Page 182 - Biaxial Multiaxial Fatigue and Fracture

Sequenced Axial and Torsional Cumulative Fatigue: ... 167

percent composition of the superalloy was: e 0.002 sulfur, 0.003 boron, < 0.005 phosphorus,
0.052 lanthanum, 0.09 carbon, 0.35 silicon, 0.8 manganese, 1.17 iron, 14.06 tungsten, 22.1 1
chromium, 22.66 nickel, with the balance made up of cobalt.
AI1 experiments were performed on thin walled tubes with nominal gage section dimensions
of 26 mm outer diameter, 22 mm inner diameter, 41 mm straight section and 25 mm gage
length. The interior surfaces of the tubes were honed in an attempt to preclude crack initiation
on the inner surface of the specimen. Outer surfaces were polished with final polishing
direction parallel to the specimen axis. Further details on the specimen geometry and
machining specifications can be found in Ref. [8]. The baseline axial and torsional fatigue
lives for this material, specimen geometry, and test temperature can be found in Ref. [I].
The specimens were heated to 538°C with an induction heating system. A11 specimens were
subjected to sequential constant amplitude fatigue loading under strain control. A
commercially available, water-cooled, biaxial, contacting extensometer with a 25 mm gage
length, designed specifically for axial-torsion testing, was used. The loading actuator that was
not being used for fatigue strain cycling (the torsional actuator during axial cycling or the axial
actuator during torsional cycling) was maintained in load control at zero load. This procedure
allowed strains to accumulate in the zero load direction. During the axial strain cycling
segments, relatively small mean strains in the load controlled torsional axis were observed.
However, the torsional strain cycling segments always showed increasing mean axial strains.
When torsional strains were applied in the first segment, the magnitudes of these axial strains
were recorded and then electronically set to zero prior to commencing the second loading
segment.
The specimen failure criterion programmed into the testing software was a 10% drop in the
measured load in the strain controlled direction. Five experiments were terminated due to a
controller interlock. Details on the testing system and test control procedures can be found in
Ref. [l].

TEST MATRIX

The test matrix for this study is shown in Table 1. Seventeen different combinations of low
amplitude followed by high amplitude, two load level experiments were performed. The
loading sequences were axial followed by axial (axiallaxial), torsion followed by torsion
(torsionltorsion), axial followed by torsion (axialltorsion), and torsion followed by axial
(torsionlaxial), with at least four different, first load level life fractions imposed in each
combination. A fifth life fraction was imposed in the torsion/torsion subset. One torsional
experiment was repeated as a cursory check on the expected specimen-to-specimen variability
in fatigue life. This summed to a total of 18 tests performed for this study. Table 1 also
contains the stress range and mean stress at half-life for each load level, the number of cycles
imposed, and the final crack orientation.