Sensitivity Analysis of Cylinder Head Fatigue Life Prediction Using Statistical Models

Abstract

For the efficient virtual development of combustion engine cylinder heads in terms of high cycle fatigue (HCF) it is highly important to have a reliable development process that represents reality in the best possible way. Most of today’s standard HCF procedures are capable of delivering high quality results for a specific load combination. However, loads are usually subject to variation. This is also valid for loads the cylinder head is subjected to. Assembly loads and operating loads considered during the virtual development process are widely determined by the production process which again is subject to variation due to certain tolerances, wear of the tooling equipment etc. As it is highly important to ensure the fatigue design of a cylinder head, there is the need for new analysis models capable of capturing every possible load variation. Within the framework of this paper the influence of different variable loading parameters on the cylinder head HCF margin of a heavy duty diesel engine will be discussed. A design of experiments (DoE) analysis is used together with the 3-d finite element method (FEM) for the investigations. Furthermore a methodology for the probabilistic assessment of the cylinder head HCF margin based on stochastic loading data is introduced. At the same time an effective methodology for the identification of the worst case boundary conditions for HCF analysis will be presented. With the presented probabilistic method it is possible to achieve a highly accurate prediction of the HCF design margin. Due to the probabilistic approach a better understanding of the entire system is possible, as the interaction between input and output parameters can be illustrated. Therefore HCF optimization problems can be encountered more effectively. Furthermore the presented methodology can be used for error estimation of analysis results and assessment of the result sensitivity. Thus, a borderline layout of the cylinder head can be achieved. Also the minimum input information quality, which is required for a profound HCF analysis, can be assessed by using the sensitivity analysis presented. Therefore the proposed methods enable a fast and reliable development of cylinder heads and other combustion engine components.