PSI - Issue 54

Jenny Köckritz et al. / Procedia Structural Integrity 54 (2024) 423–430 J. Köckritz / Structural Integrity Procedia 00 (2019) 000 – 000

426

4

Tests were performed for two different multiaxial load cases with proportional bending and torsion, see Table 1. The load case 1 is based on the loading of a critical weld in the cargo bicycle frame and consists of 68% bending moment and 32% torsion moment. It is applied with constant amplitude and =−1 and with the load collective as described in chapter 2.1. The sensitivity of numerical evaluation to different proportions of bending and torsion is tested by load case 2 with a higher torsion moment proportion of 51% at constant amplitudes and =−1 . T-joint specimens were tested under strictly alternating bending to gain material data for the numerical evaluation. Table 1. Overview of testes specimens, load cases and type of loading

Test series

Specimen Tube-plate Tube-plate Tube-plate

Load case

Loading

Proportion bending moment

Proportion torsion moment

Ⅰ

Multiaxial load case 1 Multiaxial load case 1 Multiaxial load case 2

Constant amplitude

68 % 68 % 49 %

32 % 32 % 51 %

Ⅱ

Load collective

Ⅲ Ⅳ

Constant amplitude Constant amplitude

T-joint

Pure Bending

100 %

-

The fatigue testing was conducted with a hydraulic cylinder with a maximum force of 25 kN, force-controlled by an MTS® control system. The multiaxial load was realized with a bearing lever. The cycles to failure for each test were derived during postprocessing at a deflection amplitude increase of 5%. Above 2 ∙ 10 6 cycles, a specimen was considered a runout. Due to a limited number of specimens, only the results of load case 1 were processed by the horizon method (DIN 50100-12, 2016). The remaining results were processed with to the pearl string method. S-N curves for a 50% survival certainty, scatter ranges and standard deviations were calculated for the experimental results on the basis of the assumption of a double logarithmic normal distribution (DIN 50100-12, 2016). 2.3. Finite element modelling The numerical fatigue assessment of the welds was performed with the HSS and EN method (Hobbacher, 2016) and the SWF method for all load cases with the finite element software Altair® and the solver OptiStruct®. Schematic representations of the weld modelling are displayed in Fig. 2. For the HSS method, the specimen were modelled with quad shell elements for bulk and weld material with an element size of 2 mm . The HSS and resulting life with FAT class 36 is calculated in postprocessing by exponential extrapolation according to Niemi et al. (2018). The EN method model is modelled with hexagonal solid elements and effective radii of 1 mm at the weld toes. The minimum element size at the weld toes is 0.15 mm . The fatigue life according to the EN method was calculated for the highest stress with the FAT class 71, cp. Hobbacher (2016). FAT classes for both HSS and EN were adjusted to represent a 50% chance of survival for comparability, utilizing a logarithmic standard deviation =0.3 (Hobbacher, 2016). For the application of the SWF method, the specimen was modelled similarly to the HSS model. A deviating weld seam position was applied, with the weld element attached at the edge of the shell tube and not at the recommended position, as shown in Fig. 2 (c) and (d). Such a placement allows simplified modelling in complex structures like the cargo bicycle frame. The divergent placement showed to have no great effect on the computed lifetime.

(a)

(b)

(c)

(d)

Fig. 2. Schematic representation of modelling for hot spot stress method (a), effective notch method (b), seam weld fatigue method with recommended weld placement (c) and seam weld fatigue method with adjusted weld placement (d)