Graph-based similarity analysis of BOM data to identify unnecessary inner product variance

Michael Schmidt, Benedikt Gehring, Jan Sebastian Gerber, Johannes Michael Stocker, Matthias Kreimeyer, Markus Lienkamp

Research output: Contribution to journalConference articlepeer-review

6 Scopus citations


This paper contributes to the fields of variant management and product family design. The focus lies on analysing historically grown product portfolios in order to reduce unnecessary inner variety. Such inner variety adds no value to the customer, yet it induces complexity costs within the whole company. Increasing transparency in documented product variants is key when applying standardisation or modularisation methods as part of variant management. Studies of literature and industrial practice at a major German truck manufacturer show that analysing product structure information from BOM data yields the potential to point out promising candidates in companies' portfolios for effective standardization or modularisation. For modelling and analysing highly variant and complex product structures, we employ graph-based modelling of BOM data in combination with a state-of-the-art tree matching algorithm for similarity calculations. Actual product data of a truck manufacturer serves as a case study. Thereby, we propose a generally applicable approach that enables intuitive handling of large amounts of product family data and that effectively supports variety reduction efforts.

Original languageEnglish
Pages (from-to)489-498
Number of pages10
JournalProceedings of the International Conference on Engineering Design, ICED
Issue numberDS87-1
StatePublished - 2017
Event21st International Conference on Engineering Design, ICED 2017 - Vancouver, Canada
Duration: 21 Aug 201725 Aug 2017


  • Complexity
  • Data analysis
  • Data visualization
  • Product families
  • Product structuring


Dive into the research topics of 'Graph-based similarity analysis of BOM data to identify unnecessary inner product variance'. Together they form a unique fingerprint.

Cite this