**3. Damage mechanisms in thermo-mechanical fatigue**

The phenomena during thermo-mechanical fatigue are influenced by a variety of processes within different temperature ranges during a thermal cycle, where especially at elevated temperatures the mentioned damage mechanisms can occur either individually or in mutual interaction. Thus the predominant damage processes are thermally activated gliding of

By their very nature, cyclic thermal loads appear with relatively low numbers of cycles in the low cycle fatigue (LCF) region so that the application of strain-based concepts (e.g. strain life diagrams etc.) is self-evident. If the loading is large enough to produce plastic strain, the number of cycles to failure is relatively low, in the order of less than 10,000 cycles. This total strain predominantly consists of plastic strain, which dominates the fatigue life. Widely used methods to determine the material behaviour are total strain based fatigue tests, whereby the resulting cyclic stress-strain hystereses are investigated. The resulting cyclic stress-strain curves as well as strain S-N curves are the basis for further lifetime evaluation where, depending on the material behaviour, softening or/and hardening effects can be

Depending on the application, further influences like temperature, mean strain, strain rate, atmosphere or aging-conditions must be considered. The components are primarily obtained by casting and defects such as pores, shrink holes or oxide inclusions ensued during this process have a negative influence on the lifetime. While these influences are extensively studied for isothermal conditions (Fagschlunger et al., 2006, Oberwinkler et al., 2010, Powazka et al, 2010), scientific understanding of the same for TMF is very limited. While LCF tests are always conducted under isothermal conditions, TMF tests are additionally loaded by thermal cycles, normally defined by a minimum and maximum

As TMF experiments are both very cost-intensive and time-consuming, it is often attempted in practice to estimate the fatigue life of components under thermo-mechanical load by means of more common isothermal LCF experiments. However, this approach may lead to non-conservative fatigue life estimates if the cyclic stress-strain behaviour or the effective damaging mechanisms under TMF loading differ considerably from the material behaviour under isothermal conditions. Furthermore LCF and TMF test results might not correlate due to differing methods used for recording and interpreting the deformation behaviour. In order to avoid misinterpretations it is crucial to pay close attention to the locally and temporally fluctuating temperature field, in particular when recording the TMF deformation behaviour. Thus a fundamental examination of the stress-strain behaviour and the predominant damage mechanisms under TMF conditions is crucial in order to enable accurate fatigue life predictions under thermo-mechanical fatigue loading. This approach can also clarify to which extent the employment of isothermal data is justified (Riedler et al.,

Differences may result from the fact that under TMF loading, as opposed to isothermal LCF loading, during every cycle a broad temperature range is experienced, in which the material properties can change and the material response may differ. The key to a comparison of LCF and TMF data thus lies in the evolution of the microstructure, whose integral behaviour is

The phenomena during thermo-mechanical fatigue are influenced by a variety of processes within different temperature ranges during a thermal cycle, where especially at elevated temperatures the mentioned damage mechanisms can occur either individually or in mutual interaction. Thus the predominant damage processes are thermally activated gliding of

**2. Similarities and differences between LCF and TMF** 

temperature, dwell time and heating/cooling rates.

reflected in the shape of the stress-strain hysteresis loops.

**3. Damage mechanisms in thermo-mechanical fatigue** 

found.

2004; Riedler, 2005).

dislocations at low temperatures, cyclic ageing at medium temperatures, and diffusion creep at high temperatures. However, both under IP (temperature and stress cycle are in phase) and OP (temperature and stress cycle are out of phase) TMF loading the microstructure evolution and the oxidation processes are more often than not dominated by the temperature range close to the maximum temperature. The maximum temperature occurs in the tensile stress region in case of an IP-TMF load, and in the compressive stress region in case of an OP-TMF load, see figure 1 (according to Löhe et al., 2004) and figure 2.

On the other hand concerning OP-TMF, crack initiation and growth are linked directly with the processes in the temperature range closest to the minimum temperature where tensile stresses prevail, and are only linked indirectly with processes that occur at the maximum temperature. Nevertheless this indirect influence can be even more distinctive than it would be in isothermal experiments. For example a layer of scale might build up as a result of oxidation which takes effect predominantly at high temperatures. This layer of scale is very brittle at low temperatures and thus causes early crack initiation and accelerated crack propagation under an OP-TMF load.

