Historical measurement data reuse and similarity analysis for dimensional production tolerancing of injected plastic parts

• Published in: Journal of the Brazilian Society of Mechanical Sciences and Engineering Vol. 39; no. 10; pp. 4161 - 4175
• Main Authors: de Oliveira, Ademir Linhares, Donatelli, Gustavo Daniel
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01-10-2017; Springer Nature B.V
• Subjects: Case studies; Data recovery; Dimensional analysis; Dimensional tolerances; Engineering; Feasibility studies; Mechanical Engineering; Reuse; Similarity; Technical Paper; Tolerances; Rule-based reasoning; Similarity analysis; Production tolerancing; Historical data; Process capability; Plastic injection molding
• ISSN: 1678-5878; 1806-3691
• DOI: 10.1007/s40430-017-0888-4

Geometrical variation is present in any production process and needs to be controlled, to ensure compliance with design requirements and minimize costs. Tolerancing of injected plastic parts (IPPs) is mostly performed in practice without using consistent criteria, many times resulting in tolerances which are unnecessarily narrow and incompatible with the respective manufacturing process capabilities. In this paper, a new method is proposed to support IPPs dimensional production tolerancing in early design phases, based on historical measurement data reuse and similarity analysis. Rule-based reasoning has been employed together with a scoring system to identify reliable and similar cases stored in a database, involving attributes of both the parts and the dimensional characteristics. Two different approaches for data recovery were considered, providing the ability to choose between a simplified and a more accurate analysis. The method feasibility is demonstrated by a case study involving 20 different IPPs and 24 dimensional characteristics from three suppliers. The proposed method outperformed the use of standard DIN 16742 to estimate production tolerances. Comparisons between the estimated production tolerances and the respective process capabilities resulted in maximum mean absolute deviation of 10% and Pearson’s
r
of about 0.6.