Very high cycle fatigue of normalized carbon steels

https://doi.org/10.1016/j.ijfatigue.2005.05.016Get rights and content

Abstract

Fatigue properties in the very high cycle regime of normalized carbon steel with carbon content 0.61% C (Ck60) and 0.15% C (Ck15) have been investigated with the ultrasonic fatigue testing technique at load ratio R = −1 and cycling frequency 20 kHz. Both steels show a distinct change of slope in the SN curves at approximately 107 cycles. No specimen failed above 2.2 × 108 cycles although 25 specimens were stressed to over 109 cycles. Calorimetric measurements were used to evaluate the cyclic plastic strain amplitude at constant stress amplitude. Ck60 shows initial cyclic softening followed by cyclic hardening at stresses above and below the mean fatigue limit at 109 cycles. The lower the stress amplitude, the lower was the maximum plastic strain amplitude and the higher was the number of cycles when cyclic softening started. Cyclic hardening in Ck60 is seen to continue until 109 cycles. Ck15 undergoes cyclic softening mainly at stresses above the mean fatigue limit and specimens can fail at very low values (10−6) of the cyclic plastic strain. No sub-surface crack initiation was detected.

Introduction

Most textbooks assume that carbon steels have a fatigue limit. Japanese researchers [1], [2], [3], [4], [5], [6], [7], however, have discovered the meanwhile well-known phenomenon that high strength steels may fail at very high numbers of cycles due to cracks starting at inclusions. This leads to the question whether steels, in general, do not show a fatigue limit or if this effect is found only in steels, which are heat treated to reach high strength. Carbon steels without hardening treatment are used most frequently for structural applications and therefore this question is of great technical relevance.

The fatigue behaviour of normalized carbon steels is well documented in the literature [8], [9], [10], [11]. At room temperature (in the absence of dynamic strain ageing), constant amplitude cycling leads to the following processes: at low numbers of cycles, the material deformation is almost purely elastic, since dislocations in the ferrite grains and in the ferrite of the pearlite are blocked by carbon atoms. Then dislocation mobilization starts, and cyclic softening connected with a localization of plastic strain is observed. In ferritic–pearlitic steels, at low cyclic stresses, slip bands are restricted almost entirely to ferrite grains [10]. Cyclic softening is caused by the expansion of plastically deformed zones in which the motion of free dislocations contributes to the total plastic strain [8]. After exceeding the maximum plastic strain amplitude, the specimens work-harden with decreasing values of plastic strain until crack initiation [10]. The main differences between low and high cycle fatigue are that plastic deformation at high cycle fatigue stresses is restricted to some grains whereas others may deform in a purely elastic way, and plasticity occurs mainly in surface grains, which are less constrained than the interior ones [11].

In the present investigation, the high cycle fatigue behaviour of normalized low and high carbon steels and the involved cyclic plastic deformation have been studied. Low plastic strain amplitudes are evaluated using an indirect calorimetric method. Heat produced during ultrasonic cycling is measured and serves to determine the plastic strain using a method developed by Papakyriacou et al. [12]. The investigation should deliver answers to the questions if normalized carbon steels have a fatigue limit or not and if the evolution of fatigue damage at very high numbers of cycles is similar or essentially different from the behaviour well documented in the literature for the high cycle fatigue regime.

Section snippets

Lifetime measurements

Fatigue tests were performed in ambient air using ultrasonic fatigue testing equipment operating at a cycling resonance frequency of approximately 20 kHz. Loading of the specimens is fully reversed (R = −1), and in a sequence of pulses and pauses. Pulses of 25–100 ms (i.e. 500–2000 load cycles at 20 kHz cycling frequency) are interrupted by pauses of variable length to ensure that the temperature of the specimen remains below 40 °C. Vibration amplitudes of specimens’ ends are kept constant at

SN measurements

Fig. 3 summarizes the SN data determined for Ck60 (Fig. 3(a)) and Ck15 (Fig. 3(b)). As the cyclic stresses decrease, the numbers of cycles to failure increase. Arrows indicate specimens which did not fail within 109 cycles (runouts). Two straight lines are used to indicate 50% fracture probability in both sets of data. At high stresses, where all specimens fail, power law dependence of cyclic stresses and cycles to failure and lognormal distribution are assumed. At low stresses with failures

Conclusions

Ultrasonic fatigue testing technique was used to determine the SN data and cyclic plastic deformation of high carbon steel (0.61% C) Ck60 and low carbon steel (0.15% C) Ck15 in normalized conditions. The following conclusions may be drawn:

In contrast to the behaviour of several other steels, notably high strength steels, carbon steels Ck60 and Ck15 show a pronounced change in the slope of the SN curve at about 107 cycles. No specimen failed above 2.2 × 108 cycles although 25 specimens were

References (17)

There are more references available in the full text version of this article.

Cited by (51)

View all citing articles on Scopus
View full text