A method for assessing the robustness of mechanical designs

Reliability is recognized as an important factor relating to the quality of a product, and robustness describes a means of improving reliability. Generic robustness methods rely on data of some sort, either experimental, established performance distributions, or precise engineering knowledge based on past design attempts. Design for Reliability methods, a class of reliability tool, facilitate methodical analysis, and a similar robust design method is suggested within the present paper that helps improve robustness in industries where the aforementioned data are unavailable. Thus the method can be used before any manufacture has taken place, and yields a numerical result, thus facilitating comparison of alternative designs. The method was developed by studying failure data associated with similar mechanisms. A link was sought between the mechanism geometry and the failure data, such that a mechanism analysis, performed before and after a change, would determine the benefit, regarding robustness, associated with that change. The method was then tested using further case studies, each consisting of two similar designs that satisfied the same requirements, to see whether relative robustness could be predicted. The results indicate that the method can show which design is more robust, but cannot say by how much robustness is changed.