PSI - Issue 57

Ewelina Czerlunczakiewicz et al. / Procedia Structural Integrity 57 (2024) 743–753 / Structural Integrity Procedia 00 (2019) 000 – 000

749

7

the same acceleration time series. To achieve this, a random phase was generated to recreate a random signal in the time domain and the same phase based signal was used for each analysis, allowing for the comparison of not only cumulative damage, but also displacements, stress, and accelerations at specific time steps. 3. Results 3.1. Linear results comparison The analysis and comparison of results were divided into two steps. In the first step, all the linear analyses were compared. The second step considered the non-linear analysis results as well. Three different linear analysis were compared: - SSD (efficient, widely used, linear approach), - MTD (modal transient dynamic), - Linear DI (Linear Direct integration). The elements with the highest damage values from various components were analyzed. A comparison of the damage normalized to Linear DI results of the high temperature radiator is shown in the graph below (Fig 6a). It can be observed that each analyzed part and hot spot region exhibit the same trend. In the graph below, the linear direct implicit (linear DI) results were taken as the reference for comparison, as they are also used for comparing the nonlinear results. Additionally cycles counting comparison for different methods is presented in the Fig. 6 b.

Fig. 6. (a)Damage comparison (b) cycles counting for different linear simulations The conclusion was reached that all linear approaches provide comparable results regardless of whether they are directly integrated or mode-based and calculated in the time or frequency domain. As a result, only one of these three linear analyses is considered in the second step of the comparison. In the second step, the focus is on comparing the non-linear direct-integration dynamic analysis with the linear direct-integration dynamic analysis, which was chosen as the most similar linear approach. 3.2. Linear and non-linear results comparison The second step of the comparison started by examining the behavior of the nonlinear model. A small shift in the natural frequencies was observed, caused by the definition of the clipping connections, and its influence was investigated. To isolate the impa ct of nonlinearity, the natural frequencies of the linear model were adjusted to match those of the nonlinear model, effectively eliminating the frequency gap between the two models. It was found that the observed frequency shift had a negligible impact on the results. Further analysis of the non linear model focused on examining the transient global behavior differences using displacement plots. As shown in the plot (Fig. 7), depending on the analyzed timestep and region, the predicted non - linear results can exhibit higher or lower displacement values compared to the linear results. Additionally, noticeable phase shifts and dephasing can be observed between both curves.