The aim of this thesis is to set up a methodological approach to assess a fatigue life of components under cyclic thermal loads. Therefore, a copper mould used for continuous steel casting is considered as a case study. During the process, the molten steel passes through a water cooled mould. The inner part of the component is subjected to a huge thermal flux. Consequently large temperature gradients occur across the component, especially in the region near to the meniscus, and cause elastic and plastic strains.
The finite-element thermo-mechanical analysis is performed with a three-dimensional numerical model. One of the challenging tasks is choosing a suitable material model which is going to be applied in a simulation; since the amount of resulting plastic and elastic strain is strongly controlled by the material model implemented to perform the analysis. Therefore, four different material models (linear kinematic, combined, stabilized and accelerated material model) are investigated and compared in this thesis. It has been found that the combined model requires huge computational time to reach a stabilized stress-strain loop. On the other hand, the use of the stabilized model overestimates the plasticization phenomena already in the first cycle. Accordingly, the alternative accelerated material model, where stabilization is reached earlier, is thus proposed, proofing that it is able to give suitable and safe life estimation for design purposes.
Material coefficients for all applied material and fatigue life models are estimated from experimental, isothermal low cycle fatigue data of CuAg alloy at three temperature levels (20 °C, 250 °C, 300 °C).
A strain-based fatigue model, appropriate to assess a service life of the component, is necessary to apply once the material model is chosen and the finite-element analysis is performed. A fatigue model compatible and suitable for a daily industrial practice due to its simplicity; however in the same time able to predict precisely a fatigue life. The fatigue life of analysed component is assessed depending on different material models and fatigue models (Universal Slopes equation, Modified Universal Slopes equation, the 10% Rule and 20% Rule), as well as design curves (deterministic approach, tolerance interval, EPI)
