Motivation

The value stream mapping method is a widely used method in the manufacturing industry for analyzing and redesigning material and information flows with the aim of improving processes and reducing waste. The task of value stream analysis is to create transparency about the actual state of an existing value stream in order to identify potentials for an improved production flow. Following on from this, a low-waste target state is developed in the course of the value stream design. Due to its ease of use and high effectiveness, this method is one of the most important tools of lean production. However, its weaknesses are well known in research and practice:

• Since the modeling language is focused on simple, linear material flows, it is difficult to capture complex value streams with many branches or even multiple product families.
• Recognizing the process steps, recording the correct data and designing a target state are non-trivial and depend on the experience of the user.
• The value stream map is a static representation of production, based on snapshots and estimates, in which no variability is captured.
• The manual recording of times, inventories and key figures makes the repetition of the value stream mapping very time-consuming.

Objectives

The advancing digitalization in the production environment offers the potential to address the weaknesses of the method through the targeted use of data analysis and thus raise it to a new level. The Institute for Production Management, Technology and Machine Tools has been conducting research on this topic since the beginning of 2019 and has been cooperating with the management consultancy STAUFEN.DIGITAL NEONEX GmbH for this purpose since January 2020. The objective of the contract research is to lay the methodological foundation for deriving automated evaluation formats from operational and planning data that simplify and improve the value stream method. The result is a collection of useful analysis tools that build on a common database and are used within the framework of a proven methodological procedure. The tools and the procedure are jointly researched and developed. Their immediate use in consulting projects enables direct feedback, which serves continuous further development.

Approach

The team for carrying out the VSM will be expanded in the data-assisted procedure to include a data scientist who will support the product family definition with cluster analyses, supplement the value

stream mapping with a neutral perspective based on process mining evaluations, and enable design activities with data analyses tailored for this purpose to be more meaningful.

1. product family formation and segmentation

For the segmentation of the value stream, product families are determined based on a product-process matrix. This can be determined from the company's transaction data using data mining. The grouping of product families is accelerated and simplified by the use of clustering methods. Carrying the product family assignment along in the further data analyses allows separate analysis and design of the value stream segments without disregarding their interrelationships with other products, such as through shared resources.

2. data-assisted value stream analysis

Process mining techniques can be used to reconstruct process flows from production feedback data and calculate associated metrics. Discovery algorithms are used to support process mining. The process map generated in this process captures several product families simultaneously and can map complexity. Following on from the process discovery, performance analyses can be carried out, which make it possible to extract key figures such as throughput times, processing times or waiting times from the event data. This enables reliable quantification of the value stream analysis. In initial industrial projects, for example, it was possible to correct customer demand, the share of a product family in the annual volume produced, or the machine allocation in the case of shared resources by more than 20% in each case compared to the initial estimate of the process experts.

3. data-assisted value stream design

To support the VSD, appropriate analysis formats for the evaluation of master and transaction data are developed for each design activity. These analyses allow scenarios for alternative designs to be created and evaluated. A common data basis enables accurate consideration of all interrelationships. Capacity balancing, the design of FIFO and supermarket inventories, and the planning of (minimum setup) production sequences play a central role. The data-based approach increases the effectiveness of the method through well-founded decisions and facilitated the prioritization of Kaizen activities based on scenario analyses. In addition, the data structure created facilitates the repetition of VSD activities with little effort, so that the value stream can be continuously developed.

Acknowledgement

We would like to thank STAUFEN.DIGITAL NEONEX GmbH for supporting the project and for the always good cooperation.

Funding source