Simulation-Based Scheduling of Modular Construction Using Multi-agent Resource Allocation

Modular construction is common practice for building industrial plants in the Alberta oil sands region, Canada because of the savings in cost and schedule, and improving safety and quality. Each module represents a unique construction project. Thus, modular construction is considered as multi-project construction. Scheduling and effective allocation of resources (e.g., space, skilled crew, construction equipment) in such large-scale construction projects is a challenging process. The production schedule is expected to satisfy numerous uncertain factors and constraints posed by the site layout, resource limitations, the construction process, and various supply chains (e.g., spool fabrication shop). Traditional network-based scheduling approaches are ineffective in scheduling the multi-project environment of modular construction and optimum allocation of resources. This paper presents a hybrid approach based on discrete event simulation modeling and Multi-Agent Resource Allocation (MARA) for scheduling modular construction. Modules represent agents who bid for resources (e.g., space in the yard) to maximize their individual welfare. An auctioneer is also designed who allocates the available resources to the bidding agents by maximizing the overall welfare of the society of agents. The auctioneer can employ various combinatorial optimization algorithms (greedy and ascending-auction algorithms in this study) to allocate the resources to the agents. Auctions are held regularly until the end of the simulation model to allocate the resources among agents. This paper also presents the successful implementation of this approach in an actual case study of a module assembly yard.