Fatigue behaviour assessment of C45 steel by means of energy-based methods

Abstract

Methods and techniques presented in literature for studying the fatigue behaviour of materials involve expensive, long lasting experimental campaigns and often a data analysis providing any information about damage localisation and occurrence. To avoid a time-consuming mechanical characterisation and to provide more information about where and when damage is occurring, the full-field experimental techniques such as infrared thermography allows the assessment of parameters that are related to the energy conversion to heat. In this work, the relationship between these thermal parameters (thermal methods) sentinels of the energy dissipated per cycle and the energy absorbed in every cycle is presented for the C45 steel undergoing stepwise loading fatigue tests.

On the relationship between mechanical energy rate and heat dissipated rate during fatigue for a C45 steel depending on stress ratio

“…A little pronounced slope change at a specific value of stress (approximately at 200 MPa, at R = 0.1 and at 300 MPa for the data at R = −1) characterizes the data trends of both Figure 14. These points according to the De Finis et al 45 seem to correspond to the endurance limit of the material. In addition, W p and ∆ E d 2 ω present roughly the same behavior in each case ( R = 0.1 and R = −1) and for any test, and then it is possible to conclude that ∆ E d 2 ω can be used to estimate the heat dissipated energy and then the area under the hysteresis loop.…”

“…In Figure 10A, the increase of during the test for all the samples is continuous, even if a small slope changing few stress levels before 200 MPa can be observed. At R = −1 (Figure 10B), data presents a marked slope variation in correspondence of 300 MPa due to inelastic point transition 23,44–46 …”

“…The hysteresis loops in Figure 5A are stretched while they become wider at a higher applied stress level and after the specific loading history provided by stepwise loading (Figure 5B), while the ratcheting is present and it becomes important at higher imposed mean stresses. Moreover, the Young's modulus variation can be also observed from the first to the last test as further sentinel of the ongoing damage 45,46 …”

“…The data processing to estimate the area under the hysteresis loop is based on the procedure presented in De Finis et al 45 In particular, for each time instant, the coupled values of the vectors σ and ε were evaluated computing the approximate integral of the vectors via the trapezoidal method with unit spacing. To perform the numerical integration on discrete data sets with N + 1 evenly spaced point, the approximation is provided by the following formula: where ε M and ε m are respectively the maximum and minimum strain of the vector ε .…”

“…Moreover, the Young's modulus variation can be also observed from the first to the last test as further sentinel of the ongoing damage. 45,46 The model adopted for simulating synthetic data runs quite well at higher imposed stress levels (no difference between S 1 sim and S 1 exp in Figure 5B) while it has some difficulties to represent the data at lower applied stresses (quite significant data scatter between S 1 sim and S 1 exp in Figure 5A). This can be explained by considering that when the load is in pure elastic range, it is difficult to fit the data by using a viscoelastic model.…”