Flexible Flow-Shop Scheduling For Cutting-Tool Manufacturing In Turkish Aerospace Industries (TAI)
44. Yöneylem Araştırması ve Endüstri Mühendisliği Kongresi, Ankara, Türkiye, 25 - 27 Haziran 2025, ss.125-126, (Özet Bildiri)
- Yayın Türü: Bildiri / Özet Bildiri
- Basıldığı Şehir: Ankara
- Basıldığı Ülke: Türkiye
- Sayfa Sayıları: ss.125-126
- İstanbul Kültür Üniversitesi Adresli: Evet
Özet
Flow shop scheduling addresses the problem of determining the optimal sequence and assignment of jobs to machines in a production system where each job is processed sequentially through multiple stages. The objective is to optimize performance metrics such as makespan, total tardiness, or the number of late jobs. Since Johnson’s fundemental work on the two-machine flow shop problem, numerous approaches have been proposed in the literature for various types of flow shop scheduling problems over the past seventy years. This study focuses on a high-tech production environment where manufacturing occurs across more than two stages, and some stages contain multiple identical machines. At this production line, carbide cutting tools are manufactured. In each stage, identical machines are performing different operations, and incoming parts follow a specific routing sequence. The presence of identical machines at certain stages differentiates the problem from classical flow shop scheduling, categorizing it as a Flexible Flow Shop Scheduling Problem (FFSP) in the literature. For FFSPs, the literature presents various mathematical models—such as Mixed Integer Linear Programming (MILP), Stochastic Programming, and Fuzzy Mathematical Programming—depending on real-world constraints. Two main approaches are typically used for solving these models: exact (structural) algorithms and heuristic/metaheuristic methods. However, due to the NP-hard nature of the problem, optimal solutions are achievable only for small-sized instances under specific assumptions. While exact algorithms can be successful in limited cases, they often fail to handle medium and large-scale instances and may be overly complex for practical applications. In contrast, heuristic and metaheuristic methods—though not guaranteeing optimality—offer effective and scalable solutions to scheduling problems. This project investigates the scheduling of production and regrinding tasks on a multi-machine carbide tool production line, where each machine performs distinct operations. Due to the availability of identical machines at some stages, the problem qualifies as a flexible flow shop. MILP models from the literature were revised and extended by incorporating constraints specific to the industrial partner, TAI. To solve the model, exact solutions were first obtained using the GAMS CPLEX solver for problems of varying sizes. Furthermore, for large-scale problem instances where exact solutions cannot be found, heuristics like SPT, LPT, and metaheuristic methods (namely, Evolutionary Algorithms) are designed and implemented in Python. The performance of these approaches—exact, heuristic, and metaheuristic—will be compared using real-world datasets of different scales obtained from TAI. This study aims to improve the continuity of flow, increase the machine utilization, and reduce delivery times on TAI’s carbide cutting tool production line. The developed scheduling framework is intended to offer a more effective alternative to current manual planning processes, while contributing a generalizable solution approach to FFSPs. Furthermore, a user-friendly interface will be developed to improve data flow between technicians and planners and to generate reports for production scheduling. This approach is expected to provide lasting improvements in TAI’s production processes and boost overall efficiency.