Abstract
Alternating temperatures induce thermomechanical stresses in thick-walled components such as turbine rotors or housings, which can lead to fatigue and superimposed creep. Subsequently, damage can occur at their heated surfaces. Under the nowadays prevailing operating conditions of power plants with multiple cold, warm and hot starts as reaction to the high volatility of electric demand from fossil fired power plants for ensuring grid stability, methods for lifetime assessment are coming more into the focus of investigations and research. Engineers are trying to estimate the residual lifetimes of in-service components and operators of power plants ask for strategies to minimize the calculative material damage while simultaneously providing a maximum flexibility with shortest response times on altered demands. Among constitutive models, which are not subject of this paper, accumulative models for lifetime assessment were introduced several decades ago and are partially considered in applicable standards. Such models based on a damage accumulation are easy to apply but they are considered to be either very imprecise or very conservative, while the conservatism reflects the necessity of large safety margins.
This paper summarize a few measures, which are suitable to improve the predictive quality of models based on a simple time-fraction rule. The proposed model is based on a synthesis of hysteresis loops for isothermal and non-isothermal conditions, concepts for consideration of cyclic softening or hardening during lifetime, concepts for dealing with internal back stresses, mean strains or stresses, and for accounting of creep-fatigue interaction. The latter is based on a so-called transition time concept, where the creep damage during dwell times partially attributes to the portion of fatigue damage, which in turn is determined from fatigue life curves for dwell time experiments. In addition, the model comprises a concept for the post-processing of transient FEM calculations and dealing with multiaxial loading conditions.
Since the essentials of the proposed method with the transition time concept were published more than 10 years ago, the listed modifications improve the benefit for daily engineering usage. Validation experiments provide evidence of the models predicting capabilities with acceptable uncertainty.