Operations Research 2018 Abstract Submission

A maturity model for the classification of real-world applications of data analytics in the manufacturing environment

Invited abstract in session WB-6: Business Track, stream Business Track.

Wednesday, 11:00-12:40
Room: 1g. Budapest

Authors (first author is the speaker)

1. Thomas Pschybilla
Central Department Digital Transformation, TRUMPF GmbH & Co. KG
2. Daniel Baumann
3. Wolf Wenger
DHBW Stuttgart
4. Dirk Wagner
5. Stephan Manz
Product Management, TRUMPF Laser- und Systemtechnik GmbH
6. Thomas Bauernhansl
Institut für Industrielle Fertigung und Fabrikbetrieb (IFF)


With the progressing digitalization in manufacturing continuously increasing amounts of data are being generated. This opens up various possibilities to utilize these data to improve production processes by supporting decision-making. The field of data analytics plays an important role in this context. Data analytics advances the acquisition of knowledge from data and, thus, decision-making in manufacturing and related processes such as maintenance.

Even if the importance of data analytics in the manufacturing context is undisputed, manufacturing companies are still in the early stages of the implementation and also predominantly use backward-looking types of data analytics. Identifying the current status of data analytics in the manufacturing environment reveals potential in this area and builds the basis for future developments.

This paper presents a theory-driven maturity model for the classification of data analytics use cases in the context of data utilization for analytics in manufacturing. Furthermore, the model aims to offer a subcategorization of the vast and complex topic of data analytics for manufacturing purposes. Different types like descriptive, predictive and prescriptive data analytics are integrated into the model. Up to the level of predictive data analytics, data-driven approaches like data mining play a major role. The potential of decision support beyond predictive data analytics can be achieved through the integration of prescriptive data analytics and the combination of data-driven and model-driven approaches. By its very nature, model-driven approaches need a model of the analyzed problem beforehand. The knowledge derived from data analytics allows for enhancing model-driven approaches by using its insights as input, e.g. for the implementation of specific operations research models and methods. Based on this differentiation, the stages of the maturity model are derived.

The model is verified using the example of predictive maintenance in the laser application of TRUMPF GmbH & Co. KG. With TRUMPF Condition Monitoring, both algorithms and service experts monitor the conditions of customer laser devices for the purpose of reducing unplanned downtime and thus increasing availability and productivity of connected laser systems. If a laser is at risk of malfunction, e.g. due to a decreasing cooling water level or due to a polluted filter, a warning is given proactively.

The use case for Condition Monitoring of connected lasers in the manufacturing environment is examined and classified in the developed model. The major potential of predictive data analytics is highlighted and first ideas towards prescriptive data analytics are presented. However, the use case also emphasizes the importance of human experience and interaction, factors that still play a significant role in the process of generating knowledge and deriving decisions.


Status: accepted

Back to the list of papers