Fig. 1. Active damaging mechanisms during an OP- and IP-TMF cycle (Löhe et al., 2004)

Fig. 2. History of temperature, strain and time under OP-TMF loading (on the left) and IP-TMF loading (on the right)

Comparison of Energy-Based and Damage-Related

Pinto et al., 1995).

Constant temperature

Single/multiple

Table 1. Test matrix

**5.1 Influence of an in lying drilled hole** 

**5. Investigated influences** 

Fatigue Life Models for Aluminium Components Under TMF Loading 333

it is possible to identify actual TMF cycle shapes, which are translated to the test specimen as *industrial cycles* with certain phase shifts between thermal and mechanical strains (Engler-

The description of the creep, TMF and LCF testing rigs used for the following experiments as well as a detailed material characterisation can be found in previous papers (Riedler,

Firstly, it is important to clarify the governing damage mechanisms that occur in out-ofphase TMF cycles in cylinder heads. Therefore the tests on specimens were specifically designed to take the real circumstances in components as best possible into account – with the aim of using the derived models for lifetime estimations of TMF loaded components. Investigated influences are amongst others (see test matrix in Table 1) mean and local strains, cyclic and constant temperatures, dwell times, pre-aging and aging during service life, HCF-interaction, strain and temperature rates as well as the ratio of mechanical and thermal strain. Further single and multiple step creep tests have been carried out to take into account the stress relaxation phenomena. Additional tests in an argon atmosphere have finally enabled the isolation of the predominating damage mechanism in cylinder heads. All analyses are done in the manner of hysteresis loops, stress-cycle and plastic strain-cycle

Quasistatic - Creep LCF TMF HCF Metallographic

Stress relaxation Strain rate Temperature rate Porosity Porosity

Maximum temperature

Pre-aging Dwell time Dwell time Notch effect Microstructure Strain rate Pre-aging Pre-aging Pre-aging Chem. analysis

 Argon atmosphere HCF interaction Type of loading Striations Incr. step test Phase shift Stress amplitude

The aim of this study is to investigate the effects of an in lying drilled hole that is used for an improved quality of the temperature control device, presented in (Riedler & Eichlseder, 2004). The behaviour of the hollow drilled sample is calculated with the finite element method, tested with special LCF test series as well as analyzed by means of fractured surfaces on one wrought and one cast alloy. Whereas the influence on AlCuBiPb is visible, even though, marginally in respect on the lifetime behaviour analyzed with the Manson-Coffin-Basquin (Manson, 1954; Lemaitre & Chaboche, 1985; Basquin, 1910). Approach and the cyclic deformation behaviour analyzed with the Ramberg-Osgood (Ramber & Osgood, 1943) approach, at the Aluminium cast alloy AlSi7MgCu0.5 no difference can be ascertained between the test series of the hollow and solid samples (Riedler & Eichlseder, 2004).

Strain amplitude Strain constraining Frequency Grain size

Constant

spacing

controlled Mean stress Precipitations

temperature Fractured surface

Dendrite arm spacing

2005; Riedler & Eichlseder, 2004; Minichmayr et al., 2005; Minichmayr, 2005).

plots, lifetime diagrams and cyclic deformation behaviour diagrams.

step Mean strain Mean strain Dendrite arm

In lying hole Rigid clamped -

Constant temperature

Crack initiation in various Al-Si-alloys occurs preferentially at the interface between Almatrix and Si-phase. A meso-scale modelling of the microstructure under thermomechanical loading conditions shows stress concentration in the Al-Si-interface (Thalmair, 2009). The repeated recurring thermo-mechanical cycles cause micro-stresses and hence the preferred crack initiation.

Fig. 3. Crack tip of a TMF-specimen (left) and a thermo-shocked cylinder head (right) of AlSi8Cu3 (Thalmair, 2009)

TMF-test as thermo-shock test of a cylinder head shows very similar crack behaviour, where cracking of the eutectic phase and an inter-dendritic crack-propagation along the interfaces is observed, see figure 3.
