Comparison of damage parameter based Finite Element fatigue life analysis results to combustion chamber type TMF panel test results

Abstract

The inner liner of a regeneratively cooled liquid rocket engine (LRE) combustion chamber is usually exposed to the high temperature of the hot gas and the low temperature of the coolant. It must withstand the pressure within the combustion chamber and the cooling channels. A cyclic operation of a LRE causes a local thinning of the combustion chamber wall (the so called doghouse effect) after a very low number of cycles. Thermomechanical Fatigue (TMF) panels, small actively cooled sections of the hot gas wall of the real engine in combination with cyclic laser heating, are used to study the doghouse effect without the need for testing a full scale engine. An already published numerical Finite Element analysis of the TMF panel on the basis of post-processing Coffin-Manson law resulted in a considerable underestimation of the fatigue life of this TMF panel. In order to improve the prediction of the fatigue life of the TMF panel, a visco-plastic model coupled with isotropic ductile damage was implemented as a user-material in the commercial Finite Element package ANSYS. Furthermore, thermal ageing is implemented in the model in order to take into account the change of the material microstructure with time at elevated temperatures. The temperature dependent material parameters for creep, kinematic and isotropic hardening as well as isotropic damage are determined using data from uniaxial tensile, LCF and stress relaxation tests. The number of cycles to failure is determined numerically and compared to experimental results of the TMF panel. The damage parameter based Finite Element analysis rightly predicts the damage initiation point in the middle cooling channel of the TMF panel and reaches the critical damage (the point where a mesocrack is initiated) at the 122 cycle instead of the experimentally obtained 174 cycle. The effect of the increase of the wall thickness in the fin areas is also obtained by this numerical analysis. However, this phenomenological approach which does not take into account the crack closure effect is not able to predict the doghouse effect.