Knowledge Sharing and Productivity Improvement: An Agent-Based Modeling Approach

Abstract

Labor productivity is a major determinant of project performance in construction. Models of labor productivity in construction tend to focus on learning curve theories that assume learning is an individual process with no transfer of knowledge among crew members. This paper seeks to extend theories of individual learning to capture the crew dynamics present on construction sites. Accordingly, this paper presents an agent-based model aimed at deriving the impacts of crew composition and project schedule on knowledge sharing and, thus, on task duration. The proposed model was calibrated using field observations of 201 interactions among 12 construction workers at a construction project in Beirut, Lebanon. The results indicate that more diverse crews witness higher levels of knowledge sharing and greater productivity gains. The results also suggest that schedules keeping all the workers busy eliminate the potential for knowledge sharing and thus only benefit from the baseline gains seen in individual learning. This work contributes to the literature by developing an agent-based model that simulates knowledge sharing in the construction industry at the worker level. The study is limited by its exclusion of multiskilled workers.