PSI - Issue 38

Matthias Hecht et al. / Procedia Structural Integrity 38 (2022) 251–259 Matthias Hecht et al. / Structural Integrity Procedia 00 (2021) 000–000 3 Real damage sums for bonded joints under uniaxial load range between 0.03 ≤ real ≤ 0.17 [10]. In case of an unknown location of the knee point, the damage accumulation Palmgren-Miner modified according to Cortan and Dolan was applied. A modification of the damage accumulation according to Liu and Zenner, where higher damage is assigned, did not significantly reduce the scatter of the damage sum [11]. Damage sums under multiaxial loading are presented in this paper, too. 1.2. Mean stress transformation In the case that individual cycles have different -ratios, each cycle must be transformed to a common mean stress resp. -ratio such as = -1, as mentioned above, before a partial damage can be assigned to each cycle via Palmgren Miner [12]. The mean stress sensitivity is defined by [13] = , =−1 − , =0 , =0 (5) and can be visualized in the Haigh diagram. 2. Test procedure 2.1. Specimen preparation The hollow cylinders to be used are made of S355 steel, Figure 1 (b). Before adhesive bonding, these are first blasted with glass spheres and then coated with silicate using sandblast coating. Before and after each blasting process, the adherents are cleaned using n-heptane or acetone. Subsequently, the outer diameter , the inner diameter and the specimen length are measured. In the next step, the specimens are clamped in a manufacturing device in a way that an adhesive layer thickness of a = 0.3 mm is obtained. Then the adhesive is applied to one half of the specimen in the extended manufacturing device. The device is then closed and the specimens are hardened in the device at 180 °C in an oven for 30 min. Temperature control during hardening is achieved with a thermocouple in the immediate surroundings of the adhesive layer. The adhesive used in this process is a one-component, thermosetting, toughened structural adhesive based on epoxy. After the hardening process, the specimens are slowly cooled down with the oven door open. In order to obtain a defined adhesive layer geometry and to avoid larger scatter in the measured fatigue strength, the bead on the outer side is removed mechanically. Afterwards, the actual adhesive layer thickness is determined from the total sample length and the length of the two hollow cylinders. Finally, the specimens are stored at 23 °C and 50 % relative humidity for at least ten days before the test. 2.2. Test rig Tests were carried out under pure tension, pure torsion and multiaxial loading with ( = 90°) and without phase shift ( = 0°) under constant amplitude loading. Tests with variable amplitudes were performed under pure tension, pure torsion and multiaxial loading without phase shift ( = 0°), Table 1. The specimens tested under load control in a servo-hydraulic testing machine, failed in the adhesive bond, Fig 1 (d). The testing frequency of = 10 Hz was chosen to exclude temperature rising in the bond during the test [10, 14], which was also measured by a thermo couple at a single test. The tests under constant amplitudes are performed at the LWF of Paderborn University and the tests under variable amplitudes at Fraunhofer LBF, Figure 1. In the tests at Fraunhofer LBF the axial load is applied by a hydraulic cylinder attached to the bottom and the torsional torque through a lever gear, which is also actuated by a hydraulic cylinder. In the tests at LWF of Paderborn University, the axial and torsional load is applied on a MTS system by a biaxial servo-hydraulic cylinder, which is also attached to the bottom of the testing machine. 253