SVR and GA Aided Lean Six Sigma Method for Planning in Modular Construction

Abstract

Modular construction presents a strong alternative to traditional construction, offering advantages such as improved productivity, and better quality. However, the prefabrication of module components follows a make-to-order process, resulting in customized module components. This design customization, along with various factors such as worker skill levels, and defects in shop drawings causes significant variability in the process times for prefabricating module components at workstations. This variability leads to imbalanced production line, and idle time at workstations, which increases the overall completion time of fabricating module components. To address these challenges, this paper develops a Lean Six Sigma based method that comprises three modules. In the first module, the production line that requires improvements is identified and project objectives are defined. In the second module, the process time data of module components at workstations are collected to identify and analyse inefficiencies in the production line utilizing six sigma performance metrics. The third module focuses on improving and controlling the production line process using support vector regression (SVR) and meta-heuristic optimization. A light gauge steel (LGS) wall panel production line in Edmonton, Canada was analysed to demonstrate the use of the developed method and test its performance. The results show that, after addressing the production line bottlenecks, the sigma level improves to 1.85 σ compared to 1.41 σ earlier. This method can help production managers identify wastes and bottlenecks in the production line, enabling them to plan their processes more efficiently.