Energy-based Research On The Dynamic Limit Status Of Shell

In this paper, the energy formulation of multiple freedom system is utilized. Based on energy concept, pre-program and post-program of FEM software ANSYS are edited to extract each kind of energy dissipation and total inputting energy during the dynamic non-linear time-history calculation of the shells subjected to severe earthquakes. As a result, the dynamic response regularities and damage characteristic of such structures are obtained. The relationship between the dynamic failure and the dissipation of the inputting earthquake energy of the shells is proposed. Furthermore, the effect of earthquake duration is also investigated. The paper provides some basis for the developing of anti-seismic structure design method and Double-Failure Criteria based on the first surpass of maximum displacement and plastic accumulative damage. In order to solve the reliability of the shells, the Response Surface Methodology is introduced on this paper. The solving of the reliability and the checking point is based on the Double-Failure Criteria by MATLAB programming. The prediction of the shells'dymatic failure load is accurate, avoids solving so many times by the time-history calculation, and improves the solving efficiency. The research result shows that when the duration time of earthquakes is changed, the transformation and the proportion relation of every kind of energies will be changed. and the damping energy accounts for a large part in the elastic stage, while the proportion of the kinetic energy and the strain energy is relatively small. In the plastic stage, the proportion of the nonlinear hysteretic exhausted energy gradually increases, conversely the part of damping decreases. When the structures are on the dymatic limit state, the proportion of the nonlinear hysteretic exhausted energy will leap and the summation of every kind of energies will become larger than total inputting energy. Based on the Response Surface Methodology, the change on the variance of the basic random has little effect on the predicted results. The reliability becomes small when the variance becomes large.The suggestion is that D must larger than 0.7 in sampling points for single-layer saddle-shaped reticular shells.