Usage of Probabilistic Design and Statistical Tolerance in Crankshaft Development

Within the evolution of components in the automotive industry, several tools for development and optimization are being applied increasingly. Costs and performance parameters are then improved in such high rate that often the process limits are exceeded. In several designs these improvements are achieved through nominal/deterministic values for product properties and usage in its lifecycle, neglecting the fact that there is an inherent uncertainty. Several optimum solutions are not assured once a variation in input parameters for objective function is considered. Probabilistic tools may be applied in order to establish the link among such parameters and their expected variation, enabling alternative approaches for product development in an integrated way. Different solutions can be found once the uncertainty of process or usage is considered. Statistical tools can as well be used to minimize the sensitivity of output parameters for an expected variation, improving the robustness of the solution, and to analyze the potential benefits of improving specific inputs. Through this document, it is intended to investigate the usage of dimensional variation analysis in automotive components design, specifically illustrated for a crankshaft design. Particular methods to mention: Monte Carlo, MVFOSM (Mean-Value First Order Second Moment), FORM (First Order Reliability Method) and FOSA (First Order Saddlepoint Approximation), are briefly described. A basic application of those probabilistic tools is then exposed as an illustrative example for a crankshaft residual unbalancing analysis. In the conclusion, it is verified then, actual benefits on the usage of such probabilistic tools, some of their specific advantages and disadvantages in the design and development of automotive components.