EXPERIMENTS FOR VALIDATION OF FSI MODELS FOR SEISMIC RESPONSE OF ADVANCED REACTOR INTERNALS

Accurate prediction of the dynamic seismic response of reactor vessel internals (RVIs) submerged in fluid will be critical to the design, qualification and risk assessment of advanced nuclear reactors. Analytical solutions for submerged components using the incompressible inviscid theory or the more complicated compressible inviscid or incompressible viscous theories are available for simple geometrical shapes, idealized boundary conditions and unidirectional seismic inputs of small amplitude. The geometries and boundary conditions proposed for advanced reactors and their internals are complex and analytical solutions are generally not applicable. Physical testing of reactor vessels and internals is not feasible given their size and cost. Design, qualification and risk assessment calculations will have to rely on the use of verified and validated numerical models that are capable of capturing the interaction of internals with the surrounding fluid (fluid-structure interaction: FSI) over a wide range of shaking. A two-phase experimental program was performed on a six degrees-of-freedom earthquake simulator at the University at Buffalo to generate data that will support validation of numerical models in commercial finite element codes. Phase-I testing involved a model of a base-supported reactor vessel without head or internals. The focus of this paper is Phase-II testing that involved the Phase-I vessel with a head and internals. Phase-II experiments investigated pressure histories on reactor internals; added mass, added damping and coupling effects in internals along with strain and acceleration responses under three-component seismic excitations. The effects of seismic isolation on the dynamic responses of internals were investigated using numerically generated input motions simulating a virtual isolation system. The generated data will be curated and archived at the end of the testing program for later use by other analysts seeking to validate numerical models not evaluated by the authors.