PhD Oral Exam - Mizanoor Rahman, Building Engineering

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

The global housing crisis, driven by a growing imbalance between housing supply and demand, has become an urgent concern, making access to affordable housing increasingly difficult for many. In Canada alone, an estimated 4.8 million new homes will be needed by 2035 to address this affordability crisis. However, meeting this target through traditional construction methods may be impractical due to their low productivity and inefficient use of resources. In this context, off-site construction (OSC) offers a promising alternative, with several key benefits, including increased productivity, enhanced safety and quality, and reduced costs and project delivery times. Despite these benefits, production managers continue to face significant challenges due to inefficiencies and imbalances in the production line. These imbalances arise from several key factors: (i) unrealistic allocation of workers at each workstation (WS); (ii) inaccurate estimation of panel-wise process times at each WS; (iii) inflexible production processes; (iv) non-standard panel configurations that cause significant variability in process times across components and workstations; (v) the absence of a factory layout design method tailored to OSC; and (vi) the lack of a dynamic manufacturing scheduling method capable of responding effectively to changing site demands. These imbalances result in a range of adverse outcomes, including increased project completion time (PCT), workstation idle time (WIT), work-in-progress (WIP) duration, and reduced resource utilization and throughput. Ultimately, these inefficiencies lead to broader consequences, such as longer waiting times at construction sites, transportation delays, decreased supply chain productivity, and higher overall production costs.

To address these challenges, this research introduces a comprehensive production planning and scheduling framework consisting of three integrated modules: (i) development of an integrated planning and scheduling method that combines the Linear Scheduling Method (LSM) to allocate an ideal workforce while considering space congestion, Monte Carlo simulation to estimate realistic labor productivity, and a multi-objective optimization model to simultaneously minimize WIT, WIP, and PCT; (ii) development of a simulation-based factory layout design method for OSC to establish the optimal factory layout that maximizes throughput and resource utilization while minimizing WIP, and module idle time; and (iii) development of a comprehensive Deep Reinforcement Learning-based Dynamic Manufacturing Scheduling (DRL-based DMS) model for OSC to generate optimal schedules for new projects and re-optimize existing ones to respond to dynamic events (e.g., addition of new projects due to changes in customer delivery dates and re-sequencing requests based on site installation requirements). The model aims to minimize three conflicting performance metrics: PCT, WIT, and WIP.

To validate the developed framework, this research employs two types of case studies: (i) a wood panelized production line located in Edmonton, Canada, and (ii) Taillard benchmark projects, to compare its performance against existing scheduling approaches. The first module proves effective in generating realistic and optimal production schedules, achieving reductions of 5.2% in PCT, 41.45% in WIT, and 56.42% in WIP. The second module achieves a 42.33% increase in throughput and an 11.33% improvement in resource utilization, as well as a 93.18% reduction in WIP and a 97.95% reduction in module idle time, compared to the existing layout. The third module not only re-optimizes existing schedules in near real-time in response to dynamic events (e.g., the addition of a new project or changes in sequencing), but also consistently outperforms previous approaches. The outcomes of this research make a significant contribution to the body of knowledge by enabling automated re-optimization in near real time, thereby reducing manual intervention and enhancing overall production efficiency.