High frequency mechanical impact treatment to improve fatigue life of welds

Report, 2018

High frequency mechanical impact (HFMI) is a post-weld peening process which is carried out to improve the fatigue life of welded geometries. The increase in fatigue strength is attributed to the combination of inducing compressive residual stresses at the weld toe, a change in the weld toe geometry from the peening, as well as an increased surface hardness in the treated region. To further investigate the benficial effects of HFMI, a benchmark exercise has been developed in the Specialist Committee V.3 Materials and Fabrication Technology of the International Ship and Offshore Structures Congress (ISSC 2018). The eventual expectation is the development of design guidelines for the use of HFMI in cyclically loaded components found, for example, in the ship building industry.
The benchmark exercise specifies the use of S355, a structural steel with a minimum yield strength of 355MPa, and a particular coupon geometry, which consists of a stiffener, fillet welded to a membrane loaded plate. This geometry, provided as a finite element model by the benchmark exercise committee, is known to be sensitive to fatigue as it has a high stress concentration at the weld toe. Material and mechanical properties for the simulation of HFMI and the cyclic analysis are also specified. Chalmers University of Technology contributes to the benchmark exercise through the course TME131 – Project in applied mechanics.
In this year’s project, an efficient way of simulating HFMI treatment is investigated and studies on how cyclic loading affects the induced beneficial compressive residual stresses are carried out. The project is executed in three different stages.
Stage 1 mainly concentrates on evaluating methods to simplify the modelling of the HFMI treatment. The goal is to reduce the computational effort without compromising the accuracy of the results. Simulations are performed on a simple cuboid geometry, also provided by the benchmark exercise, with varying parameters such as the constitutive hardening model, e.g. isotropic or kinematic or Chaboche, the analysis type, e.g dynamic or quasi-static, and the indenter tool model, e.g., a single tool or several tools applied in sequence. It is concluded that the choice of analysis type impacts the residual stress state to a minor extent, while it greatly affects the computational effort. A clear trend shows that with an increasing number of indenter parts, greater computational effort is required. Using a single indenter proved to give comparable results to previous work and a uniform residual stress profile. Since it is also computationally effective, it is concluded that this method is the most suitable.
The simplifications are then carried over into Stage 2, where the HFMI treatment is applied to the welded coupon geometry. In this stage, the indenter tool models from Stage 1 are redesigned to fit the coupon and simulations are performed to further evaluate the models. Simulations are performed with a variety of indenter tool models. The simulations with several indenters moving in sequence show a greater variation in the residual stress profile, suggesting some unreliability in this method. It is determined that a five-part single impact model was the the most suitable.
Finally, in Stage 3 the coupon model with induced residual stresses from a single impact HFMI simulation is subjected to cyclic membrane loading. The residual stress state is found to not be significantly impacted by constant amplitude loading. However, after variable amplitude loading the beneficial residual compressive stresses are found to be redistributed. Furthermore, the beneficial residual compressive stresses are removed to a greater extent when increasing the maximum applied nominal stress. When applying nominal stress with a peak load of 75% of the yield limit, the beneficial residual compressive stresses are reduced by almost 100%. However, when applying a nominal stress with a peak load of 63% of the yield limit, they are only reduced by 50%.
For simulating HFMI, the results suggest using the Chaboche mixed hardening model with quasi-static analysis using a single impact indenter tool model. For future work it is recommended to perform further investigation of the single impact dynamic simulations using kinematic hardening since it showed promising results. Furthermore, to achieve better understanding of the effects of cyclic loading, simulations with a wider range of load amplitudes, and closer investigations of the stress-strain development during loading, are recommended.