ABSTRACT

Engineering vehicles, such as excavators and wheel loaders, are widely used in the fields of transportation, mining, and construction. Due to the harsh working environments and bumpy roads, the components of engineering vehicles are often subjected to complicated random loads during their full working life. The fatigue failure is one of the major failure modes of these structures (Wang et al. 2011, Rychlik et al. 1996, Wang et al. 2012). In order to conduct proper fatigue design, it is important to determine the load spectrum that considers real service loads. In general, fatigue life assessments are usually based on the extrapolation of the supposed worst load conditions. Extrapolation of a measured load to a design life has been solved by several authors (Johannesson 2001, Bruni et al. 2003, Moriarty 2012). The simplest method is to repeat the measured load block several times until fatigue failure occurs. This method is disadvantageous because only the measured load cycles will appear in the fatigue test. Johannesson (2006) proposed an efficient extrapolation method to repeat the measured load block, and modified the largest maxima and lowest minima in each block to extrapolate more large cycles than the observed ones. In this method, only the maxima above a high load level and the minima below a low load level are randomly regenerated by the Generalized Pareto Distribution (GPD) based on the POT, but the loads between the thresholds are not processed. Compared with the actual load time history, the repeated load blocks have the lower randomness of the load. Meanwhile, under a given threshold, the exceedances in each load block are same. Because the available loads above a threshold are insufficient, it will lead to a bad fit for the GPD.