A Contribution to the Development of Sustainable Target Value Streams with Machine Learning Considering Material Flow Cost

Within the scope of maximizing value creation and eliminating waste, value stream mapping is considered as a well-established lean management tool which overall results in incomplete improvements due to its sole concentration on waste types that are assessable via lead time. Since resource efficiency gains increasing importance for industrial production, existing research has extended value stream mapping by the concept of material flow cost accounting. This extension relativizes the given lead time exclusivity and enables material and energy-based wastes to be factored. During application of this extended value stream mapping significant expenses arise in terms of data acquisition and processing, as well as calculation complexity and time-cost balance. Value-adding utilization of rising volume and complexity of data for generation of new target value streams in direction of the ideal state through improvement teams seems to be no longer a viable solution. To contribute to the design of a suitable solution for the future, a machine learning-based model concept is introduced as a hypothesis in this research paper. Within the prospective application, this model concept enables to use of traditional and extended KPIs of the current value stream as input. Through defined tasks, rules, and the algorithm value-adding analysis can be performed and assist in the discovery of target value streams through the resulting output. Overall, this digital application to be developed can thus assist improvement teams at their work and can contribute to discovering waste-optimized and more sustainable target value streams in an industrial environment.