High temperature materials subjected to elevated temperature have been shown to be sensitive to dwell times, giving an increased crack growth rate. The interaction between these dwell times and rapid cyclic loads have been shown to constitute a complex problem. Many models have been developed for 1D conditions, but the application to general 3D conditions has seldom been seen, although this is the most common case in most structures. In this paper a model for taking care of the interaction between these load modes in general 3D crack growth has been developed. The model uses 1D results for extension to general 3D, thus providing for local crack front evolution with a minimum of numerical simulations. The model has been implemented for usage with finite element calculations and several different tests are simulated and compared with experimental results for the nickel based superalloy Inconel 718 at 550◦C. The simulation results show crack shapes in agreement with experimental fracture surfaces and time to failure.
Funding agencies: Swedish Energy Agency; Siemens Industrial Turbomachinery AB; GKN Aerospace Engine Systems; Royal Institute of Technology through the Swedish research programme TURBO POWER
Vid tiden för disputation förelåg publikationen som manuskript
Safe life of gas turbines is always of major concern for manufacturers in order to ensure passenger safety and stable continuous power output. An increasing amount of resources have been put into research and development to assure that all safety aspects are covered in the design of new turbines and to ensure that enough frequent service intervals are scheduled to avoid complications. Many of these issues require good knowledge of material properties and of how to use these in the design process. Some of these relate to fatigue which is of major concern in all parts of a development programme. However, while some fatigue problems have been extensively studied, some have not. One example is crack growth with influence of dwell times at elevated temperature in combination with cyclic loading. Such loading conditions have been shown to give a different cracking behaviour compared to rapid cyclic loading, increasing the growth rate significantly with respect to the number of load cycles. Improved models for predicting this behaviour is therefore of major interest for gas turbine manufacturers, and could substantially increase the reliability. As a result, more research is needed in order solve these problems.
The work presented in this dissertation has focused on how to predict life under the above-mentioned circumstances. The materials used in high temperature gas turbine applications are often nickel-based superalloys, and in this work the most common one, Inconel 718, has been studied. Mechanical experiments have been performed under operation like conditions in order to receive material data for the subsequent modelling work. The modelling approach was chosen such that the underlying physics of the dwell time cracking have been incorporated on a phenomenological basis, creating a model which can be physically motivated as well as used for industrial applications. The main feature of the modelling work has been to track material damage which is received from dwell times, how this interacts with cyclic loading and how it affects the crack growth rate, thus creating a load history dependent model.
The outcome of this work has resulted in a model which is both easy to use and which has shown to give good correlation to available experimental data. Key components such as calibration for cheap and easy parameter determination, validation on complex engine spectra loadings, three dimensional crack growth, overload influences, material scatter, thermo-mechanical fatigue crack growth and the impact of high cycle fatigue loadings, are all covered in the presented work, both as experimental findings and as continuous development of the modelling concept.
The dissertation consists of two parts. In the first an introduction with the theory and background to crack growth with dwell times is given, while the second part consists of 10 papers.