Etude des vibrations aleatoires d'origine hydrodynamique de certaines structures du bloc reacteur phenix

Abstract An important source of vibrations in the vessels of a reactor such as the sodium cooled fast French reactor Phenix is the turbulence of the liquid sodium flow. In fact, the random pressure pulsations may excite the natural modes of the structures if they are within the low frequency range. At the stage of the project, our purpose was to ascertain whether the vessels and internals of the reactor were adequately stable and if not, to determine the necessary changes to achieve this aim. Our investigation can be divided into three phases. First, we measured the vibrational characteristics of the considered structures on mechanical models at the scales of 1 10 or 1 5 in the air as well as in quiet water simulating sodium. In conjunction with these experiments, theoretical and experimental investigations for the breathing vibrations were done on the modal behaviour of various patterns of cylindrical immersed shells. The comparison of results led us to liken the structures of the reactor to simple shell schemes easy to calculate. As regards exciting forces, pressure recordings carried on hydraulic models at the scale of 1 4 were given to us. We analysed the statistical features of the pressure pulsations. On one hand, we studied on digital computer the power spectral density functions of the excitation pressure recorded by a pressure transducer located at various points of the vessel surface and on the other hand, the power co-spectral density functions relating to several pairs of pressure transducers located at more or less near intervals. The results show that on the surface of each vessel, the power spectral density function is pracically independent on the measurement point and can be analytically represented in a simple manner. Then, we calculated the response of the structures analytically by using the theory of random vibrations of the shells which we developed to account for the specificity of the problem. We determined an upper bound of the response by using only the model characteristics and the power spectral density functions. This was sufficient to guarantee the mechanical stability of structures considered, after an eventual stiffening. However, in view of future needs, we tried to develop a more elaborate calculation of the response, taking the power cospectral density function into account. Interesting results were obtained, for which, however, additional validation would be of use. Special attention was given to the difficult estimate of fluid damping and some basic damping studies were carried out. In our opinion, these investigations permit to progress with the knowledge of the vibrations of nuclear reactors. However, the models necessarily present some distortions and all phenomena cannot be considered. Therefore, our results shall be checked as much as possible during tests to be done after the erection of the reactor.