Improvements and verification of an accelerated technique for simulations of cyclically loaded structures

Zheng, Songwei
Journal Title
Journal ISSN
Volume Title
University of Delaware
In engineering applications, structures are commonly subjected to cyclic loadings. This may lead to fatigue and unexpected failures. To prevent the life-time limiting failures, understanding of the failure evolution in these structures during use is very important. For failure prediction due to cyclic loading, finite element analysis (FEA) can be used to simulate and establish the stress and strain distribution as a function of time. However, every single cycle of simulation consists of many computational increments and iterations. The whole process of structural evolution consists of a large number of cycles. Thus, for structures subjected to cyclic loadings, it is extremely time-consuming and inefficient to simulate the whole process of structural evolution. In fact, in most cases, the computational time required is typically prohibiting a complete analysis. The goal with this work is to improve upon an existing numerical scheme that aims to, in combination with simple testing, predict the life-time of structures subjected to cyclic loading. The numerical scheme is the “cycle-jump technique” developed previously. The fundamental idea of the cycle-jump technique is that there is no need to calculate each individual cycle in a cyclically loaded structure. Utilizing an extrapolation scheme for extrapolating the overall behavior to “jump” over some cycles, the cycle-jump technique may predict the overall structural behaviors with much higher time efficiency. In this work, the “cycle jump technique” will be modified and improved to eventually be used for simulating realistic designs. To overcome the limitation of the existing extrapolation scheme, an alternative more general extrapolation scheme is proposed. The numerical code is also improved for applications in simulating time-dependent material behaviors, such as time-dependent oxidation evolution and creep in thermal barrier coatings subjected to high temperature thermal cycles. To verify that the cycle jump technique can capture real-life experimental results and to demonstrate the power of the method, experimental results, the cycle-by-cycle reference simulation, and the simulation with cycle jump must be compared. Preliminary experiments guided by our preliminary simulations were performed by our collaborators. Preliminary experimental results are used to compare with the simulation results. Even though the experimental data is limited, it appears as our numerical model can predict the evolution of the test samples, and incorporated the cycle-jump technique will improve the computational efficiency.