Page 39 - Biaxial Multiaxial Fatigue and Fracture

24 G. SAVAIDIS ET AL

spot stress approach, see e.g. Bovet-Griffon et al. [5] and Fayard et al. [6]. Further experience
gathered on thin-walled plane structures of commercial vehicle components under
proportional constant amplitude normal stresses [7] revealed that fatigue analysis based on hot
spot stresses is capable of handling such components.
The local stress approach to fatigue of welded joints represents an alternative to the hot
spot stress approach. Its main advantage can be seen in the existence of an experimentally
verified universal constant amplitude local stress-life curve for steel welds. However, the
determination of local stresses causes a clearly increasing numerical effort.
This paper deals with the critical plane approach in connection with numerically
determined hot spot stresses; particularly, its application to nonproportional variable
amplitude loading is shown and discussed. In the latter part of the paper the application of the
local stress approach is shown. A comparison of fatigue lives gained from hot spot stresses,
local stresses and experimental investigation, respectively, will close the paper.

COMPONENT AND LOAD CONFIGURATION
The component under investigation is a thin-walled tube joined at both ends to forged arms by
fillet welding. This component serves to stabilise the driver’s cab of trucks. The mechanical
behaviour and the fatigue life of the component shown in Fig. 1 are calculated and compared
with corresponding test results.

Fig. 1. Component under investigation

Figure 2 shows the test rig with two actuators at the top left-hand side of the component
capable for introducing lateral forces and forces resulting from suspension. Figure 3 shows
the normalised force-time sequences Fy(t) and F,(t) that have been measured using a prototype
component during driving on a test track. The test track driving program includes quasi-static
manoeuvres like cornering, braking during cornering and braking at straight-driving as well as
straight-driving over rough road segments like potholes, washboards, belgian block, country
roads and rough highways. The corresponding spectra of the normalised forces versus
cumulative frequency of occurrence evaluated in accordance with the level crossing counting
method provide approximately straight-line shaped distributions and cumulative number of
cycles of approximately 3500 for Fy(t) and 2300 for F,(t) for one block. The running time for
the experimental simulation of one block on the test rig amounts to approximately 28 minutes.
These load components predominantly cause bending (Mx, M,) and torsion (My) which
interact nonproportionally.