Geometrical Variations Management in a Multi-Disciplinary Environment with the Jacobian-Torsor Model

In recent years, several computational tools for tolerance specification and analysis have emerged to help the designer in the specification of geometrical product requirements. This paper proposes to broaden the scope of such a tool, called the Jacobian-Torsor model, so as to embrace a wider range of variations and uncertainties. In that perspective, the paper presents an original typology of geometrical variations and uncertainties. These variations are classified according to their own properties and features. Among the features that must be dealt with regarding variations and uncertainties are the following: cumulative versus independent and required versus random or predictable. In turn, the variations or uncertainties must themselves be identified through either, standards, measurements or models. Finally, in terms of simulation, these variations and uncertainties can be addressed through deterministic or stochastic approaches, or both, with the Jacobian-Torsor model.