A simulation-optimization approach for adaptive manufacturing capacity planning in small and medium-sized enterprises

• Published in: Expert systems with applications 2021-04, Vol.168, p.114451, Article 114451
• Main Authors: Teerasoponpong, Siravat, Sopadang, Apichat
• Format: Article
• Language: English
• Subjects: Artificial neural network; Artificial neural networks; Computer simulation; Enterprise resource planning; Genetic algorithm; Genetic algorithms; Manufacturing; Manufacturing capacity planning; Optimization; Simulation; Simulation-optimization; Small & medium sized enterprises-SME; Small business; SMEs
• ISSN: 0957-4174; 1873-6793
• DOI: 10.1016/j.eswa.2020.114451

•Exploits Observational and collected data to plan the manufacturing capacity.•Proposes a data-driven approach in labor-intensive manufacturing.•Adopts the concept of smart manufacturing with artificial intelligence.•Promising implementation results in a real-world case study. Manufacturing capacity planning is one of the critical processes in every manufacturing company, and, with increasing exploitation of data and information technology, has necessarily become more efficient than before. However, the power to harness data and information for planning requires specific knowledge and resources, mostly limited to large enterprises. Small and medium-sized enterprises (SMEs) generally do not have sufficient resources to collect large amounts of data or the know-how to process and exploit data. Moreover, SMEs often fail to implement advanced techniques and tools (e.g., optimization tools or enterprise resource planning (ERP) software), owing to the cost and a lack of specific knowledge and personnel. This paper proposes a solution for reducing the burden on SMEs in collecting and utilizing data for the planning of manufacturing capacity. A simulation-optimization approach is adopted because of the complex nature of labor-intensive manufacturing in SMEs. The approach includes an artificial neural network for model simulation and data relationship recognition, combined with a genetic algorithm for optimizing manufacturing resource configuration. The proposed method can facilitate the process of planning manufacturing capacity for different yield targets, as tested in a case study of a pastry company and providing the means for the company to exploit both empirical and observational data for the purpose.