Statistical tolerance analysis of mechanisms with interactions between deviations – a methodology with 10 easy steps
Year: 2013
Editor: Udo Lindemann, Srinivasan V, Yong Se Kim, Sang Won Lee, John Clarkson, Gaetano Cascini
Author: Walter, Michael; Wartzack, Sandro
Series: ICED
Institution: FAU Erlangen-Nuremberg, Germany
Page(s): 191-200
ISBN: 978-1-904670-48-3
ISSN: 2220-4334
Abstract
The time-dependent motion behavior of a mechanism is essentially affected by different kinds of deviations. Consequently, the product developer has to analyze the mechanism and its kinematic behavior as early as possible to ensure the product’s functional capabilities. However, possible interactions between the deviations and their effects on the system’s motion, and thus the functionality are not considered yet. This paper presents a methodology, consisting of 10 easy steps, which enables the product developer to perform a statistical tolerance analysis of a system in motion, which underlies different kinds of deviations as well as several interactions among them. Therefore, the identification as well as the determination (statistically) of the interactions is required, which can be done using numerical simulations like multi-body-dynamics or manufacturing process simulations. An appropriate mathematical representation of the interactions is done using meta-models (like Artificial Neural Networks). These can be easily integrated into the tolerance analysis’ functional relation. A case study of a non-ideal cross-arm window regulator illustrates the methodology’s practical use.
Keywords: Statistical tolerance analysis, artificial neural networks, robust design, simulation, design methods