Developing a generic resource allocation framework for construction simulation

The allocation of resources over time, referred to as resource scheduling, in large-scale construction environments is a challenging problem. Although traditional network scheduling techniques are the most popular scheduling techniques in the construction industry, they are ineffective in modeling the dynamic nature and resource interactions of large projects. Simulation based modeling or optimization techniques are also time-consuming, complicated and costly to be implemented in large-scale projects. This research is focused on developing a new framework to insert artificial intelligence inside construction simulations for facilitating the resource allocation process.;The second objective of this project is to develop a generic resource allocation component for addressing optimized resource allocation in various construction projects. This component is developed in a large scale model using High Level Architecture (HLA), instead of traditional simulation environments. HLA allows splitting a large scale model, known as a federation, into a number of manageable components (i.e., federates), while maintaining interoperability between them. A generic Resource Allocation (RA) federate is designed to act as an auctioneer for federates developed based on the SBAP. Another generic federate is also built to automate the communication with the RA federate. These two generic federates can be reused in various construction federations. This framework is successfully implemented in an industrial construction process that involves different supply chains including spool fabrication, module assembly and heavy crane lifts in site construction.;The first stage in this study was developing a framework to solve resource scheduling problems in large scale construction projects. This framework, called the Simulation Based Auction Protocol (SBAP), integrates Multi-Agent Resource Allocation (MARA) in a simulation environment. This hybrid framework deploys centralized MARA (i.e., auction protocols) whereby agents bid on different combinations of resources at the start of a simulation cycle. Agents attempt to improve their individual welfare by acquiring a combination of resources. An auctioneer is designed to allocate resources to the agents by maximizing the overall welfare of the society. Simulation is also employed to track the availability of resources, and manage resource oriented activities. This framework is implemented in two large construction applications of scheduling module assembly yard and multiple heavy lift planning in modular construction.