Validation of the multiaxial racetrack amplitude filter

Abstract Amplitude filters are a most important tool in practical fatigue analyses to manage their computational cost when, as usual, the measured load history is noisy, oversampled, too long, and/or contains too many non-damaging low-amplitude cycles or events. To reduce the calculation burden, such filters should not only eliminate noise and remove redundant oversampled data from the measured signal, but also neglect small amplitudes that do not cause fatigue damage. The veteran racetrack filter can perform all such tasks efficiently, however it is limited to uniaxial load histories. Multiaxial filtering techniques have been proposed in the past, however they fail to identify the most damaging events in several non-proportional histories, in addition to losing information on the load path shape. In this work, a new, fast, and efficient multiaxial version of the traditional racetrack filter is proposed to solve these issues, synchronously filtering complex loading histories while preserving all their significant reversals and equivalent ranges, and their load path shape as well, a most important feature for multiaxial fatigue analyses. Six and three-dimensional versions of the filter are proposed, respectively for invariant-based and critical-plane damage calculation approaches. The method allows not only the proper filtering of stress/strain histories at a given material point, but also of any history of multi-dimensional quantities such as forces, moments, and/or displacements acting at different points of a structure. The filter efficiency is evaluated from tension–torsion experiments in 316L stainless steel tubular specimens with challenging non-proportional path shapes.