BATCH SCHEDULING MODEL ON THREE-STAGE FLOWSHOP WITH TWO-STAGE JOB PROCESSOR AND ONE STAGE BATCH PROCESSOR TO MINIMIZE TOTAL ACTUAL FLOWTIME

• Main Author: Poeri Suryadhini, Pratya
• Format: Dissertations
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/70875
• Institution: Institut Teknologi Bandung

This dissertation research discusses batch scheduling on a three stage flowshop that processes multi-product with two stages of assembly process using a job processor and one stage of inspection process using a batch processor. Stage I and Stage II are both assembly processes, but the difference is that in Stage I all types of products will be processed on a common machine, whereas in Stage II each type of product will be processed on a dedicated machine. Stage III is an inspection process performed on a common machine used by all product types. This inspection stage is applicable to both 100% and sample inspection methods. Non-conforming products are handled by rework, increasing production quantities, or assuming a substitute product is available. Based on the characteristics of the production process and the inspection process, as well as the use of two different types of machines, the production scheduling process must be aligned with the processing time at each stage, in order to avoid delays in delivery to consumers Process alignment between stages becomes complex due to the existence of two types of batches. The batching process of a production batch is determined by the product differentiation process, which refers to the process of determining the type of product contained in a production batch. In this dissertation, product differentiation is carried out in either Stage I or Stage II. If product differentiation occurs in the Stage I, the production batch formed in the Stage I will consist of only one type of product, resulting in the production batch formed in the Stage I being the same as the production batch formed in the Stage II. Meanwhile, if the product differentiation is in the Stage II, the production batch in the Stage I is allowed to consist one or more types of products, whereas the production batch in the Stage II is only allowed consist of one type of product. As a result, the production batch in the Stage I is not always the same as the production batch in the Stage II. The difference in differentiation will affect the batch scheduling model and its decision variables. The batching process of an inspection batch in Stage III is determined by the quality testing method used, which in this research was 100% inspection and sample inspection. The inspection batch for 100% inspection is established from a group of production batches that the number of parts cannot exceed the batch processor's capacity. While the inspection batch for sample inspection is established from a group of samples from production batches that the number of parts cannot exceed the batch processor's capacity. The development of batch scheduling models will be impacted by the difference of inspection methods. There are three stages of model development in this dissertation research: the development of a model with product differentiation in Stage I and 100% inspection, known as Model 1A, the development of a model with product differentiation in Stage I and sample inspection, known as Model 1B, and the development of a model with product differentiation in Stage II and 100% inspection, known as Model 2. These models are developed with the goal of avoiding delivery delays to customers by minimizing the total actual flowtime This three-stage batch scheduling problem can be formulated as an integer non-linear model. An integer decision variable is required for batch size and a non-linear occurs due to the multiplication between the batch processing time span and the batch size. Since each model is difficult to solve using analytical methods, this research developed a heuristic algorithm to find the best solution. Numerical examples using hypothetical data are used to evaluate the proposed algorithms. Based on the results, it can be concluded that it can generate reasonable results from all data sets tested, that is producing a schedule that does not violate the time zero or exceed the due date, and there is no overlapping schedule at all stages, and the resulting schedule is a schedule that minimizes total actual flowtime.