Comparison of Uncertainty Analyses for Crankshaft Applications

Abstract

In this paper, four convex models (interval, parallelepiped, ellipsoidal, and superellipsoidal analyses) are introduced to describe the available data on uncertainty parameters for the engine’s crankshaft. The paper first evaluates the smallest area, such as box, ellipse, parallelogram, and super ellipse, which enclose the available data. Then, the Tchebycheff inequality is employed to inflate the uncertain domain to address the problem of forecasting data, i.e., variables taking values beyond the recorded uncertain data, as a limited amount of samples are used to construct the convex models. The minimum areas before and after the inflation are evaluated. Subsequently, the maximum stresses before and after the inflation of uncertain domain based on the areas of the convex models are obtained. The domain that predicts the minimum of the maximum stresses is declared as the best bounding figure, along with the attendant uncertainty model.