826
Viet Dung Luong et. al., Vol. 7, No. 2, 2021
Fig. 8. Comparison of numerical and experimental box compression test curves.
Rigid plates
Rigid bolts
Acceleration shock pulse
Fig. 9. Boundary conditions for shock test.
(a) (b) Fig. 10. Effect of mesh refinement on variable responses recorded at top rigid plate: (a) acceleration and (b) velocity.
Figure (8) shows the comparison of the experimental and numerical compression curves of the box with a good agreement. The maximum load obtained by the homogenized model is 1716.8 N compared to the experimental value of 1569.1 N giving a relative difference of 9.4%. 4.2 Repetitive Shock Results For the simulation of shock test, the finite element model consists of a box placed between two rigid plates connected by rigid bolts as shown in Fig. (9). Top plate has a mass of 8.4 kg as in experimental test and is free to move only vertically. An acceleration shock pulse is applied to the bottom rigid plate for a short time. Acceleration and velocity change are recorded on the bottom plate during the simulations. For this simulation friction interaction between plates and box was used to model boundary conditions of system. To replicate the experimental fatigue shock tests, simulations of successive shock pulses are carried out until the box is damaged. The box is considered damaged when the equivalent plastic strains exceeds 5%. To gain confidence in the accuracy of our model, we solved the model on progressively finer meshes and compared results. Since we need the accelerations and the velocity variations to plot DBC, we have plotted in Fig. (10) the results obtained for five successive shocks for three mesh refinements (h=8, 4, 2 mm) in the case of acceleration shock pulse of 16g and shock duration of 14.2 ms. The acceleration and velocity responses recorded at top rigid plate show similar trends for the three meshes, but the amplitudes are closer for the meshes h = 4 mm and h = 2 mm. We have also plotted various model energies for the three meshes in Fig. (11). As we perform a dynamic calculation, the internal and kinetic energies change over time (Figs. 11(a) and 11(b)). Figure 11(c) shows also the energy dissipated by plasticity. The energy balance for the three meshes is shown in Fig. 10(d) which should be constant. However, in the numerical model this is only approximately constant, generally with an error of less than 1% which is the case in our simulations. After this sensitivity analysis, we selected the mesh h = 4 mm for relevant computations while keeping a reasonable computational cost.
Journal of Applied and Computational Mechanics, Vol. 7, No. 2, (2021), 820-830
Made with FlippingBook Ebook Creator