Experimental and analytical study on buckling of fluid-coupled structures during a seismic load

Seismic analysis of a liquid metal fast breeder reactor is generally made with a linear dynamic model. The structure response is a combination of modes n=1 for horizontal earthquakes and n=0 for vertical earthquakes. The resulting loading is used for stability calculation on ''static equivalent'' models. This procedure may not be correct for two main reasons: geometrical defects and large displacements could change significantly the behaviour of the structures; the dynamic buckling pressure could be different from the static buckling pressure. To answer these questions, an experimental and analytical study on buckling of fluid-coupled structures during a seismic load has been performed by Electricite de France, the Commissariat a l'Energie Atomique and the Universite de Compiegne. The experiments are made on semi-spherical and cylindrical shells on a vibrating table; the loading applied to the structures consists of a permanent pressure and of a dynamic pressure due to fluid-structure interaction. The experimental results are presented and compared with the calculations. This study presents four main points: the dynamic pressure does not vary linearly with the shaking-table acceleration; at high acceleration level, the dynamic pressure increases very slowly; the prestress due to permanent pressure can have a significant effect on the eigenfrequencies if the buckling modes are colinear to the vibration modes; for hemispherical shells studied here, the buckling is obtained when the sum of permanent pressure and dynamic pressure is greater than the static buckling pressure; this structure excited with a horizontal motion can have a response in circumferential modes and not only in mode 1 as predicted by the theory of perfect shells. The defects explain the coupling of modes. (author). 9 figs, 6 tabs, 2 photos