Vibration-fatigue analysis deals with the material fatigue of flexible structures operating close to natural frequencies. Based on the uniaxial stress response, calculated in the fre- quency domain, the high-cycle fatigue model using the S-N curve material data and the Palmgren-Miner hypothesis of damage accumulation is applied. The multiaxial criterion is used to obtain the equivalent uniaxial stress response followed by the spectral moment approach to the cycle-amplitude probability density estimation. The vibration-fatigue analysis relates the fatigue analysis in the frequency domain to the structural dynamics. However, once the stress response within a node is obtained, the physical model of the structure dictating that response is discarded and does not propagate through the fatigue-analysis procedure. The structural model can be used to evaluate how specific dynamic properties ( e.g. , damping, modal shapes) affect the damage intensity. A new approach based on modal decomposition is presented in this research that directly links the fatigue-damage intensity with the dynamic properties of the system. It thus offers a valuable insight into how different modes of vibration contribute to the total damage to the material. A numerical study was performed showing good agreement between results obtained using the newly presented approach with those obtained using the classical method, especially with regards to the distribution of damage intensity and critical point location. The presented approach also offers orders of magnitude faster calculation in comparison with the conventional procedure. Furthermore, it can be applied in a straightforward way to strain experimental modal analysis results, taking advantage of experimentally measured strains.
COBISS.SI-ID: 15513115
Random vibrations induce damage in structures, especially when they are operating close to their natural frequencies. The stationarity of the input excitation is one of the fundamental assumptions required for frequency-domain fatigue-damage theory. However, for real applications, excitation is frequently non-stationary and the identification of this non-stationarity is not easy. This study researches run-tests to identify the index of non-stationarity. Further, using excitation signals with different rates of amplitude-modulated non-stationarity, the index of non-stationarity is experimentally and theoretically researched with regards to the fatigue life. The experimental research was performed on a flexible structure that was excited close to a natural frequency. The experimental fatigue life is compared to the theoretical fatigue life under the stationarity assumption. The analysis of the experimental results reveals a close relation between the identified non-stationarity in the excitation signal and the fatigue life of the structure. It was found that amplitude-modulated non-stationary excitation results in a significantly shorter fatigue life if compared to a similar level of stationary excitation.
COBISS.SI-ID: 15586587
Structural damage in mechanical components is frequently caused by high-cycle vibration fatigue. The non-linearities, frequently observed in real structures at increased excitation levels, significantly influence the damage accumulation. As the modal analysis bases on linear theory, the non-linearities are hard to include. Based on a new experimental identification of the non-linearities, this research proposes the corrected linear damage-accumulation estimation. With the proposed correction, the linear modal analysis is used for damage estimation of structures with non-linearities. The proposed approach is applied to a real-life case of steel-sheet attached with rivets. Several samples are exposed to an accelerated vibration-fatigue test with increasing and also decreasing excitation levels. It is shown that with the experimentally identified non-linearity correction, the numerical fatigue life-time was within the 10% of the experimentally identified life-time. Experimentally, it was shown that rivets same by design, but produced by different manufacturers, have a significant difference in the fatigue life-time; this difference was clearly identified with the proposed correction to the linear damage-accumulation estimation. Further, the frequency response function based identification of the non-linearity can be identified before the structure is exposed to fatigue loads resulting in new possibilities of vibration-fatigue analysis of non-linear systems.
COBISS.SI-ID: 14413851
Random vibration excitation is a common cause of failure, especially if natural dynamics is excited. The high-cycle vibration-fatigue analysis typically requires the structural dynamics analysis, the response analysis and the fatigue analysis. The material parameters (S-N curve) are obtained at uniaxial stress state. However, in real structures the stress state is rarely uniaxial and the direct application of the S-N curve is difficult. The stress tensor is reduced to a more manageable representation using a multiaxial criterion. In this study, maximum normal stress, maximum shear stress, maximum normal-and-shear stress, C-S criterion, Projection-by- Projection and the Preumont and Piéfort criterion for multiaxial stress state are compared theoretically and experimentally. The crack location and the time-to-failure were compared. The time-to-failure was found relatively accurate with all multiaxial criteria; however, the crack-location estimation was found not to be accurate enough for either of the compared criteria. The study proves the applicability of the vibration-fatigue analysis procedure on real vibrating structures with rich structural dynamics. Random vibration excitation is a common cause of failure, especially if natural dynamics is excited. The high-cycle vibration-fatigue analysis typically requires the structural dynamics analysis, the response analysis and the fatigue analysis. The material parameters (S-N curve) are obtained at uniaxial stress state. However, in real structures the stress state is rarely uniaxial and the direct application of the S-N curve is difficult. The stress tensor is reduced to a more manageable representation using a multiaxial criterion. In this study, maximum normal stress, maximum shear stress, maximum normaland-shear stress, C-S criterion, Projection-by-Projection and the Preumont and Piéfort criterion for multiaxial stress state are compared theoretically and experimentally. The crack location and the time-to-failure were compared. The time-to-failure was found relatively accurate with all multiaxial criteria; however, the crack-location estimation was found not to be accurate enough for either of the compared criteria. The study proves the applicability of the vibration-fatigue analysis procedure on real vibrating structures with rich structural dynamics.
COBISS.SI-ID: 14504475
In order to estimate a structureʼs fatigue life when excited with an acceleration profile the fatigue parameters must be known. However, materialʼs fatigue exponent and fatigue strength are not always readily available, especially for complex structures that include riveted or welded joints for which additional fatigue tests are needed. This study introduces a new fatigue-parameter assessment method based on random vibration loading and its application to a blind-hole rivet joint that diminishes the need for additional fatigue tests. The presented procedure requires a simple experimental setup; however, a more extensive analysis of the experimental results is necessary. The method of fatigue parameter assessment is presented and applied on real, experimentally obtained data from vibration tests of rivet-joint specimens, excited with a random base-vibration load in the frequency range of a single natural frequency. Special attention was given to the modelling of the rivet joint and the uncertainties arising from the riveting process were considered. With the presented procedure it is possible to obtain the fatigue parameters solely from the results of random-vibration testing with different acceleration profiles and therefore diminishing the need for additional classic fatigue tests. The obtained fatigue parameters indirectly include the stress concentration factor and the damping-loss- factor increase during the damage accumulation. Additionally, by applying the random-vibration load the influences of the natural-frequency shift and the small nonlinearities of the structure are reduced, which can present a major issue in classic harmonic-vibration fatigue testing.
COBISS.SI-ID: 14742043