Date of Award
Industrial and Manufacturing Systems Engineering
Alternative process plans, Bacterial foraging optimization algorithm, Dual-resource problem, Flexible job-shop scheduling problem, Multi-objective optimization, Sequencing flexibility
Marketing strategists usually advocate increased product variety to attend better market demand. Furthermore, companies increasingly acquire more advanced manufacturing systems to take care of the increased product mix. Manufacturing resources with different capabilities give a competitive advantage to the industry. Proper management of the current productions resources is crucial for a thriving industry. Flexible job shop scheduling problem (FJSP) is an extension of the classical Job-shop scheduling problem (JSP) where operations can be performed by a set of candidate capable machines. An extended version of the FJSP, entitled FJSP with sequencing flexibility (FJSPS), is studied in this work. The extension considers precedence between the operations in the form of a directed acyclic graph instead of sequential order. In this work, a mixed integer programming (MILP) formulation is presented. A single objective formulation to minimize the weighted tardiness for the FJSP with sequencing flexibility is proposed. A different objective to minimize makespan is also considered. Due to the NP-hardness of the problem, a novel hybrid bacterial foraging optimization algorithm (HBFOA) is developed to tackle the FJSP with sequencing flexibility. It is inspired by the behaviour of the E. coli bacteria. It mimics the process to seek for food. The HBFOA is enhanced with simulated annealing (SA). The HBFOA has been packaged in the form of a decision support system (DSS). A case study of a small and medium-sized enterprise (SME) manufacturing industry is presented to validate the proposed HBFOA and MILP. Additional numerical experiments with instances provided by the literature are considered. The results demonstrate that the HBFOA outperformed the classical dispatching rules and the best integer solution of MILP when minimizing the weighted tardiness and offered comparable results for the makespan instances. In this dissertation, another critical aspect has been studied. In the industry, skilled workers usually are able to operate a specific set of machines. Hence, managers need to decide the best operation assignments to machines and workers. However, they need also to balance the workload between workers while accomplishing the due dates. In this research, a multi-objective mathematical model that minimizes makespan, maximal worker workload and weighted tardiness is developed. This model is entitled dual-resource FJSP with sequencing flexibility (DRFJSPS). It covers both the machine assignment and also the worker selection. Due to the intractability of the DRFJSPS, an elitist non-dominated sorting genetic algorithm (NSGA-II) is developed to solve this problem efficiently. The algorithm provides a set of Pareto-optimal solutions that the decision makers can use to evaluate the trade-offs of the conflicting objectives. New instances are introduced to demonstrate the applicability of the model and algorithm. A multi-random-start local search algorithm has been developed to assess the effectiveness of the adapted NSGA-II. The comparison of the solutions demonstrates that the modified NSGA-II provides a non-dominated efficient set in a reasonable time. Finally, a situation where there are multiple process plans available for a specific job is considered. This scenario is useful to be able to react to the current status of the shop where unpredictable circumstances (machine breakdown, current product mix, due dates, demand, etc.) can be accurately tackled. The determination of the process plan also depends on its cost. For that, a balance between cost, and the accomplishment of due dates is required. A multi-objective mathematical model that minimizes makespan, total processing cost and weighted tardiness are proposed to determine the sequence and the process plan to be used. This model is entitled flexible job-shop scheduling problem with sequencing and process plan flexibility (FJSP-2F). New instances are generated to show the applicability of the model.
Vital Soto, Alejandro Vital, "Flexible Job-shop Scheduling Problem with Sequencing Flexibility: Mathematical Models and Solution Algorithms" (2019). Electronic Theses and Dissertations. 7661.
Available for download on Monday, September 30, 2019