A novel method for sensitivity analysis and characterization in integrated engineering design

Purpose:The present research work aims to analyze and characterize processes in terms of sensitivity of their performances. Robust design techniques, generally adopted for product and process optimization, are not suited for investigating sensitivity. Then a novel approach to such engineering problem needs to be proposed.Method:The developed method integrates and extends to the analysis of manufacturing and technological processes the Performance Sensitivity Distribution (PSD) theory, primarily introduced to provide analytical and geometric description of performance sensitivity for robotic mechanisms.Result:Such novel method, named Specialized PSD, starts from the clarification of the sensitivity analysis problem by defining key parameters, i.e. Design Variables (DVs), Design Parameters (DPs) and Performance Functions (PFs). According to the PSD theory, PF sensitivity is expressed in terms of deviations of DVs and DPs and it is geometrically described by a hyperellipsoid in the n-dimensional space. Sensitivity indexes are then introduced to assess PF variation for different combinations of DVs and DPs deviations. Regression Analysis is adopted to provide the mathematical description of PFs so the PSD theory is finally specialized to be applied in a process sensitivity analysis. Injection molding of a plastic specimen is finally investigated to validate the proposed method.Discussion & Conclusion:This work specialize the PSD theory for manufacturing and technological processes, extending its original field of application thanks to a novel approach to the analytical expression of the PFs. Moreover, when 2 or 3 parameters are considered, sensitivity indexes are graphically represented through tolerance maps of colour, so the method can be easily adopt for integrated design, especially in the early stage of product and process development.