A fatigue damage assessment of welded joints under dynamic loading in automotive industry

The automotive companies demand for lightweight designed products to become competitor in business as long as technology develops. Even though there are many factors affect first time design capability, fatigue life prediction has been an on-going research area to improve design efficiency. Fatigue life prediction of welded connections that are often used in automotive industry is one of the most critical areas due to lower cost and practical. It is required to estimate fatigue life of welded connections to minimize customer claims regarding durability failures. However, prediction of fatigue life of welded joint is quite complicated because many factors affect fatigue life of welded connection. During the service life of welded structures exposed to various service loading conditions, welded joints are usually the potential fatigue failure sites due to the highest stress concentration areas and altered material properties. Thus, engineers and scientists are always interested in understanding fatigue characteristics of the welded joints, and are trying to develop analytical tools to estimate fatigue lives of welded joints. Due to complexity of manufacturing process, there are several analytical approaches to predict fatigue life of seam welds. Fatigue life of seam-welded joints is generally influenced by weld geometry, service loading history, and material properties. Thus, fatigue life calculation methods should be developed to account for those influencing factors. Traditionally, the fatigue life of welded joints was assessed with the nominal stress-based S-N curves generated from fatigue tests of welded specimens for different weld notch classifications. The complex weld geometries are placed in groups having similar fatigue strengths and identified by weld class depending on the weld types and loading conditions. These approaches are relatively simple but are limited to disclosing stresses and strains at the critical regions of the welded joint. The manufacturing effects are directly included in the large empirical database for structural steels, but the residual stress effect due to different manufacturing processes is not taken into account. It also appears difficult to determine weld class for complex weld shapes and loadings. In structural stress approaches, the structural stress is determined based on macro-behavior of a structure at the location where the fatigue crack is most likely to initiate and propagate. The structural stress is defined as the nominal stress at the weld toe or root cross-section, which excludes the local geometric (weld toe or root radius) effect in the stress calculation. Therefore, a structural stress is not a true local stress. An S-N curve is generated by fatigue testing fabricated welded laboratory specimens to failure at various structural stress levels. Thus, the fatigue life of a real welded structure can then be calculated by this structural stress at the critical welded joint and the S-N curve from laboratory testing. The notch stress approach considers the local pseudo stress in a stress concentration area as a fatigue damage parameter and requires fine meshes in the local weld toes and/or roots geometry to capture the accurate notch stresses from a linear elastic FEA. However, very large stress values can occur, depending on the notch radius. For a radius approaching zero, i.e. a sharp notch, the theoretical elastic stress even becomes infinite, i.e. singular. However, the fatigue behaviour of sharp notches is less determined by the high, localised stress peak, but more by effects of the material structure preventing excessive local yielding and supporting the notch root. These so-called microstructural support effects can be taken into account by evaluating the local stress gradient. The result is the fatigue effective notch stress. The notch stress life approach is employed for fatigue life predictions of welded joints. The synthetic notch stress life curve can be constructed by a given slope factor k and a median notch endurance limit at 2 $10^6$ cycles. The notch endurance limit can be obtained by fatigue testing welded laboratory specimens with the staircase test method for the endurance load amplitude at 2 $10^6$ cycles. A linear elastic FEA is performed to calculate the median notch endurance limit by applying the median endurance load. This is the way to convert the experimentally determined endurance load amplitude to the notch endurance limit. It is worth mentioning that residual stresses due to weld solidifications and distortions have been implicitly taken into account in the calculated notch endurance limit. The purpose of this thesis to perform an application of the notch stress approaches for fatigue damage assessment of a welded bracket of a light commercial vehicle and compare the result with traditional methods. Dynamic correlation and fatigue tests were performed as a part of this study. Fatigue tests were performed using electro-dynamic dynamic shaker to represent vehicle test conditions. An accelerated test is required that yields at least the same fatigue damage content as that seen by a component over its whole life. An accelerated test profile comprises several measured load events that are assumed typical of various real-life situations. The accelerated test profile was generated according to the measured application data from durability test vehicles. The connection brackets of heater system were failed before the targeted test cycle in the fatigue tests was completed. This failure was not expected compared to the durability analysis carried out before the fatigue tests. The prediction of fatigue life of brackets was inaccurate because dynamic behaviour of the system was not modelled correctly and fatigue properties of weld region were not taken into account. After failure of the bracket on the welded area, the fatigue life calculation using weld fatigue assessment methods was planned to assess failure risk of welded connections. Structural correlated model is required for high fidelity simulations and fatigue life calculation. Model correlation was conducted to improve simulation model quality based on reference hammer test data. The damping loss factor of the system was obtained with harmonic excitation test of the specimen. The dynamic properties of the simulation model were validated using acceleration measurement data from shaker test. After correlation process, the fatigue life of the bracket was calculated by notch stress approach. Femfat fatigue analysis modules based on notch stress approaches were used to predict fatigue life of the seam weld subjected to prescribed loading. It has been found that the damage values obtained as a result of the calculations are in accordance with damage in the test. The notch stress method provides good agreement with the test results only when the dynamic behavior are considered correctly. Determination of dynamic behavior of the system is possible only by testing the manufactured parts. In development stages, damage analyses can be performed under some assumptions so reliability of the damage analyses is low. As a result of the study, notch stress approaches provides reliable and directional results when it is compared with traditional analyses methods which not include special weld fatigue parameters. Finally, new designs were proposed using similar finite element models and calculation methods.