On-Engine Expansion Measurement of Exhaust Manifold for Calibrating Thermo-Mechanical Fatigue FEA Model

An attempt was made as part of this work to acquire on-engine measurements to identify how closely current Finite Element Analysis (FEA) models replicate actual on-engine exhaust manifold behavior. Further correlation study with FEA models was performed to understand and eliminate the gaps to improve the overall FEA process. Dry cast iron exhaust manifolds experience thermo-mechanical fatigue (TMF) during engine operation. This is one of the critical failure modes. Literature is available to perform TMF assessment of exhaust manifold e.g. [1–6]. However, it is difficult to accurately predict TMF life of exhaust manifold in FEA due to dependency on multiple factors such as non-linear material behavior [3], temperature dependent material behavior, oxidation effect, creep effect, accuracy in prediction of metal temperatures and joint friction effects. Typically, non-linear material models, creep effects and oxidation effects are accounted by advanced fatigue processing software. Non-linear material models account for material and for temperature dependent non-linearity [4]. These non-linear material model and fatigue parameters are often developed using uniaxial specimen level testing. These doesn’t account for all the complexity during on-engine test due to factors such as friction and bolt loads that can influence manifold behavior. FEA processes for exhaust manifolds are seldom calibrated with on-engine measurements due to the complexity of obtaining these measurements in an environment that has severe temperatures and vibrations. The correlation study highlighted that exhaust manifold was over constrained by excessive clamping in FEA. This raised question on the gasket coefficient of friction (COF) and working preloads. These settings were investigated to get better correlation. Using reduced COF and non-linear material model for manifold capscrews, helped to achieve better correlation. Replacing material properties of manifold capscrews with nonlinear data provided capability to simulate localized yielding of capscrews and hence the corresponding load loss. Using these new settings for few other case studies also showed improvement in correlation of manifold warpage and thermal fatigue life prediction. Outcome of this work was a refined FEA approach which showed better FEA to Test correlation for exhaust manifold subject to thermal loading.