Parametric analysis of modelling properties governing the seismic response of free-standing spent fuel racks

Spent fuel racks are steel structures designed to store the spent fuel assemblies removed from the nuclear power reactor. In order to maximize the storage capacity of the pool, rack units are spaced by only a few centimeters setting up a matrix shape to fit in the spent fuel pool with a minimum clearance. Rack units rest in free-standing conditions submerged in water at 12 m depth. During a seismic event, racks undergo large displacements namely sliding, rocking, twisting and turning. Furthermore, the response of a unit is influenced by the others due to the so-called 'water coupling effect'. An accurate estimation of their response is essential to achieve a safe pool layout and a reliable structural design. The dynamic analysis of such a rack system deals with highly nonlinear behavior, a transient dynamic response and a fluid-structure interaction problem. A transient analysis with direct integration of the equation of motion throughout the whole earthquake duration becomes therefore unavoidable. An ad-hoc methodology based on the finite element method takes advantage of dynamic contact elements and implements the hydrodynamic mass concept. The latter has traditionally been accepted as a cost-effective approach to replace the water effect by an equivalent added mass. However, some dispersion of results still remains. This paper carries out a parametric analysis of the key modelling properties for a simple two-rack system. This technique examines the behavior of the main transient outputs as a modelling parameter is systematically varied. The modelling parameters under study are the mesh discretization, the rack-to-pool and fuel-to-cell contact stiffness, the flexural rigidity of the fuel assembly and the gaps existing between the fuel assembly and the storage cell. Its influence is highlighted on outputs as maximal and minimal relative displacements, maximal vertical force on support and CPU time. These numerical results provide a source of insight into the general behavior of the rack systems and an effective tool to propose a reliable modeling and meshing. The trade-off between outputs and computational cost and is also discussed.