Degree Name

Doctor of Philosophy


Department of Materials Engineering


This study was motivated by a desire to understand more about the fatigue properties of HRLC and HSLA steels and the manufacturing and metallurgical property variables which influence fatigue response. This knowledge was needed by the steel industry to improve the resistance of hot rolled strip steels to fatigue failure, and by the motor vehicle manufacturing industry to improve methods of design against fatigue failure in components made from these steels.

A review of the literature identified a fatigue design methodology, the local strain approach to fatigue design, which was most suited to accurate fatigue life prediction. A vital input to this design method was strain controlled fatigue test data. To provide these data, the fatigue properties of a group of ten steels, representative of the commercially available range of H R L C and HSLA strip steels available in Australia have been characterised in this study. The steel grades chosen were, two low strength steels, XA1010 and C1012W, four Australian HSLA Xtraform steels, and four Japanese steels.

Characterisation of fatigue properties was performed using the Ramberg-Osgood cyclic stress-strain relationship, and the Coffin-Manson cyclic strain-life relationship. The fit of these relationships was found to be quite good. However, an improved fit was obtained by using a modification to the traditional relationships proposed by Weng. These relationships also had the advantage of recognising the fatigue limit phenomenon. By characterising of the fatigue properties of the steels, the goal of providing industry with the needed input data for fatigue design, was satisfied.

While, other researchers established that a steel grade could be characterised by sampling from a single position in a coil, and from a single heat, the effects of strip thickness, skin passing, and anisotropy of properties on fatigue response were unclear and have consequently been examined in the present study. Thus, integral with the process of characterising the fatigue properties of the steels, was the testing of several variants of each steel, which were chosen to represent various treatments levels of these manufacturing variables. In this way, the effects of these variables on fatigue properties were examined.

The test results were arranged into a series of balanced factorial experimental designs to facilitate linear regression statistical analysis. The Ramberg-Osgood relationship was used as the basis of the cyclic stress-strain analysis, and an alternative description to the Coffin-Manson formulation was developed for the cyclic strain-life relationship. A notional (Neuber) stress-life analysis, which compared steels under similar service conditions was also employed.

The effect of strip thickness was examined in 3.2 and 5.0 mm thickness C1012W and XA1010 steels. The lack of a significant thickness effect in either the cyclic stress-strain or strain-life behaviour in these steels suggested that thickness does not have a large effect on the properties of hot rolled strip steels.

Test piece orientation had a significant effect on cyclic stress-strain behaviour. In each case, the cyclic strength of the transverse test pieces w a s higher than that of the longitudinal test pieces. There was no significant effect of orientation on cyclic strain-life behaviour, however, when compared on the basis of notional stress amplitude, the transverse test pieces also had longer lives.

While no significant effect of skin passing was observed on cyclic stress-strain behaviour in the Xtraform steels, a significant effect was observed in cyclic strain-life behaviour. The effect was to slightly reduce fatigue lives in the two lower strength grades. However, in the high strength grades, while fatigue lives were reduced by skinpassing at higher stress/strain amplitudes, at low stress/strain amplitudes fatigue life was improved. Consequently, the conjecture that skin-passing is beneficial to fatigue life has been demonstrated only for Xtraform 400, although implied for Xtraform 500.

Equivalent Xtraform and Japanese steels were compared on the basis of notional stress-life response, however, no steel w a s clearly superior to its counterpart.

Thus this part of the study achieved the goal of developing and applying methods of statistical analysis which could discriminate between the fatigue responses of. similar steels, and provide a systematic analysis of the influence of manufacturing variables on fatigue response.

Since uncertainty also existed about the influence of hardness, strength, composition, and microstructure on fatigue response, these effects were examined. The goal achieved in this part of the study was to develop methods to predict difficult to obtain fatigue properties from readily available measurements such as the non-fatigue variables listed above. These relationships would be of value both in alloy design and in design against fatigue.

Relationships with the parameters which define both the Ramberg-Osgood, and W e ng forms of the cyclic stress-strain relationship, including a number of reputedly successful relationships, were examined. None of these potential relationships with the non-fatigue variables, were found to hold. Relationships between 0.2% cyclic proof strength and hardness, tensile strength, and proof strength were, however, found. It w a s also established that stress amplitude at 0.1% and 0.4% plastic strain amplitude could be effectively related to a combination of composition variables and proof strength. This pair of relationships then allowed the prediction of the cyclic stress-strain relationships in the absence of data derived from fatigue tests.

Despite the reported relationships between the parameters of the Coffin-Manson form of the cyclic stress-strain relationship and strength, in the present study none of these relationships, nor those with the other available non-fatigue variables, proved to hold. Relationships were, however, found with elastic, plastic and total strain amplitudes at various lives allowing the cyclic strength and cyclic ductility relationships to be estimated.

For the fatigue strength line, relationships were found at short and long endurances. At short life, the relationship w a s with tensile strength, while at long life, the relationship was with a combination of composition variables and hardness. Additional relationships were used to identify the locus of the fatigue ductility line. Transition life could be estimated from a relationship with yield strength, and the strain at this life estimated from the fatigue strength relationship. The fatigue ductility relationship could then be estimated either by applying the average value for fatigue ductility exponent, or by estimating the locus of a point at one million cycles, from relationships with total and elastic strain, at this life. This estimation method, while not particularly accurate, w as found to be quite acceptable where no fatigue data were available experimentally.

In summary, this study has characterised the fatigue properties of the formable hot rolled strip steels, available in Australia. In addition, methods have been developed, and used to determine the effects of the manufacturing variables, thickness, orientation to rolling direction, and skin-passing, on fatigue performance. Finally, relationships have been developed between fatigue properties and the more readily determined material properties, strength, hardness, composition and microstructure. These relationships provide insight into the best choices for design of fatigue resistant alloys, and allow the prediction of fatigue properties in the absence of fatigue test data.