Optimizing real-world factory flows using aggregated discrete event simulation modelling

• DOI: 10.1007/s10696-019-09362-7
• Publication Date: 2019-07-01T16:03:07Z
• Authors: Simon Lidberg; Tehseen Aslam; Leif Pehrsson; Amos H. C. Ng
• Keywords: 0209 industrial biotechnology; Industrial problem; Digital storage; Decision trees; 0211 other engineering and technologies; Discrete event simulation; 02 engineering and technology; Decision support systems; Trees (mathematics); Aggregation; Knowledge extraction; Industrial case study; Simulation-based optimizations; Data mining; Multiobjective optimization; Decision-tree algorithm; Industrial systems; Production Engineering, Human Work Science and Ergonomics; Agglomeration; Produktionsteknik, arbetsvetenskap och ergonomi; Decision support; Multi-objective optimization; Real-world optimization; 13. Climate action; Decision supports; Decision making

AbstractReacting quickly to changing market demands and new variants by improving and adapting industrial systems is an important business advantage. Changes to systems are costly; especially when those systems are already in place. Resources invested should be targeted so that the results of the improvements are maximized. One method allowing this is the combination of discrete event simulation, aggregated models, multi-objective optimization, and data-mining shown in this article. A real-world optimization case study of an industrial problem is conducted resulting in lowering the storage levels, reducing lead time, and lowering batch sizes, showing the potential of optimizing on the factory level. Furthermore, a base for decision-support is presented, generating clusters from the optimization results. These clusters are then used as targets for a decision tree algorithm, creating rules for reaching different solutions for a decision-maker to choose from. Thereby allowing decisions to be driven by data, and not by intuition.