Stochastic Planning And Scheduling For Reconfigurable Job Shops And Flow Lines

The uncertain and competitive market is leading manufacturers to look for fast and effective technological solutions to manage their production systems and make them highly responsive to market needs. Moreover, customers are requesting customized, high-quality products quickly and at low costs. Utilizing rigid manufacturing systems such as dedicated manufacturing systems (DMSs) or flexible manufacturing systems (FMSs) limits manufacturers’ responsiveness. Reconfigurable manufacturing systems (RMSs) were introduced to cope with these challenges. These systems are built around modularity and reconfigurability and use reconfigurable machine tools (RMTs) as their main component. The adjustable structure of RMT allows the system to adapt to market requirements. However, production management in RMSs is a particularly challenging task compared to traditional systems, which makes manufacturers skeptical about adopting these systems. To address this issue, this dissertation presents novel methodologies to manage production activities within RMSs regarding planning, scheduling, and control. The research was conducted in two main parts based on the system type (i.e., job shop or flow line). A novel mixed-integer linear programming (MILP) model for planning and scheduling is formulated for the former. Then, it was extended to a two-stage stochastic (TSS) formulation to incorporate the uncertainties in volume and machines’ productivity. A data-driven controller with predictive capabilities was developed for the latter. It collects real-time data to reschedule raw material injection time and control the inner-stage movement of work-in-process (WIP) units to optimize their levels. The applicability of the proposed models was validated using case studies adopted from the literature. The result of this dissertation showed the cost-benefits of utilizing RMSs and the effectiveness of adopting the proposed methodologies to manage RMSs.