A data-driven model with minimal information for bottleneck detection - application at Bosch thermotechnology

ABSTRACT

In the context of bottleneck detection, most data-driven approaches employ data from diverse production variables (machine processing times, machine state tags, input timestamps, etc.) for a detailed analysis of bottlenecks. However, for manufacturing companies initiating their digitalization process (i.e. requiring the smallest hardware investment), a bottom-top approach is still missing. In this work, a data-driven model based on minimal information (MI) retrieved from a manufacturing execution system is proposed for bottleneck detection. We consider MI timestamps when each product exits each station and show that this is the most elementary information from production-line operations, enough to autonomously generate an abstract manufacturing layout, and to detect and predict bottlenecks. A general abstract model of a production line is proposed, named queue directed graph (QDG). Incorporating the MI, the QDG model is able to represent a job-shop with a discrete production environment and to calculate production metrics. This work has been employed in the production system of a Bosch factory, in Portugal, using their manufacturing data sets for validation. Different variants of two well-known bottleneck detection methods were implemented and adapted to Bosch’s use case: the Active Period Method and the Average Active Period Method.

KEYWORDS:

• Minimal information
• data-driven model
• autonomous system generation
• bottleneck detection
• manufacturing production system

Highlights

• Bottleneck detection is studied using minimal information without having part-processing times or entry timestamps at locations.

• A mathematical modelling representation called queue direct graph (QDG) is introduced, which merges log data with graph theory and moving tokens, to derive bottleneck metrics.

• Variants of known bottleneck methods are adapted to the problems described by a QDG, which can be computed from the corresponding QDG state-space.

• The effectiveness of this data-driven approach is shown using real data from a Bosch manufacturing line where the streaming data pipeline flow is also outlined.

Acknowledgments

This work was conducted in the scope of the Project Augmented Humanity [POCI-01-0247-FEDER-046103], financed by Portugal 2020, under the Competitiveness and Internationalisation Operational Program, the Lisbon Regional Operational Program and by the European Regional Development Fund. The second author was partially supported by the Center for Research and Development in Mathematics and Applications (CIDMA), through the Portuguese Foundation for Science and Technology, reference UIDB/04106/2022.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

All data sets used are confidential information of Bosch company manufacturing systems, so they are not publicly available.

Notes

1. see https://www.bosch.com/stories/nexeed-smart-factory/