A holistic methodology and application for batch and worker scheduling problems in a dual resource constrained cellular manufacturing system

Cellular manufacturing systems (CMS) are efective productions systems for producing part families, which are composed of parts with similar processing requirements. Through the utilization of CMS, the setup time, work-in-process inventories, material handling cost, and lead time are reduced, while the visibility and quality are enhanced. There are several activities to be carried out in CMS before starting the production process. Scheduling is the final operational activity, and plays a key role in the smooth operation of CMS. Hence, finnding good solutions to the scheduling problems is very important for the successfull implementation of CMS. Scheduling problems in cellular manufacturing system (CMS) are widely investigated in the academic literature to increase the operational efficiency in production. These problems can mainly be classified into two groups: those that consider a single cell and those that consider multiple cells. The process for cell scheduling, known as group scheduling, can be divided into family and job scheduling stages. The family sequence and the job sequences at each family are determined simultaneously in order to compute the completion time for each job. However, the effect of employee skills on the cell cycle times, employee timetabling, and the application of one-piece flow within cells are not considered in cell scheduling. For this reason, a comprehensive batch scheduling problem that integrates employee timetabling and cell scheduling by considering employee skills together with the application of one-piece flow within cells would be a valuable addition to the existing academic literature. In the current thesis, the batch scheduling problem in a cellular manufacturing system (CMS) is addressed, considering skilled workforce assignment (SWA), which refers to employee flexibility. The CMS consists of a set of independent and parallel hybrid manufacturing cells. Unlike automated and assembly cells, both automatic and manual operations are performed in these hybrid cells, hence the reason for calling the cells hybrid. The importance of SWA for batch scheduling problems becomes especially clear for the CMS. As long as the employee flexibility is concerned for these cells, it requires taking into account the different skill levels for the employees to perform the manual operations; thus, the SWA has a crucial effect on the CMS performance whenever the manual operations have to be performed in the cells. Production systems in a dual-resource constrained (DRC) setting are the ones in which the capacity constraints on the output depend on the availability of both the machines and the employees. Employees required for manual operations are considered to be valuable resources when there are fewer employees than the number of workstations at which manual operations are performed. This kind of cellular manufacturing system (CMS) is considered to be a DRC system, which is a well-established trend in the literature. In this type of system, human operators are often constrained resources, and the movements of employees are both inter-cellular and intra-cellular. One of the most notable examples of CMS in a DRC setting in which employee resources directly affect the processing times of the batches are hybrid manufacturing cells, where both automatic and manual operations are performed. Furthermore, employee involvement is unrestricted in the hybrid cells, and the assignment of employees has a significant effect on the cell cycle times. Because of the manual operations within the hybrid cells, the cell cycle times vary significantly, and the batch processing times depend on the employees assigned to perform the operations on these batches. Considering employees as the critical resources for both their availability and effect on the manual operation times makes it more important to research this problem. In real manufacturing environments, employees are not unlimited resources, and they cannot be treated equally regarding their skill levels. In this thesis, the cells are arranged in a multi-product flow shop layout that applies one-piece flow, and the CMS has a DRC setting in such a manner that a batch can only be processed if employee resources and machine resources are available. Most studies on batch scheduling have focused on static scheduling problems. In these types of problems, employee resources are accepted to be non-renewable, and they cannot be assigned to more than one operation. In addition, the number of employees in each cell remains unchanged during the scheduling horizon. On the other hand, the employee resources are assumed to be renewable for dynamic scheduling problems, so they can be assigned to several non-overlapping operations. Therefore, the number of employees in each cell varies dynamically throughout the scheduling horizon. The current thesis focuses on the dynamic batch scheduling problem, considering employee resources to be limited. The contributions of this thesis are as follows: A comprehensive methodology is proposed to handle order release, batch scheduling and worker assignment problems all together. The batch scheduling problem considering one-piece flow application within cells for the CMS is investigated, which has not been studied in the academic literature to date. In addition, SWA with a DRC setting for CMS is considered. A new optimization model and holistic methodology are presented for the problem. A lower bound is determined to use for an experimental comparison of the meta-heuristics. Several structural properties are described, for a better explanation of the problem. The metaheuristics are compared regarding the solution quality, and the best among them is determined as the result of detailed analyses. In this thesis, the batch scheduling problem for CMS in a DRC setting is examined, considering SWA. The decision problems to be met include determining the sequence and starting times of batches, and assigning employees to cells in which operations are performed while considering employee skill levels. The development of an integer linear optimization model is described for the problem. Because the considered problem is NP-hard in the strong sense, five metaheuristics have been proposed, and compared with each other. For comparison purposes, computational experiments were conducted using a set of test instances that were generated based on real production data. The proposed methodology was applied on a real manufacturing system and the improvements were shown in this thesis.