No abstract

A B S T R A C T Many components experience combined temperature and stress loading and are designed to withstand creep. In this study, experimental creep testing was performed under both static and stepped loading conditions with constant temperature for two specimen geometries (tensile and three-point bend). The objective of this study was to evaluate whether existing damage accumulation models accurately predict creep performance when considering step loading and stress gradients. Model predictions, based on static tensile creep data and using a highly stressed volume correction for the three-point bend specimens and the experimental average damage sum, agreed well with experimental data; differences were on average within 38% (static) and 2.2 h (stepped). Comparisons showed more accurate predictions using an exponential Larson-Miller parameter curve and the Pavlou damage accumulation model. Findings of the current study have applicability to component design, where complex geometries often contain stress gradients and it is desirable to predict creep performance from static tensile creep data. N O M E N C L A T U R E a, B = Material constants in stress versus highly stressed volume relation C = Empirical constant used in Larson-Miller parameter D = Damage D * = Damage level which indicates failure E = Young's modulus K c = Creep stress state factor K t = Elastic stress concentration factor LMP = Larson-Miller parameter n = number of tests p = Fitting parameter used in Pavlou analysis R 2 = Coefficient of determination S y = Yield strength (0.2%) S u = Ultimate (tensile) strength t i = Time at stress-temperature combination i t f = Time to creep failure, rupture time T = Temperature T m = Melting temperature V = Highly stressed volume (≥ 95% of max. stress) λ = Exponent in linear damage summation lambda-model

Turbine housings for automotive turbochargers are complex castings that are subjected to transient thermal stress from exhaust gas temperature and flow variation under complex duty cycles. A fatigue model based on a cumulative damage approach (Chaboche model [1]) is used to predict crack initiation. As the stress varies with temperature, a modified Chaboche model based on a normalized stress is proposed. To determine the transient stress and temperature distributions due to the high rates of convection from the gas, and the complexity of the design, conjugate heat transfer CFD simulation is performed. The tongue of the turbine housing is a critical region in which cracks initiate within a short time. Heat transfer coefficients ( HTC ) and bulk temperature predictions from CFD, in general, can be validated by thermal measurement. But because of the geometry and the location of the tongue, it is impossible to measure the metal temperature. For this work 2 methods were presented: HTC calibrated by thermal measurement and HTC from CFD heat transfer Conjugate method, steady state analysis. Heat transfer Coefficients and bulk temperatures obtained with those 2 methods are different. The heat transfer from CFD analysis with 15 layers has an important HTC which is 3000 W/ m2 °C, it is equal to 1200 W/m2 °C for the calibrated method, but the bulk temperatures are not the same. A sensitivity study on predicted life shows that this difference results in no more than a 2 hr change for a total predicted life of 50 hrs. In the industrial approach this difference is quite acceptable. The design of a turbine housing is optimized based on this TMF methodology and shows very good results in testing, as presented here.

No abstract