Abstract: Underground caverns in water conservancy and hydropower projects are often faced with significant constraints due to a limited construction space, which severely affects the scheduling and coordination of construction machinery. This paper develops a dual-objective optimization function for minimizing both construction duration and cost, and examines various constraints on the construction, such as the number of machines, process time, and process logic. We construct a multi-agent-based simulation model for construction in confined space. It simulates complex construction processes and the queuing service process of construction machinery, and makes analysis of the parallel operations and potential conflicts of construction machinery during operation. Then we give a suggestion on how to adjust the number of construction machines based on the variables. Comparison of the simulations against the field observations of construction and its cost under different machinery configurations is conducted to validate the optimized schemes for construction machinery scheduling. A case study of the Yangtze-to-Hanjiang Water Diversion Tunnel shows a capacity of 25 t is optimal for dump truck loading. As the tunnel excavation progresses, the optimal number of mucking dump trucks in a confined space takes a dynamic hierarchical decreasing trend, while the trend is increasing for the trucks in a non-confined space. Compared with the original plan, our optimized schemes save 149.3 days in construction duration and 35.02 million yuan in cost, thus providing a better decision for construction design and scheduling and on-site management.

Key words: underground cavern, mechanery scheduling, confined space, optimization simulation