Stream of Variation Based Error Compensation Strategy in Multi-Stage Manufacturing Processes

Abstract

Linear state space Stream of Variation (SoV) models of error flow in multistation assembly and machining systems have been well studied in the past decade. SoV models were utilized for identification of process-level root causes of manufacturing errors, quantitative characterization of measurements in multistation manufacturing systems, systematic selection of measurement points and features, as well as tolerance allocation and process design. Nevertheless, natural connection of the linear state space form of SoV models with traditional control theory has not been utilized to automatically compensate observed manufacturing errors and thus close the quality control loop. Recent advances in measurement technology and flexible fixtures make such operations possible and in this paper, we propose a method for strategic elimination of root causes of quality problems based on the SoV models of the flow of manufacturing errors. Furthermore, the concept of compensability that quantitatively depicts the capacity of error compensation in a specific system is proposed. Based on this concept analogous to the controllability in the traditional control theory, compensable and non-compensable subspaces of dimensional errors are identified and quantitatively described. Theoretical results have been demonstrated using the SoV model of a real industrial process used for machining of automotive cylinder heads.