A decision support system for optimized off-site construction planning using a pull-based simulation approach

Abstract

The growing adoption of prefabrication in construction presents opportunities for enhanced productivity, yet challenges remain in accurately estimating labor hours, optimizing production sequences, and improving multi-project coordination. This dissertation addresses these challenges by developing a BIM-based hybrid optimization framework that integrates data augmentation, productivity modeling, heuristic sequencing, and Discrete Event Simulation (DES) to enhance steel fabrication efficiency.

First, a data-driven productivity estimation model is proposed, incorporating BIM-based automated quantity take-offs and data augmentation techniques to refine labor-hour predictions. This model significantly improves estimation accuracy by leveraging Bayesian linear regression and Artificial Neural Networks, achieving up to a 71% improvement over current industry practice.

Next, the research introduces a BIM-integrated heuristic sequencing model that utilizes engineering specifications and time study data to forecast labor hours and optimize workstation assignments in steel fabrication plants. By combining productivity modeling with DES, the proposed framework achieves a 99% correlation with actual production data and demonstrates a 13 % reduction in labor costs. A comparative analysis with Genetic Algorithms highlights the efficiency of the heuristic approach in optimizing fabrication sequences.

Finally, the dissertation extends production planning methodologies to multi-project steel fabrication plants, where managing resource allocation and optimizing project coordination remain critical challenges. A hybrid optimization approach incorporating similarity-based clustering, process modeling, and DES is developed to improve cost efficiency and minimize logistical issues such as missing components, component transfer delays, and late shipments. By aligning production scheduling with shipping timelines, this approach ensures that required components arrive at the job site as needed, effectively synchronizing fabrication and delivery. This demand-driven planning framework enhances multi-project coordination, streamlining prefabrication workflows through a pull-based simulation approach.

The proposed methodologies offer a systematic and scalable approach to enhancing productivity, reducing labor inefficiencies, and optimizing production planning in steel prefabrication projects. By bridging the gap between BIM-based digital modeling, productivity estimation, and advanced simulation techniques, this dissertation provides a robust decision-support framework for improving offsite construction operations. Future research directions include expanding the framework to other construction sectors and refining optimization techniques for real-time adaptive scheduling and resource management.