Date of Award


Degree Type


Degree Name

Doctor of Philosophy


Materials Science and Engineering

Major Professor

Peter K. Liaw

Committee Members

Raymond A. Buchanan, John D. Landes, Charlie R. Brooks


Extensive smooth-bar low-cycle-fatigue (LCF) and fatigue crack growth (FCG) experiments on three solid-solution-strengthened superalloys HASTELLOY X, HAYNES 230, and HAYNES 188 have been conducted at 816 and 927 C. Limited tests were run at 649 C, 871 C, and 982 C to study the temperature effect. The LCF tests were performed under a total-strain-range-control mode at Solar Turbines Inc., Metcut Research Inc., and The University of Tennessee (UT). The FCG tests were done under both the constant-load-range and constant-stress-intensity-factor-range modes at Westmoreland Testing Inc. and UT. Various hold times were imposed at the maximum strain or load in both the LCF and FCG tests, respectively, to investigate the hold-time effect. For the LCF tests, the influence of the total strain range and hold time on the cyclic stress response was determined at temperatures ranging from 816 to 982 C.

At the temperatures considered, the HASTELLOY X alloy exhibited initial cyclic hardening, followed by a saturated cyclic-stress response or cyclic softening under LCF without hold times. For LCF tests with hold times, however, the alloy showed cyclic hardening, cyclic stability, or cyclic softening, which is closely related to the test temperature and the duration of the hold time. It was also observed that the LCF life of the X alloy considerably decreased due to the introduction of strain hold times. Generally, a longer hold time would result in a greater reduction in the fatigue life. For the tests without hold times, the test temperature seems to have little influence on the fatigue life of the X alloy at the test temperatures from 816 to 927 C. However, when the test temperature increased to 982 C, the fatigue life clearly shortened. The effects of heat-to-heat variation on the cyclic stress response were illustrated. A parameter based on the hysteresis energy was proposed to rationalize the LCF life data with and without hold times. In general, the fatigue life of HAYNES 230 alloy decreased as the temperature increased. However, at total strain ranges higher than 1.0% and without a hold time, the LCF life was longer at 927 C than at 816C. This “abnormal” behavior was found to result from the smaller plastic strain amplitude at half-life at 927 C than that at 816 C. The introduction of a hold time led to a decrease in the fatigue life. At both 816 and 927 C, the material exhibited a cyclic hardening/softening behavior at higher total strain ranges and a cyclic hardening/saturation behavior at lower total strain ranges. An increase in the temperature and/or the introduction of a hold time decreased the hardening rate and increased the softening rate. The introduction of a hold time and/or the increase of the test temperature progressively changed the fracture mode from the transgranular to mixed trans/inter-granular, then to intergranular feature. Within the two phases of the fatigue process, crack initiation was more severely influenced by the change of the hold time and/or temperature.

For the LCF of HAYNES 188 alloy, in the total-strain ranges used in this investigation, increasing the test temperature from 816 to 982 C shortened the fatigue life. This effect is especially clear at total-strain ranges smaller that 1.0%. Introducing hold times at the maximum tensile strain is found to lead to a significant reduction in the fatigue life. The HAYNES 188 alloy can exhibit cyclic hardening, cyclic hardening followed by softening or the saturated stress response, and cyclic softening during LCF deformation, which is closely related to the test temperature, the imposed total strain range, and the hold time. In addition, the HAYNES 188 alloy shows the heat-to-heat variation in the cyclic stress response curve.

The FCG data were analyzed with an emphasis on hold-time and temperature effects. For both alloys, the crack grew faster at a higher temperature. It was also noted that the introduction of a hold time at the maximum load led to an increase in the cyclic crack-growth rate. The longer hold time gave the faster crack-growth rate, which was related to the gradual transition from transgranular to intergranular cracking. The crack-growth rates in the fatigue and creep tests were correlated with the stress-intensity-factor range, K, and the stress-intensity factor, K, respectively.

The crack-propagation rates in the hold-time tests were predicted from the crackgrowth rates obtained from both the fatigue and the creep crack-growth tests, using a semi-empirical linear summation model. Crack-growth-rate predictions reproduce most of the characteristics observed experimentally. The crack-growth-rate data obtained from the FCG testing under load-range and stress-intensity-factor-range control modes were compared.

In addition, time-dependent fracture-mechanics parameters, C, Ct, and (Ct)avg, were applied to correlate the crack-growth rates. For both alloys, the fatigue-cracking path was mainly transgranular at 816 and 927 C. The cracking path became dominantly intergranular if the hold time increased to 2 min. , indicating that the timedependent damage mechanisms, creep and/or oxidation, were in control. When the time-dependent damage dominated (temperature 816 C and hold time 2 min. ), the crack-growth rates can be correlated with Ct or (Ct)avg parameters. The Ct and (Ct)avg parameters were capable of consolidating data from different temperatures and alloys.

The fracture surfaces of both LCF- and FCG-tested samples were examined with the scanning-electron-microscopy (SEM). The tests in this study were conducted at high temperatures in air. Therefore, the fracture surfaces were frequently covered with a dark layer of oxides, making the fracture features difficult to identify. To overcome this problem, for the LCF testing, the failed samples were cut longitudinally through the fracture surfaces, and the sections were observed to locate secondary cracks. By combining the fractographic and metallographic results, the crack initiation and propagation for all tests were successfully investigated. However, for the FCG-failed samples, the major secondary cracks on specimen surfaces were not available. An oxide-stripping technique has been developed to overcome the oxidelayer problem. The technique consists of two steps. The sample is first boiled in a potassium permanganate solution for 1 hr, and then electrolytically cleaned in an alkaline solution for 5 min. with the sample as the cathode. Except for dislodging the carbides, the technique developed was capable of removing the oxides completely without altering the fracture-surface morphology.

Files over 3MB may be slow to open. For best results, right-click and select "save as..."