method for optimizing DUI layouts based on a multi-agent system model to maximize the benefit in DUI layouts. In this method, each DUI is represented by an agent, and deep underground space including spatial scale, ground utilization form, geological conditions and other factors was constructed, and six layout expectations of different types of DUI agents were set within the framework of the spatial relationships between DUI and ground demand, DUI and deep subsurface resource development, and different DUIs. The simulation obtained a spatial representation of the optimal DUI layout within a spatial environment that represents the deep underground space of a CBD through the exploration of DUI space and DUI–DUI interactions as well as the adaptive behaviors of the agents. The results indicate that the overall spatial utilization form of deep underground space presents a parabolic wave-like pattern with shallow central aggregation. In the vertical dimension, the facility distribution exhibits clear vertical zonation. In the horizontal dimension, the facility distribution is dominated by the ground demand, but it is greatly affected by the infrastructure of high-rise buildings and reduces the service efficiency of DUI. According to the key influencing factors affecting the DUI layout, DUI is divided into demand-driven, cost-sensitive, and supporting-service types, and targeted DUI planning and development strategies are proposed. This model can effectively describe the laws of using deep underground space to set up infrastructure and provides policy development support for the decision-making process when developing deep underground space.