Ultrasonic Fatigue Testing in the Tension-Compression Mode

Fatigue damage of structural materials is strongly connected with industrialization and mainly with use of steam engine and steam locomotives for railway transport, where a lot of metal components, mainly iron based, have been used and had to withstand various types of cyclic loading. One of the earliest tests was done by Albert (Germany 1829)¹ on welded chains for mine hoists. The loading frequency was 10 bends per minute, and the maximal tests recorded reached 100,000 loading cycles¹. Another important work was carried out by William Fairbairn in 1864. Tests were performed on wrought-iron girders with use of a static load, which was lifted by a lever and then dropped causing vibrations. The girder was loaded with gradually increasing loading stress amplitude. After reaching several hundred thousand cycles on various loading stress amplitudes, in the end the girder failed after just about five thousand loading cycles at a loading amplitude of two fifths of the ultimate tensile strength. The first comprehensive and systematic study of the influence of repeated stress on structural materials was done by August Wöhler in 1860-1870¹. For these tests, he was using torsion, bending, and axial loading modes. Wöhler designed many unique fatigue testing machines, but their disadvantage was low operation speeds, for example the fastest rotating bending machine operated at 72 rpm (1.2 Hz), thus completion of the experimental program took 12 years¹. After performing these tests, it was considered that after reaching a loading amplitude at which the material withstands 10⁷ cycles, the fatigue degradation is negligible and the material can withstand an infinite number of loading cycles. This loading amplitude was named the "fatigue limit" and became the main parameter in industrial design for many years^2,3.

Further development of new industrial machines, which required higher efficiency and cost savings, had to provide the possibility of higher loading, higher operation speeds, higher durations, and high reliability with low maintenance requirements. For example, components of the high-speed train Shinkanzen, after 10 years of operation, have to withstand approximately 10⁹ cycles and failure of a main component can have fatal consequences⁴. Furthermore, components of jet engines often operate at 12,000 rpm, and components of turbo blowers often exceed 17,000 rpm. Those high operation speeds increased requirements for fatigue life testing in the so-called ultra-high cycle region, and to assess if the fatigue strength of a material could be really considered constant for more than 10 million cycles. After the first tests performed by exceeding this endurance, it was obvious that fatigue failures can occur even at applied stress amplitudes lower than the fatigue limit, after a number of cycles much more than 10⁷, and that the damage and failure mechanism could be different from the usual ones⁵.

Creating a fatigue test program aimed at investigating the ultra-high cycle region required the development of new testing devices to strongly increase the loading frequency. A symposium focused on this topic was held in Paris in June 1998, where experimental results were presented which were obtained by Stanzl-Tschegg⁶ and Bathias⁷ at 20 kHz loading frequencies, by Ritchie⁸ with the use of 1 kHz closed loop servo-hydraulic testing machine, and by Davidson⁸ with a 1.5 kHz magneto-strictive testing machine⁴. From that time, many solutions were proposed, but still the most commonly used machine for this kind of test is based on the concept of Manson from 1950 and uses frequencies close to 20 kHz⁹. These machines exhibit a good balance between strain rate, the determination accuracy of the number of cycles, and the time of the fatigue test (10¹⁰ cycles are achieved in approximately 6 days). Other devices were able to provide even higher loading frequencies, like the one used by Girald in 1959 – 92 kHz and Kikukawa in 1965 – 199 kHz; however, these are rarely used because they create extremely high deformation rates and, since the test lasts for only few minutes, a remarkable error in the cycle counting is expected. Another important factor limiting the loading frequency of resonance devices for fatigue testing is the size of the specimen, which is in direct relation with the resonance frequency. The larger the requested loading frequency, the smaller the specimen. This is the reason why frequencies above 40 kHz are rarely used¹⁰.

Since the displacement amplitude is usually limited within the interval between 3 and 80 µm, ultrasonic fatigue testing can be successfully applied on most metallic materials, though techniques for the testing of polymeric materials such as PMMA¹¹ and composites¹² were also developed. Generally, ultrasonic fatigue testing is possible to perform in modes of axial loading: tensile – compression symmetrical cycle^13,14, tension – tension cycle¹⁵, three-point bending¹⁵, and there are also a few studies with special modifications of the system for torsion testing^15,16 and biaxial bending¹⁷. It is not possible to use arbitrary specimens, because for this method, the geometry is strictly related to achieving the resonance frequency of 20 kHz. For the axial loading, several types of specimens have been commonly used, usually with an hour-glass shape with a gauge length diameter from 3 to 5 mm. For the three-point bending, thin sheets are commonly used, and for other methods special types of specimens are designed, according to the method type and testing conditions. The method was designed for evaluation of fatigue life in the high and ultra-high cycle region, and this means that at 20 kHz loading, a million cycles is obtained in 50 s; therefore, this is usually considered the bottom limit of loading cycles which can be investigated with reasonable accuracy, with respect to the number of cycle determination. Each specimen has to be harmonized with the ultrasonic horn by changing the specimen's mass to provide the right resonance frequency of the system: ultrasonic horn with specimen.