Kurzfassung I VAL 5 Junior Prize G. Curti

MODAL DRIVEN LOAD TAILORING FOR EFFICIENT VIBRATION FATIGUE ASSESSMENT ON NON-STATIONARY PROCESSES BASED ON FATIGUE DAMAGE SPECTRUM

Giulio Curti, Università degli Studi di Perugia, Italy

Vibration fatigue represents a significant challenge in the design and operation of a range of mechanical systems. It is of paramount importance to accurately assess vibration fatigue to guarantee the reliability and longevity of components. However, traditional methods of vibration fatigue assessment can be time-consuming and computationally expensive when a time-domain approach is employed. In certain instances, frequency domain methods represent the optimal solution, offering the most expedient results. However, as signals become non-stationary, these latter approaches are found to be quite unreliable. This research project is concerned with the efficient representation of damage generated by a non-stationary signal. The approach presented is a novel tailoring method that aims to streamline the process of vibration fatigue assessment by extracting the most relevant information from the vibration data.

The proposed method is based on the Fatigue Damage Spectrum (FDS), a powerful tool for analyzing vibration loads. By identifying the most damaging events within the vibration signal, the tailoring method generates a shorter, tailored signal that retains essential information for fatigue assessment while significantly reducing data volume.

The tailoring method involves several key steps:

1. FDS Analysis: The preprocessed vibration data is analyzed using the FDS method. This involves decomposing the signal into a series of single-degree-of-freedom systems and calculating the corresponding displacements.
2. Event Identification: The responses analysis identifies the most damaging events within the vibration responses. These events are characterized by their frequency, amplitude, and duration.
3. Tailored Signal Generation: A new, shorter signal is generated by extracting and composing only the most damaging events from the original signal.
4. Damage Assessment: The tailored signal is then used to assess the fatigue damage in the component. This can be done using traditional time based fatigue analysis methods, such as the rainflow counting method.

Furthermore, a supplementary procedure developed as a means of attaining independence from the slope coefficient of the Wöhler curve, used to assess fatigue damage. It is inevitable that a shorter signal will lack the damaging cycles that are damaging relevant for low values of the slope coefficient.

The proposed complementary procedure entails conducting the damage assessment using a load spectrum that is generated as an envelope of two. One of these is obtained through a time-domain approach on the synthesized time history, which was previously obtained. The other is obtained through a frequency-domain approach using the averaged PSD, which was obtained from the original measurement.

The developed method was applied to a real measured signal from a train boogie, resulting in a significant reduction in both time history length (95%) and computational time (over 90%). This demonstrates the potential of the proposed tailoring method.

In conclusion, the tailoring method presents a promising approach for streamlining vibration fatigue assessment. By reducing computational time and improving accuracy, the method can deal efficiently with non-stationary signals. However, this production can also be seen as a starting point for further research into the topic of non-stationary signals. Parts of such procedures might be pursued to further characterize non-stationary sections of measurements, which are still to this day a great issue in the industry.