Agent-based modelling and simulation of construction crew performance

Abstract

The construction industry suffers from chronic problems in project delays, crew ineffectiveness, and productivity loss. One of the root causes of such problems is improper planning and control. Project control requires not only sufficient experience, but also proactive decision-making and awareness. In order to ensure proactivity and problem awareness, simulation coupled with Lean Construction principles are used to inform the analysis of workers’ performance and conditions on site, facilitate production control, and detect possible future shortcomings or delays. This paper presents an Agent-Based and Discrete-Event model that allows project controllers to simulate current or future project states within the Weekly Work Plan (WWP) of the Last Planner® System (LPS) in order to orient the project activities and performance as desired. This model can be used to help generate more realistic planned production rates considering LPS metrics for crew capacity and performance. Factors resulting in the non-completion of tasks are taken from data of a real project over the course of 94 weeks. The simulation model is applied to an example project to explain the goals behind the proposed model. Results indicate that the proposed model is useful as a basis for a decision support system for project planners to evaluate the reliability of their planned production rates.