Predicting Assembly Performance with Featurized Datum Flow Chains

Throughout the development cycle of an assembled product, design and manufacturing engineers commonly perform ad hoc studies to determine whether their creation will meet its dimensional objectives. More simply, they answer the question, “will it go together and work right the first time?” These studies usually involve adding the maximum tolerance allowances in a single direction to find the maximum possible error. Known as a “worst case” study, this method underestimates the accuracy of an assembly because it assumes each part is always manufactured at the extremes of its specification limits. This paper proposes and describes the featurized datum flow chain as an adaptive and comprehensive alternative for calculating predicted variation in characteristics of assemblies. The paper advances prior research performed for the Lean Aircraft Initiative at the Massachusetts Institute of Technology, and it offers an example for completing an assembly study. The featurized datum flow chain captures the load sequence, part constraint, and feature associations for an assembly in a format that can be widely understood and reused. By adding manufacturing capability tolerances to the features in the chain, a user can complete simple root-sum-square calculations to find the predicted error in the assembly. The results of the study are more accurate than worst case findings because root-sum-square mathematics combines probability distributions of the tolerances. Furthermore, the datum flow chain provides a thorough picture of the entire assembly process to its users. Such common understanding is valuable to the design community in ways that reach far beyond tolerance studies.